From 894ecf8b3c6d7eabb3baa2714b63b65aa233f2fe Mon Sep 17 00:00:00 2001
From: wenming2014 <2279939962@qq.com>
Date: Fri, 25 Sep 2020 08:39:30 +0000
Subject: [PATCH] fix ci problems

---
 0924.log                            | 9432 +++++++++++++++++++++++++++
 Testing/Temporary/CTestCostData.txt |    1 +
 Testing/Temporary/LastTest.log      |    3 +
 cinn/hlir/pe/nn.cc                  |   78 +-
 python/tests/conv2d_utils.py        |   11 +-
 python/tests/test_op_nn.py          |    6 +-
 python/tests/test_pe_reduction.py   |    2 +-
 7 files changed, 9519 insertions(+), 14 deletions(-)
 create mode 100644 0924.log
 create mode 100644 Testing/Temporary/CTestCostData.txt
 create mode 100644 Testing/Temporary/LastTest.log

diff --git a/0924.log b/0924.log
new file mode 100644
index 0000000000..74bb078c67
--- /dev/null
+++ b/0924.log
@@ -0,0 +1,9432 @@
+/home/wangyue50/CINN-my/CINN/build
+-- The C compiler identification is GNU 7.5.0
+-- The CXX compiler identification is GNU 7.5.0
+-- Check for working C compiler: /usr/bin/cc
+-- Check for working C compiler: /usr/bin/cc -- works
+-- Detecting C compiler ABI info
+-- Detecting C compiler ABI info - done
+-- Detecting C compile features
+-- Detecting C compile features - done
+-- Check for working CXX compiler: /usr/bin/c++
+-- Check for working CXX compiler: /usr/bin/c++ -- works
+-- Detecting CXX compiler ABI info
+-- Detecting CXX compiler ABI info - done
+-- Detecting CXX compile features
+-- Detecting CXX compile features - done
+-- Looking for pthread.h
+-- Looking for pthread.h - found
+-- Looking for pthread_create
+-- Looking for pthread_create - not found
+-- Looking for pthread_create in pthreads
+-- Looking for pthread_create in pthreads - not found
+-- Looking for pthread_create in pthread
+-- Looking for pthread_create in pthread - found
+-- Found Threads: TRUE  
+-- pybind path: /home/wangyue50/CINN-my/CINN/build/thirds/pybind/src/extern_pybind/include
+-- third: /home/wangyue50/CINN-my/CINN/build/thirds
+-- Protobuf protoc executable: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/bin/protoc
+-- Protobuf-lite library: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/libprotobuf-lite.a
+-- Protobuf library: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/libprotobuf.a
+-- Protoc library: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/libprotoc.a
+-- Protobuf version: 3.1.0
+-- MKLML_VER: Glibc225_vsErf_mklml_lnx_2019.0.1.20181227, MKLML_URL: http://paddlepaddledeps.bj.bcebos.com/Glibc225_vsErf_mklml_lnx_2019.0.1.20181227.tgz
+-- ginac: /usr/lib/libginac.so
+-- core_src: ;cinn/common/shared.cc;cinn/common/cinn_value.cc;cinn/common/type.cc;cinn/common/target.cc;cinn/common/object.cc;cinn/common/debug_manager.cc;cinn/common/info_registry.cc;cinn/common/graph_utils.cc;cinn/common/context.cc;cinn/common/axis.cc;cinn/common/ir_util.cc;cinn/common/test_helper.cc;cinn/common/arithmatic.cc;cinn/common/cas.cc;cinn/common/union_find.cc;cinn/utils/string.cc;cinn/utils/functional.cc;cinn/utils/dot_lang.cc;cinn/utils/timer.cc;cinn/poly/dim.cc;cinn/poly/domain.cc;cinn/poly/map.cc;cinn/poly/stage.cc;cinn/poly/isl_utils.cc;cinn/poly/schedule.cc;cinn/poly/naive_scheduler.cc;cinn/poly/poly_scheduler.cc;cinn/poly/ast_gen.cc;cinn/poly/graph.cc;cinn/poly/compute_at_transform.cc;cinn/runtime/cpu/host_intrinsics.cc;cinn/runtime/cpu/mkl_math.cc;cinn/runtime/cpu/cblas.cc;cinn/runtime/intrinsic.cc;cinn/runtime/cinn_runtime.cc;cinn/runtime/intrinsic_types.cc;cinn/ir/node.cc;cinn/ir/ir.cc;cinn/ir/node.cc;cinn/ir/ir_visitor.cc;cinn/ir/ir_printer.cc;cinn/ir/ir_mutator.cc;cinn/ir/function_definition.cc;cinn/ir/lowered_func.cc;cinn/ir/ir_operators.cc;cinn/ir/buffer.cc;cinn/ir/function_base.cc;cinn/ir/operation.cc;cinn/ir/collect_ir_nodes.cc;cinn/ir/registry.cc;cinn/ir/tensor.cc
+-- Compile core_api without CUDA support
+-- Download inference test stuff from http://paddle-inference-dist.bj.bcebos.com/lite_naive_model.tar.gz
+-- Configuring done
+-- Generating done
+-- Build files have been written to: /home/wangyue50/CINN-my/CINN/build
+Scanning dependencies of target extern_gtest
+[  0%] Creating directories for 'extern_gtest'
+[ 50%] Performing download step (git clone) for 'extern_gtest'
+[ 50%] No patch step for 'extern_gtest'
+[ 50%] No update step for 'extern_gtest'
+[100%] Performing configure step for 'extern_gtest'
+loading initial cache file /home/wangyue50/CINN-my/CINN/build/thirds/gtest/tmp/extern_gtest-cache-.cmake
+-- The C compiler identification is GNU 7.5.0
+-- The CXX compiler identification is GNU 7.5.0
+-- Check for working C compiler: /usr/bin/cc
+-- Check for working C compiler: /usr/bin/cc -- works
+-- Detecting C compiler ABI info
+-- Detecting C compiler ABI info - done
+-- Detecting C compile features
+-- Detecting C compile features - done
+-- Check for working CXX compiler: /usr/bin/c++
+-- Check for working CXX compiler: /usr/bin/c++ -- works
+-- Detecting CXX compiler ABI info
+-- Detecting CXX compiler ABI info - done
+-- Detecting CXX compile features
+-- Detecting CXX compile features - done
+-- Found PythonInterp: /usr/bin/python (found version "3.6.9") 
+-- Configuring done
+-- Generating done
+-- Build files have been written to: /home/wangyue50/CINN-my/CINN/build/thirds/gtest/src/extern_gtest-build
+[100%] Performing build step for 'extern_gtest'
+Scanning dependencies of target gmock_main
+[  9%] Building CXX object googlemock/CMakeFiles/gmock_main.dir/__/googletest/src/gtest-all.cc.o
+[ 18%] Building CXX object googlemock/CMakeFiles/gmock_main.dir/src/gmock-all.cc.o
+[ 27%] Building CXX object googlemock/CMakeFiles/gmock_main.dir/src/gmock_main.cc.o
+[ 36%] Linking CXX static library libgmock_main.a
+[ 36%] Built target gmock_main
+Scanning dependencies of target gmock
+[ 45%] Building CXX object googlemock/CMakeFiles/gmock.dir/__/googletest/src/gtest-all.cc.o
+[ 54%] Building CXX object googlemock/CMakeFiles/gmock.dir/src/gmock-all.cc.o
+[ 63%] Linking CXX static library libgmock.a
+[ 63%] Built target gmock
+Scanning dependencies of target gtest
+[ 72%] Building CXX object googlemock/gtest/CMakeFiles/gtest.dir/src/gtest-all.cc.o
+[ 81%] Linking CXX static library libgtest.a
+[ 81%] Built target gtest
+Scanning dependencies of target gtest_main
+[ 90%] Building CXX object googlemock/gtest/CMakeFiles/gtest_main.dir/src/gtest_main.cc.o
+[100%] Linking CXX static library libgtest_main.a
+[100%] Built target gtest_main
+[100%] Performing install step for 'extern_gtest'
+[ 36%] Built target gmock_main
+[ 63%] Built target gmock
+[ 81%] Built target gtest
+[100%] Built target gtest_main
+Install the project...
+-- Install configuration: ""
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/lib/libgmock.a
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/lib/libgmock_main.a
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-more-matchers.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-cardinalities.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-matchers.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-generated-function-mockers.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-generated-matchers.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-more-actions.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-generated-actions.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-generated-function-mockers.h.pump
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-generated-nice-strict.h.pump
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-generated-matchers.h.pump
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/internal
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/internal/gmock-generated-internal-utils.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/internal/gmock-internal-utils.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/internal/gmock-port.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/internal/gmock-generated-internal-utils.h.pump
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/internal/custom
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/internal/custom/gmock-matchers.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/internal/custom/gmock-generated-actions.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/internal/custom/gmock-port.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/internal/custom/gmock-generated-actions.h.pump
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-spec-builders.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-generated-actions.h.pump
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-generated-nice-strict.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gmock/gmock-actions.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/lib/libgtest.a
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/lib/libgtest_main.a
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest-param-test.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest-death-test.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest-message.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest-spi.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest-param-test.h.pump
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest-printers.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest_prod.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest-typed-test.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest_pred_impl.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-internal.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-type-util.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-param-util-generated.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-filepath.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-param-util-generated.h.pump
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-param-util.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-port-arch.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-tuple.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-port.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-death-test-internal.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-string.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/custom
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/custom/gtest-printers.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/custom/gtest-port.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/custom/gtest.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-type-util.h.pump
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-linked_ptr.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/internal/gtest-tuple.h.pump
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gtest/include/gtest/gtest-test-part.h
+[100%] Completed 'extern_gtest'
+[100%] Built target extern_gtest
+Scanning dependencies of target extern_gflags
+Scanning dependencies of target extern_mklml
+Scanning dependencies of target extern_protobuf
+Scanning dependencies of target GEN_LLVM_RUNTIME_IR_HEADER
+[  3%] Built target extern_gtest
+[  3%] Creating directories for 'extern_mklml'
+[  5%] Creating directories for 'extern_gflags'
+[  5%] Generating cinn_runtime_llvm_ir.h
+[  5%] Creating directories for 'extern_protobuf'
+[  5%] Performing download step (git clone) for 'extern_gflags'
+[  7%] Performing download step (git clone) for 'extern_protobuf'
+[  7%] Performing download step (download, verify and extract) for 'extern_mklml'
+-- Downloading...
+   dst='/home/wangyue50/CINN-my/CINN/build/thirds/mklml/src/extern_mklml/Glibc225_vsErf_mklml_lnx_2019.0.1.20181227.tgz'
+   timeout='none'
+-- Using src='http://paddlepaddledeps.bj.bcebos.com/Glibc225_vsErf_mklml_lnx_2019.0.1.20181227.tgz'
+[  7%] Built target GEN_LLVM_RUNTIME_IR_HEADER
+Submodule path 'doc': checked out '971dd2a4fadac9cdab174c523c22df79efd63aa5'
+[  9%] No patch step for 'extern_gflags'
+[  9%] No update step for 'extern_gflags'
+[  9%] Performing configure step for 'extern_gflags'
+loading initial cache file /home/wangyue50/CINN-my/CINN/build/thirds/gflags/tmp/extern_gflags-cache-.cmake
+-- The CXX compiler identification is GNU 7.5.0
+-- Check for working CXX compiler: /usr/bin/c++
+-- Check for working CXX compiler: /usr/bin/c++ -- works
+-- Detecting CXX compiler ABI info
+-- Detecting CXX compiler ABI info - done
+-- Detecting CXX compile features
+-- Detecting CXX compile features - done
+-- Looking for C++ include unistd.h
+-- Looking for C++ include unistd.h - found
+-- Looking for C++ include stdint.h
+-- Looking for C++ include stdint.h - found
+-- Looking for C++ include inttypes.h
+-- Looking for C++ include inttypes.h - found
+-- Looking for C++ include sys/types.h
+-- Looking for C++ include sys/types.h - found
+-- Looking for C++ include sys/stat.h
+-- Looking for C++ include sys/stat.h - found
+-- Looking for C++ include fnmatch.h
+-- Looking for C++ include fnmatch.h - found
+-- Looking for C++ include stddef.h
+-- Looking for C++ include stddef.h - found
+-- Check size of uint32_t
+-- Check size of uint32_t - done
+-- Looking for strtoll
+-- Looking for strtoll - found
+-- Looking for C++ include pthread.h
+-- Looking for C++ include pthread.h - found
+-- Looking for pthread_create
+-- Looking for pthread_create - not found
+-- Looking for pthread_create in pthreads
+-- Looking for pthread_create in pthreads - not found
+-- Looking for pthread_create in pthread
+-- Looking for pthread_create in pthread - found
+-- Found Threads: TRUE  
+-- Check size of pthread_rwlock_t
+-- Check size of pthread_rwlock_t - done
+-- Configuring done
+-- Generating done
+-- Build files have been written to: /home/wangyue50/CINN-my/CINN/build/thirds/gflags/src/extern_gflags-build
+[  9%] Performing build step for 'extern_gflags'
+Scanning dependencies of target gflags_nothreads_static
+Scanning dependencies of target gflags_static
+[ 12%] Building CXX object CMakeFiles/gflags_nothreads_static.dir/src/gflags.cc.o
+[ 25%] Building CXX object CMakeFiles/gflags_nothreads_static.dir/src/gflags_completions.cc.o
+[ 37%] Building CXX object CMakeFiles/gflags_nothreads_static.dir/src/gflags_reporting.cc.o
+[ 62%] Building CXX object CMakeFiles/gflags_static.dir/src/gflags.cc.o
+[ 62%] Building CXX object CMakeFiles/gflags_static.dir/src/gflags_reporting.cc.o
+[ 75%] Building CXX object CMakeFiles/gflags_static.dir/src/gflags_completions.cc.o
+[ 87%] Linking CXX static library lib/libgflags.a
+[100%] Linking CXX static library lib/libgflags_nothreads.a
+[100%] Built target gflags_static
+[100%] Built target gflags_nothreads_static
+[  9%] Performing install step for 'extern_gflags'
+[ 50%] Built target gflags_nothreads_static
+[100%] Built target gflags_static
+Install the project...
+-- Install configuration: ""
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/lib/libgflags.a
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/lib/libgflags_nothreads.a
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/include/gflags/gflags.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/include/gflags/gflags_declare.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/include/gflags/gflags_completions.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/include/gflags/gflags_gflags.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/lib/cmake/gflags/gflags-config.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/lib/cmake/gflags/gflags-config-version.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/lib/cmake/gflags/gflags-targets.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/lib/cmake/gflags/gflags-targets-noconfig.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/bin/gflags_completions.sh
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/gflags/lib/pkgconfig/gflags.pc
+-- Installing: /root/.cmake/packages/gflags/a4bd6f63375cbd4b0411e04dd0704cce
+[  9%] Completed 'extern_gflags'
+[  9%] Built target extern_gflags
+Scanning dependencies of target extern_glog
+[  9%] Creating directories for 'extern_glog'
+[  9%] Performing download step (git clone) for 'extern_glog'
+-- Downloading... done
+-- extracting...
+     src='/home/wangyue50/CINN-my/CINN/build/thirds/mklml/src/extern_mklml/Glibc225_vsErf_mklml_lnx_2019.0.1.20181227.tgz'
+     dst='/home/wangyue50/CINN-my/CINN/build/thirds/mklml/src/extern_mklml'
+-- extracting... [tar xfz]
+-- extracting... [analysis]
+-- extracting... [rename]
+-- extracting... [clean up]
+-- extracting... done
+[ 11%] No update step for 'extern_mklml'
+[ 11%] No patch step for 'extern_mklml'
+[ 11%] No configure step for 'extern_mklml'
+[ 11%] No build step for 'extern_mklml'
+[ 13%] Performing install step for 'extern_mklml'
+[ 13%] Completed 'extern_mklml'
+[ 13%] Built target extern_mklml
+[ 15%] No update step for 'extern_glog'
+[ 15%] No patch step for 'extern_glog'
+[ 15%] Performing configure step for 'extern_glog'
+loading initial cache file /home/wangyue50/CINN-my/CINN/build/thirds/glog/tmp/extern_glog-cache-.cmake
+-- The C compiler identification is GNU 7.5.0
+-- The CXX compiler identification is GNU 7.5.0
+-- Check for working C compiler: /usr/bin/cc
+-- Check for working C compiler: /usr/bin/cc -- works
+-- Detecting C compiler ABI info
+-- Detecting C compiler ABI info - done
+-- Detecting C compile features
+-- Detecting C compile features - done
+-- Check for working CXX compiler: /usr/bin/c++
+-- Check for working CXX compiler: /usr/bin/c++ -- works
+-- Detecting CXX compiler ABI info
+-- Detecting CXX compiler ABI info - done
+Branch 'mobile' set up to track remote branch 'mobile' from 'origin'.
+-- Detecting CXX compile features
+-- Detecting CXX compile features - done
+-- Looking for gflags namespace
+-- Looking for gflags namespace - gflags
+-- Looking for pthread.h
+-- Looking for pthread.h - found
+-- Looking for pthread_create
+-- Looking for pthread_create - not found
+-- Looking for pthread_create in pthreads
+-- Looking for pthread_create in pthreads - not found
+-- Looking for pthread_create in pthread
+-- Looking for pthread_create in pthread - found
+-- Found Threads: TRUE  
+-- Looking for dlfcn.h
+-- Looking for dlfcn.h - found
+-- Looking for execinfo.h
+-- Looking for execinfo.h - found
+-- Looking for glob.h
+-- Looking for glob.h - found
+-- Looking for inttypes.h
+-- Looking for inttypes.h - found
+-- Looking for libunwind.h
+-- Looking for libunwind.h - not found
+-- Looking for memory.h
+-- Looking for memory.h - found
+-- Looking for pwd.h
+-- Looking for pwd.h - found
+-- Looking for stdint.h
+-- Looking for stdint.h - found
+-- Looking for stdlib.h
+-- Looking for stdlib.h - found
+-- Looking for string.h
+-- Looking for string.h - found
+-- Looking for strings.h
+-- Looking for strings.h - found
+-- Looking for sys/stat.h
+-- Looking for sys/stat.h - found
+-- Looking for sys/syscall.h
+-- Looking for sys/syscall.h - found
+-- Looking for sys/time.h
+-- Looking for sys/time.h - found
+-- Looking for sys/types.h
+-- Looking for sys/types.h - found
+-- Looking for sys/utsname.h
+-- Looking for sys/utsname.h - found
+-- Looking for syscall.h
+-- Looking for syscall.h - found
+-- Looking for syslog.h
+-- Looking for syslog.h - found
+-- Looking for ucontext.h
+-- Looking for ucontext.h - found
+-- Looking for unistd.h
+-- Looking for unistd.h - found
+-- Looking for unwind.h
+-- Looking for unwind.h - found
+-- Looking for C++ include ext/hash_map
+-- Looking for C++ include ext/hash_map - found
+-- Looking for C++ include ext/hash_set
+-- Looking for C++ include ext/hash_set - found
+-- Looking for C++ include ext/slist
+Submodule path 'third_party/benchmark': checked out '360e66c1c4777c99402cf8cd535aa510fee16573'
+-- Looking for C++ include ext/slist - found
+-- Looking for C++ include tr1/unordered_map
+[ 15%] No update step for 'extern_protobuf'
+[ 15%] No patch step for 'extern_protobuf'
+[ 15%] Performing configure step for 'extern_protobuf'
+-- The C compiler identification is GNU 7.5.0
+-- Looking for C++ include tr1/unordered_map - found
+-- Looking for C++ include tr1/unordered_set
+-- The CXX compiler identification is GNU 7.5.0
+-- Check for working C compiler: /usr/bin/cc
+-- Check for working C compiler: /usr/bin/cc -- works
+-- Detecting C compiler ABI info
+-- Looking for C++ include tr1/unordered_set - found
+-- Looking for C++ include unordered_map
+-- Detecting C compiler ABI info - done
+-- Detecting C compile features
+-- Looking for C++ include unordered_map - found
+-- Looking for C++ include unordered_set
+-- Detecting C compile features - done
+-- Check for working CXX compiler: /usr/bin/c++
+-- Check for working CXX compiler: /usr/bin/c++ -- works
+-- Detecting CXX compiler ABI info
+-- Detecting CXX compiler ABI info - done
+-- Detecting CXX compile features
+-- Looking for C++ include unordered_set - found
+-- Looking for stddef.h
+-- Looking for stddef.h - found
+-- Check size of unsigned __int16
+-- Check size of unsigned __int16 - failed
+-- Check size of u_int16_t
+-- Check size of u_int16_t - done
+-- Check size of uint16_t
+-- Check size of uint16_t - done
+-- Looking for dladdr
+-- Looking for dladdr - not found
+-- Looking for fcntl
+-- Detecting CXX compile features - done
+-- Looking for pthread.h
+-- Looking for fcntl - found
+-- Looking for pread
+-- Looking for pthread.h - found
+-- Looking for pthread_create
+-- Looking for pread - found
+-- Looking for pwrite
+-- Looking for pthread_create - not found
+-- Looking for pthread_create in pthreads
+-- Looking for pwrite - found
+-- Looking for sigaction
+-- Looking for pthread_create in pthreads - not found
+-- Looking for pthread_create in pthread
+-- Looking for sigaction - found
+-- Looking for sigaltstack
+-- Looking for pthread_create in pthread - found
+-- Found Threads: TRUE  
+-- Configuring done
+-- Looking for sigaltstack - found
+-- Performing Test HAVE_NO_DEPRECATED
+-- Generating done
+-- Build files have been written to: /home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf-build
+[ 16%] Performing build step for 'extern_protobuf'
+-- Performing Test HAVE_NO_DEPRECATED - Success
+-- Performing Test HAVE_NO_UNNAMED_TYPE_TEMPLATE_ARGS
+-- Performing Test HAVE_NO_UNNAMED_TYPE_TEMPLATE_ARGS - Failed
+-- Looking for snprintf
+Scanning dependencies of target libprotobuf-lite
+[  0%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/arena.cc.o
+[  1%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/arenastring.cc.o
+[  2%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/extension_set.cc.o
+[  2%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/generated_message_util.cc.o
+[  2%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/coded_stream.cc.o
+[  3%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/zero_copy_stream.cc.o
+Scanning dependencies of target libprotobuf
+-- Looking for snprintf - found
+-- Looking for get_static_proc_name in unwind
+[  3%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/arena.cc.o
+-- Looking for get_static_proc_name in unwind - not found
+-- Performing Test HAVE___ATTRIBUTE__
+[  4%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/arenastring.cc.o
+[  5%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/extension_set.cc.o
+[  5%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/generated_message_util.cc.o
+-- Performing Test HAVE___ATTRIBUTE__ - Success
+-- Performing Test HAVE___ATTRIBUTE__VISIBILITY_DEFAULT
+[  5%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/coded_stream.cc.o
+[  5%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/zero_copy_stream_impl_lite.cc.o
+-- Performing Test HAVE___ATTRIBUTE__VISIBILITY_DEFAULT - Success
+-- Performing Test HAVE___ATTRIBUTE__VISIBILITY_HIDDEN
+[  6%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/message_lite.cc.o
+[  6%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/repeated_field.cc.o
+[  7%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/zero_copy_stream.cc.o
+-- Performing Test HAVE___ATTRIBUTE__VISIBILITY_HIDDEN - Success
+-- Performing Test HAVE___BUILTIN_EXPECT
+-- Performing Test HAVE___BUILTIN_EXPECT - Success
+-- Performing Test HAVE___SYNC_VAL_COMPARE_AND_SWAP
+[  8%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/atomicops_internals_x86_gcc.cc.o
+-- Performing Test HAVE___SYNC_VAL_COMPARE_AND_SWAP - Success
+-- Performing Test HAVE_RWLOCK
+[  8%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/atomicops_internals_x86_msvc.cc.o
+[  8%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/zero_copy_stream_impl_lite.cc.o
+-- Performing Test HAVE_RWLOCK - Failed
+-- Performing Test HAVE___DECLSPEC
+-- Performing Test HAVE___DECLSPEC - Failed
+-- Performing Test STL_NO_NAMESPACE
+[  9%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/message_lite.cc.o
+[  9%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/repeated_field.cc.o
+[ 10%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/atomicops_internals_x86_gcc.cc.o
+-- Performing Test STL_NO_NAMESPACE - Failed
+-- Performing Test STL_STD_NAMESPACE
+[ 10%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/atomicops_internals_x86_msvc.cc.o
+[ 11%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/bytestream.cc.o
+[ 11%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/common.cc.o
+[ 12%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/int128.cc.o
+-- Performing Test STL_STD_NAMESPACE - Success
+-- Performing Test HAVE_USING_OPERATOR
+[ 12%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/once.cc.o
+[ 13%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/status.cc.o
+[ 14%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/bytestream.cc.o
+[ 14%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/common.cc.o
+[ 15%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/int128.cc.o
+-- Performing Test HAVE_USING_OPERATOR - Success
+-- Performing Test HAVE_NAMESPACES
+[ 15%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/once.cc.o
+[ 16%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/status.cc.o
+-- Performing Test HAVE_NAMESPACES - Success
+[ 16%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/statusor.cc.o
+[ 17%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/stringprintf.cc.o
+[ 17%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/stringpiece.cc.o
+[ 18%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/structurally_valid.cc.o
+[ 19%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/strutil.cc.o
+[ 19%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/statusor.cc.o
+[ 19%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/time.cc.o
+[ 20%] Building CXX object CMakeFiles/libprotobuf-lite.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/wire_format_lite.cc.o
+[ 21%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/stringpiece.cc.o
+[ 21%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/stringprintf.cc.o
+[ 22%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/structurally_valid.cc.o
+[ 22%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/strutil.cc.o
+[ 23%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/time.cc.o
+[ 23%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/wire_format_lite.cc.o
+[ 24%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/any.cc.o
+[ 24%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/any.pb.cc.o
+-- Configuring done
+-- Generating done
+-- Build files have been written to: /home/wangyue50/CINN-my/CINN/build/thirds/glog/src/extern_glog-build
+[ 25%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/api.pb.cc.o
+[ 18%] Performing build step for 'extern_glog'
+Scanning dependencies of target glog
+[ 12%] Building CXX object CMakeFiles/glog.dir/src/demangle.cc.o
+[ 25%] Building CXX object CMakeFiles/glog.dir/src/logging.cc.o
+[ 25%] Linking CXX static library libprotobuf-lite.a
+[ 25%] Built target libprotobuf-lite
+[ 37%] Building CXX object CMakeFiles/glog.dir/src/raw_logging.cc.o
+[ 50%] Building CXX object CMakeFiles/glog.dir/src/symbolize.cc.o
+[ 75%] Building CXX object CMakeFiles/glog.dir/src/utilities.cc.o
+[ 75%] Building CXX object CMakeFiles/glog.dir/src/vlog_is_on.cc.o
+[ 25%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/importer.cc.o
+[ 26%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/parser.cc.o
+[ 87%] Building CXX object CMakeFiles/glog.dir/src/signalhandler.cc.o
+[ 26%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/descriptor.cc.o
+[ 27%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/descriptor.pb.cc.o
+[ 27%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/descriptor_database.cc.o
+[ 28%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/duration.pb.cc.o
+[ 28%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/dynamic_message.cc.o
+[100%] Linking CXX static library libglog.a
+[100%] Built target glog
+[ 20%] Performing install step for 'extern_glog'
+[100%] Built target glog
+Install the project...
+-- Install configuration: ""
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/lib/libglog.a
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/include/glog/config.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/include/glog/logging.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/include/glog/raw_logging.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/include/glog/stl_logging.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/include/glog/vlog_is_on.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/include/glog/log_severity.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/lib/cmake/glog/glog-config.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/lib/cmake/glog/glog-config-version.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/lib/cmake/glog/glog-targets.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/glog/lib/cmake/glog/glog-targets-noconfig.cmake
+[ 20%] Completed 'extern_glog'
+[ 20%] Built target extern_glog
+[ 29%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/empty.pb.cc.o
+Scanning dependencies of target cinn_gtest_main
+[ 20%] Building CXX object cinn/CMakeFiles/cinn_gtest_main.dir/gtest_main.cc.o
+[ 29%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/extension_set_heavy.cc.o
+[ 30%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/field_mask.pb.cc.o
+[ 30%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/generated_message_reflection.cc.o
+[ 20%] Linking CXX static library libcinn_gtest_main.a
+[ 20%] Built target cinn_gtest_main
+[ 31%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/gzip_stream.cc.o
+[ 31%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/printer.cc.o
+[ 32%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/strtod.cc.o
+[ 32%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/tokenizer.cc.o
+[ 33%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/io/zero_copy_stream_impl.cc.o
+[ 33%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/map_field.cc.o
+[ 34%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/message.cc.o
+[ 34%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/reflection_ops.cc.o
+[ 35%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/service.cc.o
+[ 35%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/source_context.pb.cc.o
+[ 36%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/struct.pb.cc.o
+[ 37%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/mathlimits.cc.o
+[ 37%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/stubs/substitute.cc.o
+[ 38%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/text_format.cc.o
+[ 38%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/timestamp.pb.cc.o
+[ 39%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/type.pb.cc.o
+[ 39%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/unknown_field_set.cc.o
+[ 40%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/field_comparator.cc.o
+[ 40%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/field_mask_util.cc.o
+[ 41%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/datapiece.cc.o
+[ 41%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/default_value_objectwriter.cc.o
+[ 42%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/error_listener.cc.o
+[ 42%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/field_mask_utility.cc.o
+[ 43%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/json_escaping.cc.o
+[ 43%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/json_objectwriter.cc.o
+[ 44%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/json_stream_parser.cc.o
+[ 44%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/object_writer.cc.o
+[ 45%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/proto_writer.cc.o
+[ 45%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/protostream_objectsource.cc.o
+[ 46%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/protostream_objectwriter.cc.o
+[ 46%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/type_info.cc.o
+[ 47%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/type_info_test_helper.cc.o
+[ 47%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/internal/utility.cc.o
+[ 48%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/json_util.cc.o
+[ 48%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/message_differencer.cc.o
+[ 49%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/time_util.cc.o
+[ 49%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/util/type_resolver_util.cc.o
+[ 50%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/wire_format.cc.o
+[ 50%] Building CXX object CMakeFiles/libprotobuf.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/wrappers.pb.cc.o
+[ 51%] Linking CXX static library libprotobuf.a
+[ 51%] Built target libprotobuf
+Scanning dependencies of target libprotoc
+[ 52%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/code_generator.cc.o
+[ 52%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/command_line_interface.cc.o
+[ 52%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_enum_field.cc.o
+[ 54%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_extension.cc.o
+[ 54%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_field.cc.o
+[ 54%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_enum.cc.o
+[ 55%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_file.cc.o
+[ 55%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_generator.cc.o
+[ 56%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_helpers.cc.o
+[ 56%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_map_field.cc.o
+[ 57%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_message.cc.o
+[ 57%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_message_field.cc.o
+[ 58%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_primitive_field.cc.o
+[ 58%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_service.cc.o
+[ 59%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/cpp/cpp_string_field.cc.o
+[ 59%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_doc_comment.cc.o
+[ 60%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_enum.cc.o
+[ 60%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_enum_field.cc.o
+[ 61%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_field_base.cc.o
+[ 61%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_generator.cc.o
+[ 62%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_helpers.cc.o
+[ 62%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_map_field.cc.o
+[ 63%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_message.cc.o
+[ 63%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_message_field.cc.o
+[ 64%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_primitive_field.cc.o
+[ 64%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_reflection_class.cc.o
+[ 65%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_repeated_enum_field.cc.o
+[ 65%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_repeated_message_field.cc.o
+[ 66%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_repeated_primitive_field.cc.o
+[ 66%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_source_generator_base.cc.o
+[ 67%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/csharp/csharp_wrapper_field.cc.o
+[ 67%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_context.cc.o
+[ 68%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_doc_comment.cc.o
+[ 68%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_enum.cc.o
+[ 69%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_enum_field.cc.o
+[ 69%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_enum_field_lite.cc.o
+[ 70%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_enum_lite.cc.o
+[ 70%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_extension.cc.o
+[ 71%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_extension_lite.cc.o
+[ 71%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_field.cc.o
+[ 72%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_file.cc.o
+[ 72%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_generator.cc.o
+[ 73%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_helpers.cc.o
+[ 73%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_generator_factory.cc.o
+[ 74%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_lazy_message_field.cc.o
+[ 75%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_lazy_message_field_lite.cc.o
+[ 75%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_map_field.cc.o
+[ 76%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_map_field_lite.cc.o
+[ 76%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_message.cc.o
+[ 77%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_message_builder.cc.o
+[ 77%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_message_builder_lite.cc.o
+[ 78%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_message_field.cc.o
+[ 78%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_message_field_lite.cc.o
+[ 78%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_name_resolver.cc.o
+[ 79%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_message_lite.cc.o
+[ 80%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_primitive_field.cc.o
+[ 80%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_primitive_field_lite.cc.o
+[ 81%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_service.cc.o
+[ 81%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_shared_code_generator.cc.o
+[ 82%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_string_field.cc.o
+[ 82%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/java/java_string_field_lite.cc.o
+[ 83%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_enum.cc.o
+[ 83%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_enum_field.cc.o
+[ 84%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_extension.cc.o
+[ 84%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_field.cc.o
+[ 85%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_file.cc.o
+[ 85%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_generator.cc.o
+[ 86%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_helpers.cc.o
+[ 86%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_map_field.cc.o
+[ 87%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_message.cc.o
+[ 87%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_message_field.cc.o
+[ 88%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/javanano/javanano_primitive_field.cc.o
+[ 88%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/js/js_generator.cc.o
+[ 89%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_enum.cc.o
+[ 89%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_enum_field.cc.o
+[ 90%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_extension.cc.o
+[ 90%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_field.cc.o
+[ 91%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_file.cc.o
+[ 91%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_generator.cc.o
+[ 92%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_helpers.cc.o
+[ 92%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_map_field.cc.o
+[ 93%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_message.cc.o
+[ 93%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_message_field.cc.o
+[ 94%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_oneof.cc.o
+[ 94%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/objectivec/objectivec_primitive_field.cc.o
+[ 95%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/php/php_generator.cc.o
+[ 95%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/plugin.cc.o
+[ 96%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/plugin.pb.cc.o
+[ 96%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/python/python_generator.cc.o
+[ 97%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/ruby/ruby_generator.cc.o
+[ 97%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/subprocess.cc.o
+[ 98%] Building CXX object CMakeFiles/libprotoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/zip_writer.cc.o
+[ 98%] Linking CXX static library libprotoc.a
+[ 98%] Built target libprotoc
+Scanning dependencies of target protoc
+[ 99%] Building CXX object CMakeFiles/protoc.dir/home/wangyue50/CINN-my/CINN/build/thirds/protobuf/src/extern_protobuf/src/google/protobuf/compiler/main.cc.o
+[100%] Linking CXX executable protoc
+[100%] Built target protoc
+[ 20%] Performing install step for 'extern_protobuf'
+[ 12%] Built target libprotobuf-lite
+[ 51%] Built target libprotobuf
+[ 98%] Built target libprotoc
+[100%] Built target protoc
+Install the project...
+-- Install configuration: ""
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/libprotobuf-lite.a
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/libprotobuf.a
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/libprotoc.a
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/bin/protoc
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/any.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/any.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/api.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/arena.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/arenastring.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/code_generator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/command_line_interface.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/cpp/cpp_generator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/csharp/csharp_generator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/csharp/csharp_names.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/importer.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/java/java_generator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/java/java_names.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/javanano/javanano_generator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/js/js_generator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/objectivec/objectivec_generator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/objectivec/objectivec_helpers.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/parser.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/php/php_generator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/plugin.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/plugin.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/python/python_generator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/ruby/ruby_generator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/descriptor.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/descriptor.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/descriptor_database.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/duration.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/dynamic_message.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/empty.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/extension_set.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/field_mask.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/generated_enum_reflection.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/generated_enum_util.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/generated_message_reflection.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/generated_message_util.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/has_bits.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/io/coded_stream.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/io/gzip_stream.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/io/printer.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/io/strtod.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/io/tokenizer.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/io/zero_copy_stream.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/io/zero_copy_stream_impl.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/io/zero_copy_stream_impl_lite.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/map.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/map_entry.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/map_entry_lite.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/map_field.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/map_field_inl.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/map_field_lite.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/map_type_handler.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/message.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/message_lite.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/metadata.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/reflection.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/reflection_ops.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/repeated_field.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/service.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/source_context.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/struct.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomic_sequence_num.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_arm64_gcc.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_arm_gcc.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_arm_qnx.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_atomicword_compat.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_generic_c11_atomic.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_generic_gcc.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_macosx.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_mips_gcc.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_power.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_ppc_gcc.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_solaris.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_tsan.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_x86_gcc.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/atomicops_internals_x86_msvc.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/bytestream.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/callback.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/casts.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/common.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/fastmem.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/hash.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/logging.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/macros.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/mutex.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/once.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/platform_macros.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/port.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/scoped_ptr.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/shared_ptr.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/singleton.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/status.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/stl_util.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/stringpiece.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/template_util.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/stubs/type_traits.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/text_format.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/timestamp.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/type.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/unknown_field_set.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/util/field_comparator.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/util/field_mask_util.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/util/json_util.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/util/message_differencer.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/util/time_util.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/util/type_resolver.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/util/type_resolver_util.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/wire_format.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/wire_format_lite.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/wire_format_lite_inl.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/wrappers.pb.h
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/descriptor.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/any.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/api.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/duration.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/empty.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/field_mask.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/source_context.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/struct.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/timestamp.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/type.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/wrappers.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/include/google/protobuf/compiler/plugin.proto
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/cmake/protobuf/protobuf-targets.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/cmake/protobuf/protobuf-targets-noconfig.cmake
+-- Up-to-date: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/cmake/protobuf
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/cmake/protobuf/protobuf-config-version.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/cmake/protobuf/protobuf-config.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/cmake/protobuf/protobuf-module.cmake
+-- Installing: /home/wangyue50/CINN-my/CINN/build/thirds/install/protobuf/lib/cmake/protobuf/protobuf-options.cmake
+[ 22%] Completed 'extern_protobuf'
+[ 22%] Built target extern_protobuf
+[ 24%] Running C++ protocol buffer compiler on framework.proto
+Scanning dependencies of target framework_proto
+[ 24%] Building CXX object cinn/frontend/paddle/CMakeFiles/framework_proto.dir/framework.pb.cc.o
+[ 24%] Linking CXX static library libframework_proto.a
+[ 24%] Built target framework_proto
+Scanning dependencies of target core
+[ 24%] Building CXX object CMakeFiles/core.dir/cinn/common/shared.cc.o
+[ 24%] Building CXX object CMakeFiles/core.dir/cinn/common/target.cc.o
+[ 24%] Building CXX object CMakeFiles/core.dir/cinn/common/type.cc.o
+[ 24%] Building CXX object CMakeFiles/core.dir/cinn/common/info_registry.cc.o
+[ 24%] Building CXX object CMakeFiles/core.dir/cinn/common/graph_utils.cc.o
+[ 24%] Building CXX object CMakeFiles/core.dir/cinn/common/debug_manager.cc.o
+[ 26%] Building CXX object CMakeFiles/core.dir/cinn/common/cinn_value.cc.o
+[ 28%] Building CXX object CMakeFiles/core.dir/cinn/common/object.cc.o
+[ 30%] Building CXX object CMakeFiles/core.dir/cinn/common/context.cc.o
+[ 30%] Building CXX object CMakeFiles/core.dir/cinn/common/axis.cc.o
+[ 30%] Building CXX object CMakeFiles/core.dir/cinn/common/ir_util.cc.o
+[ 32%] Building CXX object CMakeFiles/core.dir/cinn/common/test_helper.cc.o
+[ 32%] Building CXX object CMakeFiles/core.dir/cinn/common/arithmatic.cc.o
+[ 32%] Building CXX object CMakeFiles/core.dir/cinn/common/cas.cc.o
+[ 32%] Building CXX object CMakeFiles/core.dir/cinn/common/union_find.cc.o
+[ 33%] Building CXX object CMakeFiles/core.dir/cinn/utils/string.cc.o
+[ 33%] Building CXX object CMakeFiles/core.dir/cinn/utils/functional.cc.o
+[ 33%] Building CXX object CMakeFiles/core.dir/cinn/utils/dot_lang.cc.o
+[ 35%] Building CXX object CMakeFiles/core.dir/cinn/utils/timer.cc.o
+[ 35%] Building CXX object CMakeFiles/core.dir/cinn/poly/dim.cc.o
+[ 35%] Building CXX object CMakeFiles/core.dir/cinn/poly/domain.cc.o
+[ 35%] Building CXX object CMakeFiles/core.dir/cinn/poly/map.cc.o
+[ 37%] Building CXX object CMakeFiles/core.dir/cinn/poly/stage.cc.o
+[ 37%] Building CXX object CMakeFiles/core.dir/cinn/poly/isl_utils.cc.o
+[ 37%] Building CXX object CMakeFiles/core.dir/cinn/poly/schedule.cc.o
+[ 37%] Building CXX object CMakeFiles/core.dir/cinn/poly/poly_scheduler.cc.o
+[ 39%] Building CXX object CMakeFiles/core.dir/cinn/poly/naive_scheduler.cc.o
+[ 39%] Building CXX object CMakeFiles/core.dir/cinn/poly/ast_gen.cc.o
+[ 39%] Building CXX object CMakeFiles/core.dir/cinn/poly/graph.cc.o
+[ 41%] Building CXX object CMakeFiles/core.dir/cinn/poly/compute_at_transform.cc.o
+[ 41%] Building CXX object CMakeFiles/core.dir/cinn/runtime/cpu/host_intrinsics.cc.o
+[ 41%] Building CXX object CMakeFiles/core.dir/cinn/runtime/cpu/mkl_math.cc.o
+[ 43%] Building CXX object CMakeFiles/core.dir/cinn/runtime/cpu/cblas.cc.o
+[ 43%] Building CXX object CMakeFiles/core.dir/cinn/runtime/intrinsic.cc.o
+[ 43%] Building CXX object CMakeFiles/core.dir/cinn/runtime/cinn_runtime.cc.o
+[ 45%] Building CXX object CMakeFiles/core.dir/cinn/runtime/intrinsic_types.cc.o
+[ 45%] Building CXX object CMakeFiles/core.dir/cinn/ir/node.cc.o
+[ 45%] Building CXX object CMakeFiles/core.dir/cinn/ir/ir.cc.o
+[ 45%] Building CXX object CMakeFiles/core.dir/cinn/ir/ir_visitor.cc.o
+[ 47%] Building CXX object CMakeFiles/core.dir/cinn/ir/ir_printer.cc.o
+[ 47%] Building CXX object CMakeFiles/core.dir/cinn/ir/ir_mutator.cc.o
+[ 47%] Building CXX object CMakeFiles/core.dir/cinn/ir/function_definition.cc.o
+[ 49%] Building CXX object CMakeFiles/core.dir/cinn/ir/lowered_func.cc.o
+[ 49%] Building CXX object CMakeFiles/core.dir/cinn/ir/ir_operators.cc.o
+[ 49%] Building CXX object CMakeFiles/core.dir/cinn/ir/buffer.cc.o
+[ 49%] Building CXX object CMakeFiles/core.dir/cinn/ir/function_base.cc.o
+[ 50%] Building CXX object CMakeFiles/core.dir/cinn/ir/operation.cc.o
+[ 50%] Building CXX object CMakeFiles/core.dir/cinn/ir/collect_ir_nodes.cc.o
+[ 50%] Building CXX object CMakeFiles/core.dir/cinn/ir/registry.cc.o
+[ 52%] Building CXX object CMakeFiles/core.dir/cinn/ir/tensor.cc.o
+[ 52%] Building CXX object CMakeFiles/core.dir/cinn/backends/llvm/llvm_util.cc.o
+[ 52%] Building CXX object CMakeFiles/core.dir/cinn/backends/llvm/runtime_symbol_registry.cc.o
+[ 52%] Building CXX object CMakeFiles/core.dir/cinn/backends/llvm/codegen_llvm.cc.o
+[ 54%] Building CXX object CMakeFiles/core.dir/cinn/backends/llvm/codegen_x86.cc.o
+[ 54%] Building CXX object CMakeFiles/core.dir/cinn/backends/llvm/simple_jit.cc.o
+[ 54%] Building CXX object CMakeFiles/core.dir/cinn/backends/llvm/execution_engine.cc.o
+[ 56%] Building CXX object CMakeFiles/core.dir/cinn/backends/llvm/llvm_optimizer.cc.o
+[ 56%] Building CXX object CMakeFiles/core.dir/cinn/backends/outputs.cc.o
+[ 56%] Building CXX object CMakeFiles/core.dir/cinn/backends/codegen_c.cc.o
+[ 56%] Building CXX object CMakeFiles/core.dir/cinn/backends/codegen_c_x86.cc.o
+[ 58%] Building CXX object CMakeFiles/core.dir/cinn/backends/codegen_c_x86_test.cc.o
+[ 58%] Building CXX object CMakeFiles/core.dir/cinn/backends/codegen_cuda_dev.cc.o
+[ 58%] Building CXX object CMakeFiles/core.dir/cinn/backends/codegen_cuda_host.cc.o
+[ 60%] Building CXX object CMakeFiles/core.dir/cinn/backends/extern_func_emitter.cc.o
+[ 60%] Building CXX object CMakeFiles/core.dir/cinn/backends/extern_func_emitter_builtin.cc.o
+[ 60%] Building CXX object CMakeFiles/core.dir/cinn/backends/function_prototype.cc.o
+[ 60%] Building CXX object CMakeFiles/core.dir/cinn/backends/extern_func_protos.cc.o
+[ 62%] Building CXX object CMakeFiles/core.dir/cinn/backends/extern_func_jit_register.cc.o
+[ 62%] Building CXX object CMakeFiles/core.dir/cinn/backends/modular.cc.o
+[ 62%] Building CXX object CMakeFiles/core.dir/cinn/backends/compiler.cc.o
+[ 64%] Building CXX object CMakeFiles/core.dir/cinn/lang/buffer.cc.o
+[ 64%] Building CXX object CMakeFiles/core.dir/cinn/lang/compute.cc.o
+[ 64%] Building CXX object CMakeFiles/core.dir/cinn/lang/placeholder.cc.o
+[ 64%] Building CXX object CMakeFiles/core.dir/cinn/lang/module.cc.o
+[ 66%] Building CXX object CMakeFiles/core.dir/cinn/lang/lower.cc.o
+[ 66%] Building CXX object CMakeFiles/core.dir/cinn/lang/builtin.cc.o
+[ 66%] Building CXX object CMakeFiles/core.dir/cinn/lang/lower_impl.cc.o
+[ 67%] Building CXX object CMakeFiles/core.dir/cinn/lang/compute_at_postprocess.cc.o
+[ 67%] Building CXX object CMakeFiles/core.dir/cinn/lang/packed_func.cc.o
+[ 67%] Building CXX object CMakeFiles/core.dir/cinn/optim/remove_nested_block.cc.o
+[ 69%] Building CXX object CMakeFiles/core.dir/cinn/optim/replace_call_with_expr.cc.o
+[ 69%] Building CXX object CMakeFiles/core.dir/cinn/optim/ir_copy.cc.o
+[ 69%] Building CXX object CMakeFiles/core.dir/cinn/optim/ir_replace.cc.o
+[ 69%] Building CXX object CMakeFiles/core.dir/cinn/optim/replace_var_with_expr.cc.o
+[ 71%] Building CXX object CMakeFiles/core.dir/cinn/optim/tensor_write_tell.cc.o
+[ 71%] Building CXX object CMakeFiles/core.dir/cinn/optim/ir_simplify.cc.o
+[ 71%] Building CXX object CMakeFiles/core.dir/cinn/optim/optimize.cc.o
+[ 73%] Building CXX object CMakeFiles/core.dir/cinn/optim/vectorize_loops.cc.o
+[ 73%] Building CXX object CMakeFiles/core.dir/cinn/optim/unroll_loops.cc.o
+[ 73%] Building CXX object CMakeFiles/core.dir/cinn/optim/transform_polyfor_to_for.cc.o
+[ 73%] Building CXX object CMakeFiles/core.dir/cinn/optim/eliminate_broadcast_in_forloop.cc.o
+[ 75%] Building CXX object CMakeFiles/core.dir/cinn/optim/fold_cinn_call_arguments.cc.o
+[ 75%] Building CXX object CMakeFiles/core.dir/cinn/optim/call_arg_list_to_pod_value.cc.o
+[ 75%] Building CXX object CMakeFiles/core.dir/cinn/optim/insert_debug_log_callee.cc.o
+[ 77%] Building CXX object CMakeFiles/core.dir/cinn/optim/lower_function_call_bind_vars.cc.o
+[ 77%] Building CXX object CMakeFiles/core.dir/cinn/optim/extern_call_process.cc.o
+[ 77%] Building CXX object CMakeFiles/core.dir/cinn/optim/map_extern_call.cc.o
+[ 77%] Building CXX object CMakeFiles/core.dir/cinn/optim/cache_read_write_replace.cc.o
+[ 79%] Building CXX object CMakeFiles/core.dir/cinn/optim/compute_inline_expand.cc.o
+[ 79%] Building CXX object CMakeFiles/core.dir/cinn/optim/buffer_assign.cc.o
+[ 79%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/tensor.cc.o
+[ 81%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/scope.cc.o
+[ 81%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/variable.cc.o
+[ 81%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/buffer.cc.o
+[ 81%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/memory.cc.o
+[ 83%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/instruction.cc.o
+[ 83%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/graph_compiler.cc.o
+[ 83%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/graph.cc.o
+[ 84%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/node.cc.o
+[ 84%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/pass.cc.o
+[ 84%] Building CXX object CMakeFiles/core.dir/cinn/hlir/framework/op_strategy.cc.o
+[ 84%] Building CXX object CMakeFiles/core.dir/cinn/hlir/pe/broadcast.cc.o
+[ 86%] Building CXX object CMakeFiles/core.dir/cinn/hlir/pe/elementwise.cc.o
+[ 86%] Building CXX object CMakeFiles/core.dir/cinn/hlir/pe/nn.cc.o
+[ 86%] Building CXX object CMakeFiles/core.dir/cinn/hlir/pe/reduction.cc.o
+[ 88%] Building CXX object CMakeFiles/core.dir/cinn/hlir/pe/schedule.cc.o
+[ 88%] Building CXX object CMakeFiles/core.dir/cinn/hlir/pe/transform.cc.o
+[ 88%] Building CXX object CMakeFiles/core.dir/cinn/hlir/pe/vision.cc.o
+[ 88%] Building CXX object CMakeFiles/core.dir/cinn/hlir/op/nn.cc.o
+[ 90%] Building CXX object CMakeFiles/core.dir/cinn/hlir/op/broadcast.cc.o
+[ 90%] Building CXX object CMakeFiles/core.dir/cinn/hlir/op/transform.cc.o
+[ 90%] Building CXX object CMakeFiles/core.dir/cinn/hlir/pass/infershape.cc.o
+[ 92%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle/cpp/var_desc.cc.o
+[ 92%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle/cpp/op_desc.cc.o
+[ 92%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle/cpp/block_desc.cc.o
+[ 94%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle/cpp/program_desc.cc.o
+[ 94%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle/pb/var_desc.cc.o
+[ 94%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle/pb/op_desc.cc.o
+[ 94%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle/pb/block_desc.cc.o
+[ 96%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle/pb/program_desc.cc.o
+[ 96%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle/model_parser.cc.o
+[ 96%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle/compatible_pb.cc.o
+[ 98%] Building CXX object CMakeFiles/core.dir/cinn/frontend/syntax.cc.o
+[ 98%] Building CXX object CMakeFiles/core.dir/cinn/frontend/paddle_model_to_program.cc.o
+[ 98%] Building CXX object CMakeFiles/core.dir/cinn/frontend/executor.cc.o
+[ 98%] Linking CXX static library libcore.a
+[ 98%] Built target core
+Scanning dependencies of target test01_elementwise_add_main
+[ 98%] Building CXX object tests/CMakeFiles/test01_elementwise_add_main.dir/test01_elementwise_add_main.cc.o
+[100%] Linking CXX executable test01_elementwise_add_main
+[100%] Built target test01_elementwise_add_main
+[  5%] Built target extern_protobuf
+[  9%] Built target extern_gtest
+[ 16%] Built target extern_mklml
+[ 16%] Built target extern_gflags
+[ 16%] Built target GEN_LLVM_RUNTIME_IR_HEADER
+[ 22%] Built target extern_glog
+[ 22%] Built target cinn_gtest_main
+[ 24%] Built target framework_proto
+[ 98%] Built target core
+Scanning dependencies of target test02_matmul_main
+[100%] Building CXX object tests/CMakeFiles/test02_matmul_main.dir/test02_matmul_main.cc.o
+[100%] Linking CXX executable test02_matmul_main
+[100%] Built target test02_matmul_main
+[  5%] Built target extern_protobuf
+[  9%] Built target extern_gflags
+[  9%] Built target GEN_LLVM_RUNTIME_IR_HEADER
+[ 13%] Built target extern_mklml
+[ 17%] Built target extern_gtest
+[ 23%] Built target extern_glog
+[ 23%] Built target cinn_gtest_main
+[ 25%] Built target framework_proto
+[100%] Built target core
+Scanning dependencies of target test03_conv_main
+[100%] Building CXX object tests/CMakeFiles/test03_conv_main.dir/test03_convolution_main.cc.o
+[100%] Linking CXX executable test03_conv_main
+[100%] Built target test03_conv_main
+Test project /home/wangyue50/CINN-my/CINN/build
+    Start 51: test01_elementwise_add_main
+1/1 Test #51: test01_elementwise_add_main ......   Passed    0.09 sec
+
+100% tests passed, 0 tests failed out of 1
+
+Total Test time (real) =   0.09 sec
+Test project /home/wangyue50/CINN-my/CINN/build
+    Start 53: test02_matmul_main
+1/1 Test #53: test02_matmul_main ...............   Passed    0.52 sec
+
+100% tests passed, 0 tests failed out of 1
+
+Total Test time (real) =   0.52 sec
+Test project /home/wangyue50/CINN-my/CINN/build
+    Start 55: test03_conv_main
+1/1 Test #55: test03_conv_main .................   Passed    0.13 sec
+
+100% tests passed, 0 tests failed out of 1
+
+Total Test time (real) =   0.13 sec
+Scanning dependencies of target extern_pybind
+Scanning dependencies of target extern_lite_download_lite_naive_model_tar_gz
+[  2%] Built target extern_mklml
+[  5%] Built target extern_gflags
+[  5%] Creating directories for 'extern_pybind'
+[  5%] Built target GEN_LLVM_RUNTIME_IR_HEADER
+[  8%] Built target extern_protobuf
+[ 10%] Creating directories for 'extern_lite_download_lite_naive_model_tar_gz'
+[ 11%] Built target extern_gtest
+[ 14%] Built target extern_glog
+Scanning dependencies of target __x86_source_fake_lib
+Scanning dependencies of target cinn_runtime
+[ 14%] Building CXX object cinn/backends/CMakeFiles/__x86_source_fake_lib.dir/_x86_builtin_source.cc.o
+[ 15%] Built target cinn_gtest_main
+[ 15%] Built target framework_proto
+[ 16%] Building CXX object cinn/runtime/CMakeFiles/cinn_runtime.dir/buffer.cc.o
+[ 16%] Building CXX object cinn/runtime/CMakeFiles/cinn_runtime.dir/cinn_runtime.cc.o
+[ 16%] Performing download step (git clone) for 'extern_pybind'
+[ 16%] Performing download step for 'extern_lite_download_lite_naive_model_tar_gz'
+[ 55%] Built target core
+Scanning dependencies of target test_shared
+Scanning dependencies of target test_graph_utils
+Scanning dependencies of target test_cinn_value
+[ 55%] Building CXX object cinn/common/CMakeFiles/test_shared.dir/shared_test.cc.o
+[ 55%] Building CXX object cinn/common/CMakeFiles/test_graph_utils.dir/graph_utils_test.cc.o
+[ 55%] Building CXX object cinn/common/CMakeFiles/test_cinn_value.dir/cinn_value_test.cc.o
+Scanning dependencies of target test_arithmatic
+[ 56%] Building CXX object cinn/common/CMakeFiles/test_arithmatic.dir/arithmatic_test.cc.o
+[ 57%] Linking CXX static library lib__x86_source_fake_lib.a
+[ 57%] Linking CXX static library libcinn_runtime.a
+[ 57%] Built target __x86_source_fake_lib
+Scanning dependencies of target test_cas
+[ 57%] Built target cinn_runtime
+[ 57%] Building CXX object cinn/common/CMakeFiles/test_cas.dir/cas_test.cc.o
+Scanning dependencies of target test_string
+[ 57%] Building CXX object cinn/utils/CMakeFiles/test_string.dir/string_test.cc.o
+[ 58%] Linking CXX executable test_shared
+[ 59%] No patch step for 'extern_lite_download_lite_naive_model_tar_gz'
+[ 59%] No update step for 'extern_lite_download_lite_naive_model_tar_gz'
+[ 60%] Linking CXX executable test_string
+[ 60%] No configure step for 'extern_lite_download_lite_naive_model_tar_gz'
+[ 60%] No build step for 'extern_lite_download_lite_naive_model_tar_gz'
+[ 60%] No install step for 'extern_lite_download_lite_naive_model_tar_gz'
+[ 60%] Completed 'extern_lite_download_lite_naive_model_tar_gz'
+[ 60%] Built target extern_lite_download_lite_naive_model_tar_gz
+Scanning dependencies of target test_schedule
+[ 61%] Building CXX object cinn/poly/CMakeFiles/test_schedule.dir/schedule_test.cc.o
+[ 61%] Built target test_shared
+Scanning dependencies of target test_compute_at_transform
+[ 61%] Building CXX object cinn/poly/CMakeFiles/test_compute_at_transform.dir/compute_at_transform_test.cc.o
+[ 61%] Built target test_string
+Scanning dependencies of target test_stage
+[ 61%] Building CXX object cinn/poly/CMakeFiles/test_stage.dir/stage_test.cc.o
+[ 61%] Linking CXX executable test_graph_utils
+[ 62%] Linking CXX executable test_cinn_value
+[ 62%] Linking CXX executable test_arithmatic
+[ 63%] Linking CXX executable test_cas
+[ 63%] Linking CXX executable test_compute_at_transform
+[ 63%] Linking CXX executable test_schedule
+[ 63%] Built target test_graph_utils
+Scanning dependencies of target test_cinn_runtime
+[ 63%] Building CXX object cinn/runtime/CMakeFiles/test_cinn_runtime.dir/cinn_runtime_test.cc.o
+[ 63%] Built target test_cinn_value
+Scanning dependencies of target test_mkl_math
+[ 64%] Building CXX object cinn/runtime/cpu/CMakeFiles/test_mkl_math.dir/mkl_math_test.cc.o
+[ 64%] Linking CXX executable test_cinn_runtime
+[ 64%] Built target test_arithmatic
+[ 64%] Building CXX object cinn/runtime/cpu/CMakeFiles/test_mkl_math.dir/mkl_math.cc.o
+[ 64%] Built target test_cinn_runtime
+Scanning dependencies of target test_host_intrinsics
+[ 65%] Building CXX object cinn/runtime/cpu/CMakeFiles/test_host_intrinsics.dir/host_intrinsics_test.cc.o
+[ 65%] Built target test_compute_at_transform
+[ 65%] Built target test_cas
+Scanning dependencies of target test_tensor
+Scanning dependencies of target test_collect_ir_nodes
+[ 65%] Building CXX object cinn/ir/CMakeFiles/test_tensor.dir/tensor_test.cc.o
+[ 65%] Building CXX object cinn/ir/CMakeFiles/test_collect_ir_nodes.dir/collect_ir_nodes_test.cc.o
+Submodule path 'tools/clang': checked out '6a00cbc4a9b8e68b71caf7f774b3f9c753ae84d5'
+[ 65%] No patch step for 'extern_pybind'
+[ 65%] No update step for 'extern_pybind'
+[ 66%] No configure step for 'extern_pybind'
+[ 66%] No build step for 'extern_pybind'
+[ 66%] No install step for 'extern_pybind'
+[ 66%] No test step for 'extern_pybind'
+[ 66%] Completed 'extern_pybind'
+[ 66%] Built target test_schedule
+Scanning dependencies of target test_buffer
+[ 66%] Built target extern_pybind
+[ 66%] Building CXX object cinn/ir/CMakeFiles/test_buffer.dir/buffer_test.cc.o
+Scanning dependencies of target test_codegen_c
+[ 66%] Building CXX object cinn/backends/CMakeFiles/test_codegen_c.dir/codegen_c_test.cc.o
+[ 67%] Linking CXX executable test_collect_ir_nodes
+[ 67%] Linking CXX executable test_buffer
+Scanning dependencies of target test_codegen_c_x86
+[ 68%] Building CXX object cinn/backends/CMakeFiles/test_codegen_c_x86.dir/codegen_c_x86_test.cc.o
+[ 68%] Built target test_collect_ir_nodes
+Scanning dependencies of target test_compiler
+[ 68%] Building CXX object cinn/backends/CMakeFiles/test_compiler.dir/compiler_test.cc.o
+[ 68%] Built target test_buffer
+Scanning dependencies of target test_codegen_llvm
+[ 68%] Building CXX object cinn/backends/llvm/CMakeFiles/test_codegen_llvm.dir/codegen_llvm_test.cc.o
+[ 68%] Linking CXX executable test_codegen_c_x86
+[ 68%] Linking CXX executable test_stage
+[ 68%] Linking CXX executable test_codegen_c
+[ 68%] Built target test_codegen_c_x86
+Scanning dependencies of target test_execution_engine
+[ 69%] Building CXX object cinn/backends/llvm/CMakeFiles/test_execution_engine.dir/execution_engine_test.cc.o
+[ 69%] Built target test_stage
+[ 69%] Linking CXX executable test_mkl_math
+[ 69%] Built target test_codegen_c
+Scanning dependencies of target test_placeholder
+Scanning dependencies of target test_codegen_x86
+[ 69%] Building CXX object cinn/lang/CMakeFiles/test_placeholder.dir/placeholder_test.cc.o
+[ 69%] Building CXX object cinn/backends/llvm/CMakeFiles/test_codegen_x86.dir/codegen_x86_test.cc.o
+[ 69%] Linking CXX executable test_tensor
+[ 69%] Linking CXX executable test_host_intrinsics
+[ 69%] Linking CXX executable test_placeholder
+[ 69%] Built target test_tensor
+[ 69%] Built target test_mkl_math
+Scanning dependencies of target test_compute
+Scanning dependencies of target test_lower
+[ 70%] Building CXX object cinn/lang/CMakeFiles/test_compute.dir/compute_test.cc.o
+[ 70%] Building CXX object cinn/lang/CMakeFiles/test_lower.dir/lower_test.cc.o
+[ 71%] Linking CXX executable test_codegen_llvm
+[ 71%] Built target test_host_intrinsics
+Scanning dependencies of target test_lower_impl
+[ 71%] Building CXX object cinn/lang/CMakeFiles/test_lower_impl.dir/lower_impl_test.cc.o
+[ 71%] Built target test_placeholder
+Scanning dependencies of target test_packed_func
+[ 72%] Building CXX object cinn/lang/CMakeFiles/test_packed_func.dir/packed_func_test.cc.o
+[ 72%] Linking CXX executable test_compute
+[ 72%] Built target test_codegen_llvm
+Scanning dependencies of target test_cache_read_write_replace
+[ 73%] Building CXX object cinn/optim/CMakeFiles/test_cache_read_write_replace.dir/cache_read_write_replace_test.cc.o
+[ 74%] Linking CXX executable test_lower
+[ 74%] Linking CXX executable test_compiler
+[ 74%] Linking CXX executable test_lower_impl
+[ 74%] Linking CXX executable test_packed_func
+[ 74%] Built target test_compute
+Scanning dependencies of target test_optimize
+[ 74%] Building CXX object cinn/optim/CMakeFiles/test_optimize.dir/optimize_test.cc.o
+[ 74%] Linking CXX executable test_cache_read_write_replace
+[ 74%] Built target test_lower
+Scanning dependencies of target test_ir_copy
+[ 74%] Building CXX object cinn/optim/CMakeFiles/test_ir_copy.dir/ir_copy_test.cc.o
+[ 74%] Built target test_packed_func
+Scanning dependencies of target test_remove_nested_block
+[ 75%] Building CXX object cinn/optim/CMakeFiles/test_remove_nested_block.dir/remove_nested_block_test.cc.o
+[ 75%] Built target test_compiler
+Scanning dependencies of target test_replace_call_with_expr
+[ 75%] Building CXX object cinn/optim/CMakeFiles/test_replace_call_with_expr.dir/replace_call_with_expr_test.cc.o
+[ 75%] Linking CXX executable test_codegen_x86
+[ 75%] Linking CXX executable test_optimize
+[ 75%] Built target test_lower_impl
+Scanning dependencies of target test_ir_simplify
+[ 76%] Building CXX object cinn/optim/CMakeFiles/test_ir_simplify.dir/ir_simplify_test.cc.o
+[ 76%] Built target test_cache_read_write_replace
+Scanning dependencies of target test_vectorize_loops
+[ 76%] Building CXX object cinn/optim/CMakeFiles/test_vectorize_loops.dir/vectorize_loops_test.cc.o
+[ 76%] Linking CXX executable test_execution_engine
+[ 76%] Linking CXX executable test_ir_copy
+[ 76%] Linking CXX executable test_remove_nested_block
+[ 76%] Linking CXX executable test_replace_call_with_expr
+[ 76%] Built target test_codegen_x86
+[ 76%] Linking CXX executable test_ir_simplify
+Scanning dependencies of target test_transform_polyfor_to_for
+[ 77%] Building CXX object cinn/optim/CMakeFiles/test_transform_polyfor_to_for.dir/transform_polyfor_to_for_test.cc.o
+[ 77%] Built target test_optimize
+Scanning dependencies of target test_hlir_framework_print_graph_pass
+[ 77%] Building CXX object cinn/hlir/framework/CMakeFiles/test_hlir_framework_print_graph_pass.dir/print_graph_pass_test.cc.o
+[ 77%] Built target test_ir_copy
+[ 78%] Linking CXX executable test_vectorize_loops
+Scanning dependencies of target test_hlir_framework_infershape_pass
+[ 78%] Building CXX object cinn/hlir/framework/CMakeFiles/test_hlir_framework_infershape_pass.dir/infershape_pass_test.cc.o
+[ 78%] Built target test_remove_nested_block
+[ 78%] Built target test_replace_call_with_expr
+Scanning dependencies of target test_hlir_framework_buffer
+[ 79%] Building CXX object cinn/hlir/framework/CMakeFiles/test_hlir_framework_buffer.dir/buffer_test.cc.o
+Scanning dependencies of target test_hlir_framework_scope
+[ 79%] Building CXX object cinn/hlir/framework/CMakeFiles/test_hlir_framework_scope.dir/scope_test.cc.o
+[ 79%] Built target test_execution_engine
+Scanning dependencies of target test_hlir_framework_tensor
+[ 79%] Building CXX object cinn/hlir/framework/CMakeFiles/test_hlir_framework_tensor.dir/tensor_test.cc.o
+[ 79%] Linking CXX executable test_hlir_framework_buffer
+[ 79%] Linking CXX executable test_transform_polyfor_to_for
+[ 79%] Built target test_hlir_framework_buffer
+Scanning dependencies of target test_hlir_framework_instruction
+[ 80%] Building CXX object cinn/hlir/framework/CMakeFiles/test_hlir_framework_instruction.dir/instruction_test.cc.o
+[ 80%] Built target test_ir_simplify
+[ 80%] Linking CXX executable test_hlir_framework_tensor
+[ 81%] Linking CXX executable test_hlir_framework_scope
+Scanning dependencies of target test_hlir_framework_op
+[ 81%] Building CXX object cinn/hlir/framework/CMakeFiles/test_hlir_framework_op.dir/op_test.cc.o
+[ 81%] Linking CXX executable test_hlir_framework_print_graph_pass
+[ 81%] Built target test_vectorize_loops
+Scanning dependencies of target test_pe_transform
+[ 81%] Building CXX object cinn/hlir/pe/CMakeFiles/test_pe_transform.dir/pe_transform_test.cc.o
+[ 81%] Built target test_hlir_framework_tensor
+[ 81%] Built target test_hlir_framework_scope
+[ 81%] Built target test_transform_polyfor_to_for
+Scanning dependencies of target test_pe_elementwise
+[ 81%] Building CXX object cinn/hlir/pe/CMakeFiles/test_pe_elementwise.dir/pe_elementwise_test.cc.o
+Scanning dependencies of target test_op_broadcast
+Scanning dependencies of target test_pe_broadcast
+[ 81%] Building CXX object cinn/hlir/op/CMakeFiles/test_op_broadcast.dir/op_broadcast_test.cc.o
+[ 81%] Building CXX object cinn/hlir/pe/CMakeFiles/test_pe_broadcast.dir/pe_broadcast_test.cc.o
+[ 81%] Built target test_hlir_framework_print_graph_pass
+Scanning dependencies of target test_op_nn
+[ 82%] Building CXX object cinn/hlir/op/CMakeFiles/test_op_nn.dir/op_nn_test.cc.o
+[ 83%] Linking CXX executable test_hlir_framework_op
+[ 83%] Linking CXX executable test_op_broadcast
+[ 83%] Built target test_hlir_framework_op
+Scanning dependencies of target core_api
+[ 84%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/runtime.cc.o
+[ 84%] Linking CXX executable test_hlir_framework_infershape_pass
+[ 84%] Built target test_op_broadcast
+[ 84%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/common.cc.o
+[ 85%] Linking CXX executable test_pe_transform
+[ 85%] Linking CXX executable test_hlir_framework_instruction
+[ 85%] Built target test_hlir_framework_infershape_pass
+[ 85%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/lang.cc.o
+[ 86%] Linking CXX executable test_pe_broadcast
+[ 86%] Linking CXX executable test_pe_elementwise
+[ 86%] Built target test_pe_transform
+[ 87%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/ir.cc.o
+[ 87%] Built target test_hlir_framework_instruction
+[ 87%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/poly.cc.o
+[ 87%] Linking CXX executable test_op_nn
+Scanning dependencies of target test_frontend_syntax
+[ 87%] Building CXX object cinn/frontend/CMakeFiles/test_frontend_syntax.dir/syntax_test.cc.o
+[ 87%] Built target test_pe_elementwise
+[ 87%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/backends.cc.o
+[ 87%] Built target test_pe_broadcast
+[ 87%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/bind.cc.o
+[ 87%] Built target test_op_nn
+[ 88%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/optim.cc.o
+Scanning dependencies of target test_frontend_executor
+[ 88%] Building CXX object cinn/frontend/CMakeFiles/test_frontend_executor.dir/executor_test.cc.o
+Scanning dependencies of target test_model_parser
+[ 88%] Building CXX object cinn/frontend/paddle/CMakeFiles/test_model_parser.dir/model_parser_test.cc.o
+[ 89%] Built target test01_elementwise_add_main
+[ 89%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/pe.cc.o
+[ 89%] Built target test03_conv_main
+[ 89%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/frontend.cc.o
+[ 90%] Built target test02_matmul_main
+[ 91%] Building CXX object cinn/pybind/CMakeFiles/core_api.dir/framework.cc.o
+[ 92%] Linking CXX executable test_model_parser
+[ 92%] Built target test_model_parser
+Scanning dependencies of target test_generated1
+[ 92%] Building CXX object cinn/backends/CMakeFiles/test_generated1.dir/generated_module1.cc.o
+[ 93%] Linking CXX executable test_generated1
+[ 93%] Built target test_generated1
+Scanning dependencies of target test01_elementwise_add_case
+[ 94%] Building CXX object tests/CMakeFiles/test01_elementwise_add_case.dir/test01_elementwise_add_case.cc.o
+[ 94%] Linking CXX executable test_frontend_syntax
+[ 94%] Building CXX object tests/CMakeFiles/test01_elementwise_add_case.dir/test01_elementwise_add.cc.o
+Scanning dependencies of target test03_conv_case
+[ 94%] Building CXX object tests/CMakeFiles/test03_conv_case.dir/test03_convolution_case.cc.o
+[ 94%] Building CXX object tests/CMakeFiles/test03_conv_case.dir/test03_convolution.cc.o
+[ 94%] Building CXX object tests/CMakeFiles/test01_elementwise_add_case.dir/test01_elementwise_add_compute_at.cc.o
+[ 95%] Linking CXX executable test03_conv_case
+Scanning dependencies of target test02_matmul_case
+[ 95%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_case.cc.o
+[ 95%] Built target test03_conv_case
+[ 95%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul.cc.o
+[ 95%] Building CXX object tests/CMakeFiles/test01_elementwise_add_case.dir/test01_elementwise_add_compute_at_level1.cc.o
+[ 95%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_tile.cc.o
+[ 95%] Built target test_frontend_syntax
+[ 96%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_split.cc.o
+[ 97%] Building CXX object tests/CMakeFiles/test01_elementwise_add_case.dir/test01_elementwise_add_vectorize.cc.o
+[ 97%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_block.cc.o
+[ 97%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_vectorize.cc.o
+[ 98%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_loop_permutation.cc.o
+[ 98%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_array_packing.cc.o
+[ 99%] Linking CXX executable test_frontend_executor
+[ 99%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_varient_shape.cc.o
+[ 99%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_varient_shape_tile.cc.o
+[100%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_array_packing_dynamic_shape.cc.o
+[100%] Building CXX object tests/CMakeFiles/test02_matmul_case.dir/test02_matmul_call.cc.o
+[100%] Linking CXX executable test01_elementwise_add_case
+[100%] Built target test_frontend_executor
+[100%] Built target test01_elementwise_add_case
+[100%] Linking CXX executable test02_matmul_case
+[100%] Built target test02_matmul_case
+[100%] Linking CXX shared library core_api.so
+[100%] Built target core_api
+Scanning dependencies of target COPY_CINN_CORE_API
+[100%] Generating cinn/core_api.so
+[100%] Built target COPY_CINN_CORE_API
+ResNet18/
+ResNet18/params
+ResNet18/__model__
+MobileNetV2/
+MobileNetV2/params
+MobileNetV2/__model__
+res fc_0.tmp_2
+res fc_5.tmp_2
+res relu_0.tmp_0
+UpdateCTestConfiguration  from :/home/wangyue50/CINN-my/CINN/build/DartConfiguration.tcl
+UpdateCTestConfiguration  from :/home/wangyue50/CINN-my/CINN/build/DartConfiguration.tcl
+Test project /home/wangyue50/CINN-my/CINN/build
+Constructing a list of tests
+Done constructing a list of tests
+Updating test list for fixtures
+Added 0 tests to meet fixture requirements
+Checking test dependency graph...
+Checking test dependency graph end
+test 53
+      Start 53: test02_matmul_main
+
+53: Test command: /home/wangyue50/CINN-my/CINN/build/tests/test02_matmul_main "--cinn_x86_builtin_code_root=/home/wangyue50/CINN-my/CINN/cinn/backends"
+53: Test timeout computed to be: 600
+test 55
+      Start 55: test03_conv_main
+
+55: Test command: /home/wangyue50/CINN-my/CINN/build/tests/test03_conv_main "--cinn_x86_builtin_code_root=/home/wangyue50/CINN-my/CINN/cinn/backends"
+55: Test timeout computed to be: 600
+test 51
+      Start 51: test01_elementwise_add_main
+
+51: Test command: /home/wangyue50/CINN-my/CINN/build/tests/test01_elementwise_add_main "--cinn_x86_builtin_code_root=/home/wangyue50/CINN-my/CINN/cinn/backends"
+51: Test timeout computed to be: 600
+test 1
+      Start  1: test_cinn_value
+
+1: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/common/test_cinn_value ""
+1: Test timeout computed to be: 600
+test 2
+      Start  2: test_shared
+
+2: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/common/test_shared ""
+2: Test timeout computed to be: 600
+test 3
+      Start  3: test_graph_utils
+
+3: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/common/test_graph_utils ""
+3: Test timeout computed to be: 600
+test 4
+      Start  4: test_arithmatic
+
+4: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/common/test_arithmatic ""
+4: Test timeout computed to be: 600
+test 5
+      Start  5: test_cas
+
+5: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/common/test_cas ""
+5: Test timeout computed to be: 600
+test 6
+      Start  6: test_string
+
+6: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/utils/test_string ""
+6: Test timeout computed to be: 600
+test 7
+      Start  7: test_schedule
+
+7: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/poly/test_schedule ""
+7: Test timeout computed to be: 600
+55: [==========] Running 1 test from 1 test case.
+55: [----------] Global test environment set-up.
+55: [----------] 1 test from test03_conv
+55: [ RUN      ] test03_conv.basic
+55: WARNING: Logging before InitGoogleLogging() is written to STDERR
+55: W0924 13:32:18.210212 27787 codegen_c.cc:24] Output C header to file ./test03_convolution.h
+55: W0924 13:32:18.233403 27787 codegen_c.cc:33] Output C source to file ./test03_convolution.cc
+55: [       OK ] test03_conv.basic (74 ms)
+55: [----------] 1 test from test03_conv (74 ms total)
+55: 
+55: [----------] Global test environment tear-down
+55: [==========] 1 test from 1 test case ran. (74 ms total)
+55: [  PASSED  ] 1 test.
+53: [==========] Running 9 tests from 2 test cases.
+53: [----------] Global test environment set-up.
+53: [----------] 1 test from test02_matmul
+53: [ RUN      ] test02_matmul.basic
+53: WARNING: Logging before InitGoogleLogging() is written to STDERR
+53: W0924 13:32:18.175216 27786 codegen_c.cc:24] Output C header to file ./test02_matmul.h
+53: W0924 13:32:18.176230 27786 codegen_c.cc:33] Output C source to file ./test02_matmul.cc
+53: W0924 13:32:18.207154 27786 codegen_c.cc:24] Output C header to file ./test02_matmul_tile.h
+53: W0924 13:32:18.211670 27786 codegen_c.cc:33] Output C source to file ./test02_matmul_tile.cc
+53: [       OK ] test02_matmul.basic (52 ms)
+53: [----------] 1 test from test02_matmul (52 ms total)
+53: 
+53: [----------] 8 tests from matmul
+53: [ RUN      ] matmul.Split
+53: W0924 13:32:18.232223 27786 codegen_c.cc:24] Output C header to file ./test02_matmul_split.h
+53: W0924 13:32:18.235391 27786 codegen_c.cc:33] Output C source to file ./test02_matmul_split.cc
+53: [       OK ] matmul.Split (24 ms)
+53: [ RUN      ] matmul.Blocking
+53: W0924 13:32:18.278457 27786 codegen_c.cc:24] Output C header to file ./test02_matmul_block.h
+53: W0924 13:32:18.286087 27786 codegen_c.cc:33] Output C source to file ./test02_matmul_block.cc
+53: [       OK ] matmul.Blocking (51 ms)
+53: [ RUN      ] matmul.Vectorization
+53: W0924 13:32:18.332374 27786 codegen_c.cc:24] Output C header to file ./test02_matmul_vectorize.h
+51: [==========] Running 4 tests from 2 test cases.
+51: [----------] Global test environment set-up.
+51: [----------] 2 tests from test01_elementwise_add
+51: [ RUN      ] test01_elementwise_add.basic
+51: WARNING: Logging before InitGoogleLogging() is written to STDERR
+51: W0924 13:32:18.168052 27788 codegen_c.cc:24] Output C header to file ./test01_elementwise_add.h
+51: W0924 13:32:18.168761 27788 codegen_c.cc:33] Output C source to file ./test01_elementwise_add.cc
+51: [       OK ] test01_elementwise_add.basic (9 ms)
+51: [ RUN      ] test01_elementwise_add.vectorize
+51: I0924 13:32:18.175801 27788 test01_elementwise_add_main.cc:56] after optim:
+51: function add1_vectorize (_A, _B, _C)
+51: {
+51:   for (i, 100)
+51:   {
+51:     for (j, 4)
+51:     {
+51:       C[Broadcast(i,8), Ramp((8 * j),1,8)] = (A[i, Ramp((8 * j),1,8)] + B[i, Ramp((8 * j),1,8)])
+51:     }
+51:   }
+51: }
+51: W0924 13:32:18.176316 27788 codegen_c.cc:24] Output C header to file ./test01_elementwise_add_vectorize.h
+51: W0924 13:32:18.177737 27788 codegen_c.cc:33] Output C source to file ./test01_elementwise_add_vectorize.cc
+51: [       OK ] test01_elementwise_add.vectorize (9 ms)
+51: [----------] 2 tests from test01_elementwise_add (18 ms total)
+51: 
+51: [----------] 2 tests from elementwise_add
+51: [ RUN      ] elementwise_add.compute_at
+51: W0924 13:32:18.197803 27788 codegen_c.cc:24] Output C header to file ./test01_elementwise_add_compute_at.h
+51: W0924 13:32:18.200814 27788 codegen_c.cc:33] Output C source to file ./test01_elementwise_add_compute_at.cc
+51: [       OK ] elementwise_add.compute_at (23 ms)
+51: [ RUN      ] elementwise_add.compute_at1
+51: W0924 13:32:18.227854 27788 codegen_c.cc:24] Output C header to file ./test01_elementwise_add_compute_at_level1.h
+51: W0924 13:32:18.230847 27788 codegen_c.cc:33] Output C source to file ./test01_elementwise_add_compute_at_level1.cc
+51: [       OK ] elementwise_add.compute_at1 (30 ms)
+51: [----------] 2 tests from elementwise_add (53 ms total)
+51: 
+51: [----------] Global test environment tear-down
+51: [==========] 4 tests from 2 test cases ran. (71 ms total)
+51: [  PASSED  ] 4 tests.
+1: [==========] Running 3 tests from 1 test case.
+1: [----------] Global test environment set-up.
+1: [----------] 3 tests from CINNValue
+1: [ RUN      ] CINNValue.test
+1: [       OK ] CINNValue.test (0 ms)
+1: [ RUN      ] CINNValue.buffer
+1: [       OK ] CINNValue.buffer (0 ms)
+1: [ RUN      ] CINNValue.Expr
+1: [       OK ] CINNValue.Expr (0 ms)
+1: [----------] 3 tests from CINNValue (0 ms total)
+1: 
+1: [----------] Global test environment tear-down
+1: [==========] 3 tests from 1 test case ran. (0 ms total)
+1: [  PASSED  ] 3 tests.
+2: [==========] Running 2 tests from 1 test case.
+2: [----------] Global test environment set-up.
+2: [----------] 2 tests from Shared
+2: [ RUN      ] Shared.test
+2: [       OK ] Shared.test (0 ms)
+2: [ RUN      ] Shared.cycle_share
+2: [       OK ] Shared.cycle_share (0 ms)
+2: [----------] 2 tests from Shared (0 ms total)
+2: 
+2: [----------] Global test environment tear-down
+2: [==========] 2 tests from 1 test case ran. (0 ms total)
+2: [  PASSED  ] 2 tests.
+3: [==========] Running 2 tests from 1 test case.
+3: [----------] Global test environment set-up.
+3: [----------] 2 tests from Graph
+3: [ RUN      ] Graph.Visualize
+3: WARNING: Logging before InitGoogleLogging() is written to STDERR
+3: I0924 13:32:18.161446 27791 graph_utils_test.cc:58] graph:
+3: digraph G {
+3:    node_0[label="A"]
+3:    node_1[label="B"]
+3:    node_2[label="C"]
+3:    node_3[label="D"]
+3:    node_4[label="E"]
+3:    node_0->node_2
+3:    node_0->node_1
+3:    node_1->node_3
+3:    node_2->node_4
+3:    node_2->node_3
+3: } // end G
+3: [       OK ] Graph.Visualize (0 ms)
+3: [ RUN      ] Graph.simple
+3: I0924 13:32:18.161548 27791 graph_utils_test.cc:67] graph1 digraph G {
+3:    node_5[label="A"]
+3:    node_6[label="B"]
+3:    node_6->node_5
+3: } // end G
+3: [       OK ] Graph.simple (0 ms)
+3: [----------] 2 tests from Graph (0 ms total)
+3: 
+3: [----------] Global test environment tear-down
+3: [==========] 2 tests from 1 test case ran. (0 ms total)
+3: [  PASSED  ] 2 tests.
+4: [==========] Running 5 tests from 2 test cases.
+4: [----------] Global test environment set-up.
+4: [----------] 3 tests from GiNaC
+4: [ RUN      ] GiNaC.simplify
+4: WARNING: Logging before InitGoogleLogging() is written to STDERR
+4: I0924 13:32:18.162345 27792 arithmatic_test.cc:26] e: -94+9*y
+4: [       OK ] GiNaC.simplify (0 ms)
+4: [ RUN      ] GiNaC.diff
+4: I0924 13:32:18.162425 27792 arithmatic_test.cc:37] e: 1
+4: I0924 13:32:18.162429 27792 arithmatic_test.cc:38] e1: 0
+4: [       OK ] GiNaC.diff (0 ms)
+4: [ RUN      ] GiNaC.solve
+4: I0924 13:32:18.162444 27792 arithmatic_test.cc:48] solve: {x==8}
+4: I0924 13:32:18.162505 27792 arithmatic_test.cc:49] 2
+4: [       OK ] GiNaC.solve (0 ms)
+4: [----------] 3 tests from GiNaC (0 ms total)
+4: 
+4: [----------] 2 tests from Solve
+4: [ RUN      ] Solve.basic
+4: I0924 13:32:18.162674 27792 arithmatic_test.cc:59] res: 200
+4: [       OK ] Solve.basic (0 ms)
+4: [ RUN      ] Solve.basic1
+4: I0924 13:32:18.162784 27792 arithmatic_test.cc:72] res -400
+4: [       OK ] Solve.basic1 (0 ms)
+4: [----------] 2 tests from Solve (0 ms total)
+4: 
+4: [----------] Global test environment tear-down
+4: [==========] 5 tests from 2 test cases ran. (0 ms total)
+4: [  PASSED  ] 5 tests.
+5: [==========] Running 12 tests from 2 test cases.
+5: [----------] Global test environment set-up.
+5: [----------] 11 tests from CAS
+5: [ RUN      ] CAS.SimplifyPower_0
+5: WARNING: Logging before InitGoogleLogging() is written to STDERR
+5: I0924 13:32:18.163086 27793 cas_test.cc:25] p0 (x^0)
+5: I0924 13:32:18.163147 27793 cas_test.cc:27] simplified 1
+5: I0924 13:32:18.163197 27793 cas_test.cc:47] p0 (1^x)
+5: I0924 13:32:18.163206 27793 cas_test.cc:49] simplified 1
+5: I0924 13:32:18.163215 27793 cas_test.cc:56] p0 (0^x)
+5: I0924 13:32:18.163221 27793 cas_test.cc:58] simplified 0
+5: I0924 13:32:18.163229 27793 cas_test.cc:65] p0 (0^x)
+5: I0924 13:32:18.163236 27793 cas_test.cc:67] simplified 0
+5: [       OK ] CAS.SimplifyPower_0 (0 ms)
+5: [ RUN      ] CAS.number_cal
+5: I0924 13:32:18.163265 27793 cas_test.cc:75] ((1 * 100 * -1) + 0 + 1001)
+5: [       OK ] CAS.number_cal (0 ms)
+5: [ RUN      ] CAS.SimplifyPower
+5: I0924 13:32:18.163280 27793 cas_test.cc:81] p0 (x^2)
+5: I0924 13:32:18.163285 27793 cas_test.cc:84] power: ((x^2)^3)
+5: I0924 13:32:18.163316 27793 cas_test.cc:87] simplified: (x^(6))
+5: [       OK ] CAS.SimplifyPower (0 ms)
+5: [ RUN      ] CAS.cmp
+5: [       OK ] CAS.cmp (0 ms)
+5: [ RUN      ] CAS.SimplifySum
+5: [       OK ] CAS.SimplifySum (2 ms)
+5: [ RUN      ] CAS.SimplifyProduct
+5: I0924 13:32:18.166296 27793 cas_test.cc:164] (x^5)
+5: [       OK ] CAS.SimplifyProduct (1 ms)
+5: [ RUN      ] CAS.SimplifyMod
+5: [       OK ] CAS.SimplifyMod (0 ms)
+5: [ RUN      ] CAS.ConvertCinnToCAS
+5: I0924 13:32:18.167304 27793 cas_test.cc:196] body ((((A[i, j] + 0) + 1) + (2 * B[i, j])) + ((0 * B[i, j]) * A[i, j]))
+5: [       OK ] CAS.ConvertCinnToCAS (1 ms)
+5: [ RUN      ] CAS.FracOp
+5: [       OK ] CAS.FracOp (1 ms)
+5: [ RUN      ] CAS.Mod
+5: I0924 13:32:18.172488 27793 cas_test.cc:273] ((128 * k) + (x + ((32768 * y) + (32 * z))))
+5: [       OK ] CAS.Mod (4 ms)
+5: [ RUN      ] CAS.IntConnerCase
+5: I0924 13:32:18.173523 27793 cas.cc:1231] found y Interval[2, 3]  ai 1
+5: I0924 13:32:18.173576 27793 cas.cc:1231] found y Interval[0, 3]  ai 1
+5: I0924 13:32:18.173614 27793 cas.cc:1231] found y Interval[0, 3]  ai 1
+5: [       OK ] CAS.IntConnerCase (1 ms)
+5: [----------] 11 tests from CAS (10 ms total)
+5: 
+5: [----------] 1 test from SolveInequality
+5: [ RUN      ] SolveInequality.basic
+5: [       OK ] SolveInequality.basic (3 ms)
+5: [----------] 1 test from SolveInequality (3 ms total)
+5: 
+5: [----------] Global test environment tear-down
+5: [==========] 12 tests from 2 test cases ran. (13 ms total)
+5: [  PASSED  ] 12 tests.
+6: [==========] Running 2 tests from 1 test case.
+6: [----------] Global test environment set-up.
+6: [----------] 2 tests from string
+6: [ RUN      ] string.Endswith
+6: [       OK ] string.Endswith (0 ms)
+6: [ RUN      ] string.Startswith
+6: [       OK ] string.Startswith (0 ms)
+6: [----------] 2 tests from string (0 ms total)
+6: 
+6: [----------] Global test environment tear-down
+6: [==========] 2 tests from 1 test case ran. (0 ms total)
+6: [  PASSED  ] 2 tests.
+7: [==========] Running 2 tests from 1 test case.
+7: [----------] Global test environment set-up.
+7: [----------] 2 tests from CreateStages
+7: [ RUN      ] CreateStages.compute_at
+7: {
+7:   for (i, 100)
+7:   {
+7:     for (j, 100)
+7:     {
+7:       B[i, j] = (1 + A[i, j])
+7:       for (k, 100)
+7:       {
+7:         C[i, j, k] = (B[i, j] * B[j, k])
+7:       }
+7:     }
+7:   }
+7: }
+7: [       OK ] CreateStages.compute_at (18 ms)
+7: [ RUN      ] CreateStages.buffer_bind_to_multiple_tensors_schedule
+7: {
+7:   for (i, 100)
+7:   {
+7:     for (j, 100)
+7:     {
+7:       B[i, j] = (1 + A[i, j])
+7:     }
+7:   }
+7:   for (i, 100)
+7:   {
+7:     for (j, 100)
+7:     {
+7:       C[i, j] = (1 + A[i, j])
+7:     }
+7:   }
+7:   for (i, 100)
+7:   {
+7:     for (j, 100)
+7:     {
+7:       D[i, j] = (1 + A[i, j])
+7:     }
+7:   }
+7: }
+7: [       OK ] CreateStages.buffer_bind_to_multiple_tensors_schedule (15 ms)
+7: [----------] 2 tests from CreateStages (33 ms total)
+7: 
+7: [----------] Global test environment tear-down
+7: [==========] 2 tests from 1 test case ran. (33 ms total)
+7: [  PASSED  ] 2 tests.
+ 1/68 Test #55: test03_conv_main .......................   Passed    0.09 sec
+ 2/68 Test #51: test01_elementwise_add_main ............   Passed    0.19 sec
+ 3/68 Test  #1: test_cinn_value ........................   Passed    0.19 sec
+ 4/68 Test  #2: test_shared ............................   Passed    0.19 sec
+ 5/68 Test  #3: test_graph_utils .......................   Passed    0.19 sec
+ 6/68 Test  #4: test_arithmatic ........................   Passed    0.19 sec
+ 7/68 Test  #5: test_cas ...............................   Passed    0.19 sec
+ 8/68 Test  #6: test_string ............................   Passed    0.19 sec
+ 9/68 Test  #7: test_schedule ..........................   Passed    0.18 sec
+test 8
+      Start  8: test_stage
+
+8: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/poly/test_stage ""
+8: Test timeout computed to be: 600
+test 9
+      Start  9: test_compute_at_transform
+
+9: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/poly/test_compute_at_transform ""
+9: Test timeout computed to be: 600
+test 10
+      Start 10: test_cinn_runtime
+
+10: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/runtime/test_cinn_runtime ""
+10: Test timeout computed to be: 600
+test 11
+      Start 11: test_mkl_math
+
+11: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/runtime/cpu/test_mkl_math ""
+11: Test timeout computed to be: 600
+test 12
+      Start 12: test_host_intrinsics
+
+12: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/runtime/cpu/test_host_intrinsics ""
+12: Test timeout computed to be: 600
+test 13
+      Start 13: test_collect_ir_nodes
+
+13: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/ir/test_collect_ir_nodes ""
+13: Test timeout computed to be: 600
+test 14
+      Start 14: test_buffer
+
+14: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/ir/test_buffer "--cinn_x86_builtin_code_root=/home/wangyue50/CINN-my/CINN/cinn/backends"
+14: Test timeout computed to be: 600
+test 15
+      Start 15: test_tensor
+
+15: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/ir/test_tensor ""
+15: Test timeout computed to be: 600
+test 16
+      Start 16: test_codegen_c
+
+16: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/backends/test_codegen_c "--cinn_x86_builtin_code_root=/home/wangyue50/CINN-my/CINN/cinn/backends"
+16: Test timeout computed to be: 600
+14: [==========] Running 2 tests from 1 test case.
+14: [----------] Global test environment set-up.
+14: [----------] 2 tests from Buffer
+14: [ RUN      ] Buffer.basic
+14: [       OK ] Buffer.basic (0 ms)
+14: [ RUN      ] Buffer.bind_to_multiple_tensors
+14: codegen C:
+14: #include <cinn_runtime.h>
+14: #include <stdio.h>
+14: 
+14: ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+14: ///                     Predefined utilities in CINN BEGIN(
+14: ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+14: 
+14: #include <immintrin.h>
+14: #include <stdint.h>
+14: 
+14: #include <vector>
+14: 
+14: //! Vector in stack, this can only used in generated .cc file.
+14: template <typename T, size_t Num>
+14: struct StackVec {
+14:   typedef T value_type;
+14:   typedef StackVec<T, Num> self_type;
+14: 
+14:   self_type& operator=(const StackVec& src) {
+14:     if (this != &src) {
+14:       memcpy(data_, src.data_, num_bytes());
+14:     }
+14:     return *this;
+14:   }
+14: 
+14:   StackVec() { memset(data_, 0, num_bytes()); }
+14: 
+14:   explicit StackVec(const T* externl) : external_data_(externl) {}
+14: 
+14:   static self_type Broadcast(const value_type& v) {
+14:     self_type res;
+14:     for (size_t i = 0; i < Num; i++) res.data_[i] = v;
+14:     return res;
+14:   }
+14: 
+14:   static self_type Ramp(const value_type& base, const value_type& stride) {
+14:     self_type res;
+14:     for (size_t i = 0; i < Num; i++) {
+14:       res.data_[i] = base + stride * i;
+14:     }
+14:   }
+14: 
+14:   static self_type Load(const void* base, int32_t offset) {
+14:     self_type res;
+14:     memcpy(&res.data_[0], (const value_type*)base + offset, num_bytes());
+14:   }
+14: 
+14:   static self_type Load(const void* base, const StackVec<int32_t, Num>& offset) {
+14:     self_type res;
+14:     for (size_t i = 0; i < Num; i++) {
+14:       res.data_[i] = ((const value_type*)base)[offset[i]];
+14:     }
+14:   }
+14: 
+14:   void Store(void* base, int32_t offset) const {
+14:     mempcpy((value_type*)base + offset, &data_[0], num_bytes());  // NOLINT
+14:   }
+14: 
+14:   inline value_type& operator[](size_t i) { return data_[i]; }
+14:   inline value_type operator[](size_t i) const { return data_[i]; }
+14: 
+14:   // binary operator between two vectors
+14:   // @{
+14: #define __(op__)                                                           \
+14:   friend self_type operator op__(const self_type& a, const self_type& b) { \
+14:     self_type res;                                                         \
+14:     for (size_t i = 0; i < Num; i++) {                                     \
+14:       res.data_[i] = a[i] op__ b[i];                                       \
+14:     }                                                                      \
+14:     return res;                                                            \
+14:   }
+14:   __(+)
+14:   __(-)
+14:   __(*)
+14:   __(/)
+14:   __(%)
+14:   // @}
+14: #undef __
+14: 
+14:   // binary operator between a vector and a scalar
+14:   // @{
+14: #define __(op__)                                                            \
+14:   friend self_type operator op__(const self_type& a, const value_type& b) { \
+14:     self_type res;                                                          \
+14:     for (size_t i = 0; i < Num; i++) {                                      \
+14:       res.data_[i] = a[i] op__ b;                                           \
+14:     }                                                                       \
+14:     return res;                                                             \
+14:   }
+14:   __(+)
+14:   __(-)
+14:   __(*)
+14:   __(/)
+14:   __(%)
+14: #undef __
+14:   // @}
+14: 
+14:   static constexpr size_t num_bytes() { return sizeof(data_); }
+14: 
+14:  private:
+14:   T data_[Num];
+14:   T* external_data_{nullptr};
+14: };
+14: 
+14: /**
+14:  * The vector with external data.
+14:  */
+14: template <typename T, size_t Num>
+14: struct ExternalVec {
+14:   typedef T value_type;
+14:   typedef ExternalVec<T, Num> self_type;
+14: 
+14:   explicit ExternalVec(T* data) : data_(data) {}
+14: 
+14:   self_type& operator=(const self_type& src) {
+14:     if (data_ != src.data_) {
+14:       memcpy(data_, src.data_, num_bytes());
+14:     }
+14:     return *this;
+14:   }
+14: 
+14:   static self_type Load(const void* base, int32_t offset) {
+14:     self_type res((T*)base + offset);  // NOLINT
+14:     return res;
+14:   }
+14: 
+14:   static constexpr size_t num_bytes() { return sizeof(value_type) * Num; }
+14: 
+14:  private:
+14:   T* data_{nullptr};
+14: };
+14: 
+14: // AVX256 load
+14: //@{
+14: inline __m256 cinn_avx256_load(const float* dst) { return _mm256_load_ps(dst); }
+14: inline __m256d cinn_avx256_load(const double* dst) { return _mm256_load_pd(dst); }
+14: //@}
+14: // AVX512 load
+14: //@{
+14: inline __m512 cinn_avx512_load(const float* dst) { return _mm512_load_ps(dst); }
+14: inline __m512d cinn_avx512_load(const double* dst) { return _mm512_load_pd(dst); }
+14: //@}
+14: 
+14: // FP32x8 * FP32x8
+14: // @{
+14: inline void cinn_avx256_add(float* dst, float* a, float* b) {
+14:   _mm256_store_ps(dst, _mm256_add_ps(_mm256_load_ps(a), _mm256_load_ps(b)));
+14: }
+14: inline void cinn_avx256_sub(float* dst, float* a, float* b) {
+14:   _mm256_store_ps(dst, _mm256_sub_ps(_mm256_load_ps(a), _mm256_load_ps(b)));
+14: }
+14: inline void cinn_avx256_mul(float* dst, float* a, float* b) {
+14:   _mm256_store_ps(dst, _mm256_mul_ps(_mm256_load_ps(a), _mm256_load_ps(b)));
+14: }
+14: inline void cinn_avx256_div(float* dst, float* a, float* b) {
+14:   _mm256_store_ps(dst, _mm256_div_ps(_mm256_load_ps(a), _mm256_load_ps(b)));
+14: }
+14: // @}
+14: 
+14: // FP32x4 * float
+14: // @{
+14: inline void cinn_avx256_add(float* dst, float* a, float b) {
+14:   _mm256_store_ps(dst, _mm256_add_ps(_mm256_load_ps(a), _mm256_set1_ps(b)));
+14: }
+14: inline void cinn_avx256_sub(float* dst, float* a, float b) {
+14:   _mm256_store_ps(dst, _mm256_sub_ps(_mm256_load_ps(a), _mm256_set1_ps(b)));
+14: }
+14: inline void cinn_avx256_mul(float* dst, float* a, float b) {
+14:   _mm256_store_ps(dst, _mm256_mul_ps(_mm256_load_ps(a), _mm256_set1_ps(b)));
+14: }
+14: inline void cinn_avx256_div(float* dst, float* a, float b) {
+14:   _mm256_store_ps(dst, _mm256_div_ps(_mm256_load_ps(a), _mm256_set1_ps(b)));
+14: }
+14: // @}
+14: 
+14: // float * FP32x4
+14: // @{
+14: inline void cinn_avx256_add(float* dst, float a, float* b) {
+14:   _mm256_store_ps(dst, _mm256_add_ps(_mm256_set1_ps(a), _mm256_load_ps(b)));
+14: }
+14: inline void cinn_avx256_sub(float* dst, float a, float* b) {
+14:   _mm256_store_ps(dst, _mm256_sub_ps(_mm256_set1_ps(a), _mm256_load_ps(b)));
+14: }
+14: inline void cinn_avx256_mul(float* dst, float a, float* b) {
+14:   _mm256_store_ps(dst, _mm256_mul_ps(_mm256_set1_ps(a), _mm256_load_ps(b)));
+14: }
+14: inline void cinn_avx256_div(float* dst, float a, float* b) {
+14:   _mm256_store_ps(dst, _mm256_div_ps(_mm256_set1_ps(a), _mm256_load_ps(b)));
+14: }
+14: // @}
+14: 
+14: // 4 x float64
+14: // @{
+14: inline void cinn_avx256_add(double* dst, double* a, double* b) {
+14:   _mm256_store_pd(dst, _mm256_add_pd(_mm256_load_pd(a), _mm256_load_pd(b)));
+14: }
+14: inline void cinn_avx256_sub(double* dst, double* a, double* b) {
+14:   _mm256_store_pd(dst, _mm256_sub_pd(_mm256_load_pd(a), _mm256_load_pd(b)));
+14: }
+14: inline void cinn_avx256_mul(double* dst, double* a, double* b) {
+14:   _mm256_store_pd(dst, _mm256_mul_pd(_mm256_load_pd(a), _mm256_load_pd(b)));
+14: }
+14: inline void cinn_avx256_div(double* dst, double* a, double* b) {
+14:   _mm256_store_pd(dst, _mm256_div_pd(_mm256_load_pd(a), _mm256_load_pd(b)));
+14: }
+14: // @}
+14: 
+14: // FP32x4 * FP64
+14: // @{
+14: inline void cinn_avx256_add(double* dst, double* a, double b) {
+14:   _mm256_store_pd(dst, _mm256_add_pd(_mm256_load_pd(a), _mm256_set1_pd(b)));
+14: }
+14: inline void cinn_avx256_sub(double* dst, double* a, double b) {
+14:   _mm256_store_pd(dst, _mm256_sub_pd(_mm256_load_pd(a), _mm256_set1_pd(b)));
+14: }
+14: inline void cinn_avx256_mul(double* dst, double* a, double b) {
+14:   _mm256_store_pd(dst, _mm256_mul_pd(_mm256_load_pd(a), _mm256_set1_pd(b)));
+14: }
+14: inline void cinn_avx256_div(double* dst, double* a, double b) {
+14:   _mm256_store_pd(dst, _mm256_div_pd(_mm256_load_pd(a), _mm256_set1_pd(b)));
+14: }
+14: // @}
+14: 
+14: // float * FP32x4
+14: // @{
+14: inline void cinn_avx256_add(double* dst, double a, double* b) {
+14:   _mm256_store_pd(dst, _mm256_add_pd(_mm256_set1_pd(a), _mm256_load_pd(b)));
+14: }
+14: inline void cinn_avx256_sub(double* dst, double a, double* b) {
+14:   _mm256_store_pd(dst, _mm256_sub_pd(_mm256_set1_pd(a), _mm256_load_pd(b)));
+14: }
+14: inline void cinn_avx256_mul(double* dst, double a, double* b) {
+14:   _mm256_store_pd(dst, _mm256_mul_pd(_mm256_set1_pd(a), _mm256_load_pd(b)));
+14: }
+14: inline void cinn_avx256_div(double* dst, double a, double* b) {
+14:   _mm256_store_pd(dst, _mm256_div_pd(_mm256_set1_pd(a), _mm256_load_pd(b)));
+14: }
+14: // @}
+14: 
+14: //! 32 x float32 operations.
+14: // @{
+14: inline void cinn_avx512_add(float* dst, float* a, float* b) {
+14:   _mm512_store_ps(dst, _mm512_add_ps(_mm512_load_ps(a), _mm512_load_ps(b)));
+14: }
+14: inline void cinn_avx512_sub(float* dst, float* a, float* b) {
+14:   _mm512_store_ps(dst, _mm512_sub_ps(_mm512_load_ps(a), _mm512_load_ps(b)));
+14: }
+14: inline void cinn_avx512_mul(float* dst, float* a, float* b) {
+14:   _mm512_store_ps(dst, _mm512_mul_ps(_mm512_load_ps(a), _mm512_load_ps(b)));
+14: }
+14: inline void cinn_avx512_div(float* dst, float* a, float* b) {
+14:   _mm512_store_ps(dst, _mm512_div_ps(_mm512_load_ps(a), _mm512_load_ps(b)));
+14: }
+14: // @}
+14: 
+14: // FP32x4 * FP64
+14: // @{
+14: inline void cinn_avx512_add(float* dst, float* a, float b) {
+14:   _mm512_store_pd(dst, _mm512_add_pd(_mm512_load_pd(a), _mm512_set1_pd(b)));
+14: }
+14: inline void cinn_avx512_sub(float* dst, float* a, float b) {
+14:   _mm512_store_pd(dst, _mm512_sub_pd(_mm512_load_pd(a), _mm512_set1_pd(b)));
+14: }
+14: inline void cinn_avx512_mul(float* dst, float* a, float b) {
+14:   _mm512_store_pd(dst, _mm512_mul_pd(_mm512_load_pd(a), _mm512_set1_pd(b)));
+14: }
+14: inline void cinn_avx512_div(float* dst, float* a, float b) {
+14:   _mm512_store_pd(dst, _mm512_div_pd(_mm512_load_pd(a), _mm512_set1_pd(b)));
+14: }
+14: // @}
+14: 
+14: // float * FP32x4
+14: // @{
+14: inline void cinn_avx512_add(float* dst, float a, float* b) {
+14:   _mm512_store_pd(dst, _mm512_add_pd(_mm512_set1_pd(a), _mm512_load_pd(b)));
+14: }
+14: inline void cinn_avx512_sub(float* dst, float a, float* b) {
+14:   _mm512_store_pd(dst, _mm512_sub_pd(_mm512_set1_pd(a), _mm512_load_pd(b)));
+14: }
+14: inline void cinn_avx512_mul(float* dst, float a, float* b) {
+14:   _mm512_store_pd(dst, _mm512_mul_pd(_mm512_set1_pd(a), _mm512_load_pd(b)));
+14: }
+14: inline void cinn_avx512_div(float* dst, float a, float* b) {
+14:   _mm512_store_pd(dst, _mm512_div_pd(_mm512_set1_pd(a), _mm512_load_pd(b)));
+14: }
+14: // @}
+14: 
+14: //! 16 x float32 operations.
+14: // @{
+14: inline void cinn_avx512_add(double* dst, double* a, double* b) {
+14:   _mm512_store_pd(dst, _mm512_add_pd(_mm512_load_pd(a), _mm512_load_pd(b)));
+14: }
+14: inline void cinn_avx512_sub(double* dst, double* a, double* b) {
+14:   _mm512_store_pd(dst, _mm512_sub_pd(_mm512_load_pd(a), _mm512_load_pd(b)));
+14: }
+14: inline void cinn_avx512_mul(double* dst, double* a, double* b) {
+14:   _mm512_store_pd(dst, _mm512_mul_pd(_mm512_load_pd(a), _mm512_load_pd(b)));
+14: }
+14: inline void cinn_avx512_div(double* dst, double* a, double* b) {
+14:   _mm512_store_pd(dst, _mm512_div_pd(_mm512_load_pd(a), _mm512_load_pd(b)));
+14: }
+14: // @}
+14: 
+14: inline __m512 cinn_avx512_add(const __m512& a, const __m512& b);
+14: 
+14: inline __m256 cinn_avx256_add_float(const __m256& a, const __m256& b) { return _mm256_add_ps(a, b); }
+14: inline __m256d cinn_avx256_add_double(const __m256d& a, const __m256d& b) { return _mm256_add_pd(a, b); }
+14: inline __m512 cinn_avx512_add_float(const __m512& a, const __m512& b) { return _mm512_add_ps(a, b); }
+14: inline __m512d cinn_avx512_add_double(const __m512d& a, const __m512d& b) { return _mm512_add_pd(a, b); }
+14: 
+14: //! set1
+14: // @{
+14: inline __m256 cinn_avx256_set1(float value) { return _mm256_set1_ps(value); }
+14: inline __m256d cinn_avx256_set1(double value) { return _mm256_set1_pd(value); }
+14: inline __m512 cinn_avx512_set1(float value) { return _mm512_set1_ps(value); }
+14: inline __m512d cinn_avx512_set1(double value) { return _mm512_set1_pd(value); }
+14: // @}
+14: 
+14: //! store
+14: // @{
+14: inline void cinn_avx512_store(float* dst, const __m512& x) { _mm512_store_ps(dst, x); }
+14: inline void cinn_avx512_store(double* dst, const __m512d& x) { _mm512_store_pd(dst, x); }
+14: inline void cinn_avx256_store(float* dst, const __m256& x) { _mm256_store_ps(dst, x); }
+14: inline void cinn_avx256_store(double* dst, const __m256d& x) { _mm256_store_pd(dst, x); }
+14: // @}
+14: 
+14: //! add
+14: // @{
+14: inline __m256 cinn_avx256_add(const __m256& a, const __m256& b) { return _mm256_add_ps(a, b); }
+14: inline __m256d cinn_avx256_add(const __m256d& a, const __m256d& b) { return _mm256_add_pd(a, b); }
+14: inline __m512 cinn_avx512_add(const __m512& a, const __m512& b) { return _mm512_add_ps(a, b); }
+14: inline __m512d cinn_avx512_add(const __m512d& a, const __m512d& b) { return _mm512_add_pd(a, b); }
+14: // @}
+14: 
+14: //! mul
+14: // @{
+14: inline __m256 cinn_avx256_mul(const __m256& a, const __m256& b) { return _mm256_mul_ps(a, b); }
+14: inline __m256d cinn_avx256_mul(const __m256d& a, const __m256d& b) { return _mm256_mul_pd(a, b); }
+14: inline __m512 cinn_avx512_mul(const __m512& a, const __m512& b) { return _mm512_mul_ps(a, b); }
+14: inline __m512d cinn_avx512_mul(const __m512d& a, const __m512d& b) { return _mm512_mul_pd(a, b); }
+14: // @}
+14: 
+14: ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+14: ///                     )END Predefined utilities in CINN
+14: ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+14: 
+14: cinn_buffer_t* _A = cinn_buffer_t::new_((cinn_device_kind_t)(0)/*target*/, cinn_float32_t(), { 100, 20 }, 32/*align*/);
+14: void func1(void* _args, int32_t num_args)
+14: {
+14:   cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+14:   cinn_buffer_malloc((void*)(0), _A);
+14:   float* A = ((float*)(_A->memory));
+14:   float* B = ((float*)(_A->memory));
+14:   for (int32_t i = 0; i < 100; i += 1) {
+14:     for (int32_t j = 0; j < 20; j += 1) {
+14:       A[((20 * i) + j)] = 0;
+14:     };
+14:   };
+14:   for (int32_t i = 0; i < 100; i += 1) {
+14:     for (int32_t j = 0; j < 20; j += 1) {
+14:       B[((20 * i) + j)] = 1;
+14:     };
+14:   };
+14:   cinn_buffer_free((void*)(0), _A);
+14: }
+14: 
+14: 
+14: [       OK ] Buffer.bind_to_multiple_tensors (10 ms)
+14: [----------] 2 tests from Buffer (10 ms total)
+14: 
+14: [----------] Global test environment tear-down
+14: [==========] 2 tests from 1 test case ran. (11 ms total)
+14: [  PASSED  ] 2 tests.
+53: W0924 13:32:18.341225 27786 codegen_c.cc:33] Output C source to file ./test02_matmul_vectorize.cc
+53: [       OK ] matmul.Vectorization (55 ms)
+53: [ RUN      ] matmul.LoopPermutation
+53: W0924 13:32:18.403461 27786 codegen_c.cc:24] Output C header to file ./test02_matmul_loop_permutation.h
+53: W0924 13:32:18.427978 27786 codegen_c.cc:33] Output C source to file ./test02_matmul_loop_permutation.cc
+53: [       OK ] matmul.LoopPermutation (87 ms)
+53: [ RUN      ] matmul.ArrayPacking
+13: [==========] Running 2 tests from 1 test case.
+13: [----------] Global test environment set-up.
+13: [----------] 2 tests from CollectIRNodes
+13: [ RUN      ] CollectIRNodes.basic0
+13: [       OK ] CollectIRNodes.basic0 (0 ms)
+13: [ RUN      ] CollectIRNodes.basic
+13: WARNING: Logging before InitGoogleLogging() is written to STDERR
+13: I0924 13:32:18.361411 27801 collect_ir_nodes_test.cc:32] fn:
+13: function fn (_A, _B, _C)
+13: {
+13:   for (i, 100)
+13:   {
+13:     for (j, 200)
+13:     {
+13:       C[i, j] = (A[i, j] + B[i, j])
+13:     }
+13:   }
+13: }
+13: I0924 13:32:18.361507 27801 collect_ir_nodes_test.cc:38] fn.body:
+13: {
+13:   for (i, 100)
+13:   {
+13:     for (j, 200)
+13:     {
+13:       C[i, j] = (A[i, j] + B[i, j])
+13:     }
+13:   }
+13: }
+13: [       OK ] CollectIRNodes.basic (7 ms)
+13: [----------] 2 tests from CollectIRNodes (7 ms total)
+13: 
+13: [----------] Global test environment tear-down
+13: [==========] 2 tests from 1 test case ran. (7 ms total)
+13: [  PASSED  ] 2 tests.
+12: [==========] Running 1 test from 1 test case.
+12: [----------] Global test environment set-up.
+12: [----------] 1 test from tanh
+12: [ RUN      ] tanh.basic
+12: WARNING: Logging before InitGoogleLogging() is written to STDERR
+12: I0924 13:32:18.361735 27800 host_intrinsics_test.cc:30] fn:
+12: function fn (_x, _tensor)
+12: {
+12:   for (i, 10)
+12:   {
+12:     for (j, 20)
+12:     {
+12:       tensor[i, j] = cinn_cpu_tanh_fp32(x[i, j])
+12:     }
+12:   }
+12: }
+12: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 800
+12: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 800
+12: I0924 13:32:18.421118 27800 host_intrinsics_test.cc:49] 0.654482
+12: I0924 13:32:18.421147 27800 host_intrinsics_test.cc:49] 0.663164
+12: I0924 13:32:18.421150 27800 host_intrinsics_test.cc:49] 0.721922
+12: [       OK ] tanh.basic (67 ms)
+12: [----------] 1 test from tanh (67 ms total)
+12: 
+12: [----------] Global test environment tear-down
+12: [==========] 1 test from 1 test case ran. (67 ms total)
+12: [  PASSED  ] 1 test.
+11: [==========] Running 27 tests from 2 test cases.
+11: [----------] Global test environment set-up.
+11: [----------] 26 tests from mkl_math
+11: [ RUN      ] mkl_math.exp
+11: WARNING: Logging before InitGoogleLogging() is written to STDERR
+11: I0924 13:32:18.364267 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_exp_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_exp_fp32_out[i, j] = cinn_cpu_exp_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.exp (33 ms)
+11: [ RUN      ] mkl_math.erf
+11: I0924 13:32:18.396248 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_erf_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_erf_fp32_out[i, j] = cinn_cpu_erf_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.erf (29 ms)
+11: [ RUN      ] mkl_math.sqrt
+11: I0924 13:32:18.424861 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_sqrt_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_sqrt_fp32_out[i, j] = cinn_cpu_sqrt_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.sqrt (29 ms)
+11: [ RUN      ] mkl_math.log
+11: I0924 13:32:18.453444 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_log_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_log_fp32_out[i, j] = cinn_cpu_log_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.log (28 ms)
+11: [ RUN      ] mkl_math.log2
+11: I0924 13:32:18.481892 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_log2_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_log2_fp32_out[i, j] = cinn_cpu_log2_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.log2 (29 ms)
+11: [ RUN      ] mkl_math.log10
+11: I0924 13:32:18.510504 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_log10_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_log10_fp32_out[i, j] = cinn_cpu_log10_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.log10 (28 ms)
+11: [ RUN      ] mkl_math.floor
+11: I0924 13:32:18.538941 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_floor_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_floor_fp32_out[i, j] = cinn_cpu_floor_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.floor (29 ms)
+11: [ RUN      ] mkl_math.ceil
+11: I0924 13:32:18.567494 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_ceil_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_ceil_fp32_out[i, j] = cinn_cpu_ceil_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+10: [==========] Running 3 tests from 3 test cases.
+10: [----------] Global test environment set-up.
+10: [----------] 1 test from buffer
+10: [ RUN      ] buffer.basic
+10: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 120
+10: [       OK ] buffer.basic (0 ms)
+10: [----------] 1 test from buffer (0 ms total)
+10: 
+10: [----------] 1 test from cinn_print_debug_string
+10: [ RUN      ] cinn_print_debug_string.basic
+10: hello world
+10: should be 1, 1
+10: should be pointer, 0x7ffc871d9534
+10: should be 1, 1
+10: v3[1 2 3], 
+10: [       OK ] cinn_print_debug_string.basic (0 ms)
+10: [----------] 1 test from cinn_print_debug_string (0 ms total)
+10: 
+10: [----------] 1 test from cinn_args_construct
+10: [ RUN      ] cinn_args_construct.basic
+10: [       OK ] cinn_args_construct.basic (0 ms)
+10: [----------] 1 test from cinn_args_construct (0 ms total)
+10: 
+10: [----------] Global test environment tear-down
+10: [==========] 3 tests from 3 test cases ran. (0 ms total)
+10: [  PASSED  ] 3 tests.
+9: [==========] Running 1 test from 1 test case.
+9: [----------] Global test environment set-up.
+9: [----------] 1 test from ComputeAtTransform2
+9: [ RUN      ] ComputeAtTransform2.basic
+9: WARNING: Logging before InitGoogleLogging() is written to STDERR
+9: I0924 13:32:18.362177 27797 compute_at_transform.cc:117] code:
+9: 
+9: {
+9:   if (_cp_c_0 >= 0 && _cp_c_0 <= 12 && _cp_c_1 == 0)
+9:     for (int c0 = 0; c0 <= 12; c0 += 1)
+9:       for (int c4 = 0; c4 <= 49; c4 += 1)
+9:         p(4 * _cp_c_0, c4);
+9:   for (int c0 = 0; c0 <= 12; c0 += 1)
+9:     for (int c2 = 0; c2 <= 49; c2 += 1)
+9:       for (int c3 = 0; c3 <= 49; c3 += 1)
+9:         c(4 * c0, c2, c3);
+9: }
+9: I0924 13:32:18.376412 27797 compute_at_transform.cc:117] code:
+9: 
+9: for (int c0 = 0; c0 <= 12; c0 += 1) {
+9:   if (_cp_c_0 >= 0 && _cp_c_0 <= 12 && _cp_c_1 == 0)
+9:     for (int c5 = 0; c5 <= 49; c5 += 1)
+9:       p(4 * _cp_c_0, c5);
+9:   for (int c3 = 0; c3 <= 49; c3 += 1)
+9:     for (int c4 = 0; c4 <= 49; c4 += 1)
+9:       c(4 * c0, c3, c4);
+9: }
+9: I0924 13:32:18.376475 27797 compute_at_transform_test.cc:31] shape:
+9: I0924 13:32:18.376705 27797 compute_at_transform_test.cc:34] 2
+9: I0924 13:32:18.376709 27797 compute_at_transform_test.cc:34] 50
+9: [       OK ] ComputeAtTransform2.basic (24 ms)
+9: [----------] 1 test from ComputeAtTransform2 (24 ms total)
+9: 
+9: [----------] Global test environment tear-down
+9: [==========] 1 test from 1 test case ran. (24 ms total)
+9: [  PASSED  ] 1 test.
+8: [==========] Running 23 tests from 10 test cases.
+8: [----------] Global test environment set-up.
+8: [----------] 6 tests from Stage
+8: [ RUN      ] Stage.split
+8: WARNING: Logging before InitGoogleLogging() is written to STDERR
+8: I0924 13:32:18.353860 27796 stage_test.cc:31] { S[i, j] -> S[i_outer, i_inner, j' = j] : (-i + i_inner) mod 4 = 0 and -3 + i <= 4i_outer <= i and 0 <= i_inner <= 3 }
+8: [       OK ] Stage.split (1 ms)
+8: [ RUN      ] Stage.tile
+8: I0924 13:32:18.355557 27796 stage_test.cc:46] { S[i, j, k] -> S[i_outer, i_inner, j_outer, j_inner, k' = k] : (-i + i_inner) mod 4 = 0 and (-j + j_inner) mod 6 = 0 and -3 + i <= 4i_outer <= i and 0 <= i_inner <= 3 and -5 + j <= 6j_outer <= j and 0 <= j_inner <= 5 }
+8: [       OK ] Stage.tile (2 ms)
+8: [ RUN      ] Stage.reorder
+8: I0924 13:32:18.356016 27796 stage_test.cc:63] { S[i, j, k] -> S[i' = i, k' = k, j' = j] }
+8: [       OK ] Stage.reorder (1 ms)
+8: [ RUN      ] Stage.split_reorder
+8: I0924 13:32:18.357061 27796 stage_test.cc:74] { S[i, j, k] -> S[i_outer, k' = k, i_inner, j' = j] : (-i + i_inner) mod 4 = 0 and -3 + i <= 4i_outer <= i and 0 <= i_inner <= 3 }
+8: [       OK ] Stage.split_reorder (1 ms)
+8: [ RUN      ] Stage.Fuse
+8: I0924 13:32:18.357940 27796 stage_test.cc:96] split: { S[i, j, k] -> S[i_outer, i_inner, j' = j, k' = k] : (-i + i_inner) mod 4 = 0 and -3 + i <= 4i_outer <= i and 0 <= i_inner <= 3 }
+8: I0924 13:32:18.358453 27796 stage_test.cc:98] fused: { S[i, j, k] -> S[i_outer_i_inner_fused, j' = j, k' = k] : (-i + i_outer_i_inner_fused) mod 4 = 0 and -3 + i_outer_i_inner_fused <= 4*floor((i)/4) <= i_outer_i_inner_fused }
+8: [       OK ] Stage.Fuse (1 ms)
+8: [ RUN      ] Stage.Fuse1
+8: I0924 13:32:18.358978 27796 stage_test.cc:108] fused: { S[i, j, k] -> S[i_j_fused = 101i + j, k' = k] }
+8: [       OK ] Stage.Fuse1 (0 ms)
+8: [----------] 6 tests from Stage (6 ms total)
+8: 
+8: [----------] 1 test from ComputeAtRelation
+8: [ RUN      ] ComputeAtRelation.basic
+8: [       OK ] ComputeAtRelation.basic (0 ms)
+8: [----------] 1 test from ComputeAtRelation (0 ms total)
+8: 
+8: [----------] 5 tests from ComputeAt
+8: [ RUN      ] ComputeAt.Before
+8: I0924 13:32:18.385377 27796 stage_test.cc:126] fn:
+8: function fn (_A, _B, _cache, _C)
+8: {
+8:   for (po0, 10)
+8:   {
+8:     for (po1, 10)
+8:     {
+8:       if (((((po0 >= 0) and (po0 <= 9)) and (po1 >= 0)) and (po1 <= 9))) {
+8:         cache[0, 0] = A[po0, po1]
+8:       }
+8:       C[po0, po1] = (cache[0, 0] + B[po0, po1])
+8:     }
+8:   }
+8: }
+8: [       OK ] ComputeAt.Before (26 ms)
+8: [ RUN      ] ComputeAt.level0
+8: I0924 13:32:18.412464 27796 stage_test.cc:165] fn:
+8: function fn (_A, _cache, _C)
+8: {
+8:   for (po0, 10)
+8:   {
+8:     if (((po0 >= 0) and (po0 <= 9))) {
+8:       for (j, 11)
+8:       {
+8:         for (k, 10)
+8:         {
+8:           cache[0, j, k] = A[po0, j, k]
+8:         }
+8:       }
+8:     }
+8:     for (i, 10)
+8:     {
+8:       for (j, 10)
+8:       {
+8:         C[po0, i, j] = select((i < 9), (cache[0, i, j] + cache[0, (1 + i), j]), 0)
+8:       }
+8:     }
+8:   }
+8: }
+8: I0924 13:32:18.412554 27796 stage_test.cc:198] C code:
+8: #include <cinn_runtime.h>
+8: #include <stdio.h>
+8: 
+8: void fn(void* _args, int32_t num_args)
+8: {
+8:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+8:   cinn_buffer_t* _cache = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+8:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+8:   cinn_buffer_malloc((void*)(0), _cache);
+8:   cinn_buffer_malloc((void*)(0), _C);
+8:   const float* A = ((const float*)(_A->memory));
+8:   float* C = ((float*)(_C->memory));
+8:   float* cache = ((float*)(_cache->memory));
+8:   for (int32_t po0 = 0; po0 < 10; po0 += 1) {
+8:     if (((po0 >= 0) && (po0 <= 9))) {
+8:       for (int32_t j = 0; j < 11; j += 1) {
+8:         for (int32_t k = 0; k < 10; k += 1) {
+8:           cache[((10 * j) + k)] = A[((25 * j) + ((750 * po0) + k))];
+8:         };
+8:       };
+8:     };
+8:     for (int32_t i = 0; i < 10; i += 1) {
+8:       for (int32_t j = 0; j < 10; j += 1) {
+8:         C[((10 * i) + ((100 * po0) + j))] = (((i < 9)) ? (cache[((10 * i) + j)] + cache[(10 + ((10 * i) + j))]) : 0);
+8:       };
+8:     };
+8:   };
+8:   cinn_buffer_free((void*)(0), _cache);
+8:   cinn_buffer_free((void*)(0), _C);
+8: }
+8: 
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 30000
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4000
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 440
+8: [       OK ] ComputeAt.level0 (56 ms)
+8: [ RUN      ] ComputeAt.level1
+8: I0924 13:32:18.474115 27796 stage_test.cc:250] fn:
+8: function fn (_A, _B, _cache, _C)
+8: {
+8:   for (po0, 10)
+8:   {
+8:     for (po1, 10)
+8:     {
+8:       if (((((po0 >= 0) and (po0 <= 9)) and (po1 >= 0)) and (po1 <= 9))) {
+8:         for (i, (1 + int32((1 + (po0 - cinn_max(0, (po0 - 1)))))))
+8:         {
+8:           cache[i, 0] = A[(i + cinn_max(0, (po0 - 1))), po1]
+8:         }
+8:       }
+8:       C[po0, po1] = select((po0 < 10), (cache[-1, 0] + (cache[0, 0] + (cache[1, 0] + B[po0, po1]))), 0)
+8:     }
+8:   }
+8: }
+8: I0924 13:32:18.474181 27796 stage_test.cc:276] source:
+8: #include <cinn_runtime.h>
+8: #include <stdio.h>
+8: 
+8: void fn(void* _args, int32_t num_args)
+8: {
+8:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+8:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+8:   cinn_buffer_t* _cache = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+8:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+8:   cinn_buffer_malloc((void*)(0), _cache);
+8:   cinn_buffer_malloc((void*)(0), _C);
+8:   const float* A = ((const float*)(_A->memory));
+8:   const float* B = ((const float*)(_B->memory));
+8:   float* C = ((float*)(_C->memory));
+8:   float* cache = ((float*)(_cache->memory));
+8:   for (int32_t po0 = 0; po0 < 10; po0 += 1) {
+8:     for (int32_t po1 = 0; po1 < 10; po1 += 1) {
+8:       if (((((po0 >= 0) && (po0 <= 9)) && (po1 >= 0)) && (po1 <= 9))) {
+8:         for (int32_t i = 0; i < (1 + ((int32_t)((1 + (po0 - cinn_max(0, (po0 - 1))))))); i += 1) {
+8:           cache[i] = A[((200 * i) + ((200 * cinn_max(0, (po0 - 1))) + po1))];
+8:         };
+8:       };
+8:       C[((10 * po0) + po1)] = (((po0 < 10)) ? (cache[-1] + (cache[0] + (cache[1] + B[((200 * po0) + po1)]))) : 0);
+8:     };
+8:   };
+8:   cinn_buffer_free((void*)(0), _cache);
+8:   cinn_buffer_free((void*)(0), _C);
+8: }
+8: 
+8: [       OK ] ComputeAt.level1 (35 ms)
+8: [ RUN      ] ComputeAt.simple
+8: I0924 13:32:18.531344 27796 stage_test.cc:313] fn:
+8: function fn (_A, _A1, _B)
+8: {
+8:   for (po0, 2)
+8:   {
+8:     for (po1, 16)
+8:     {
+8:       if (((((po1 >= 0) and (((16 * po0) + po1) >= 0)) and (po1 <= 15)) and (((16 * po0) + po1) <= 31))) {
+8:         for (i, 3)
+8:         {
+8:           for (j, 32)
+8:           {
+8:             A1[i, j] = A[(i + ((16 * po0) + po1)), j]
+8:           }
+8:         }
+8:       }
+8:       for (i, 32)
+8:       {
+8:         B[((16 * po0) + po1), i] = (A1[0, i] + (A1[1, i] + A1[2, i]))
+8:       }
+8:     }
+8:   }
+8: }
+8: I0924 13:32:18.531433 27796 stage_test.cc:346] source:
+8: #include <cinn_runtime.h>
+8: #include <stdio.h>
+8: 
+8: void fn(void* _args, int32_t num_args)
+8: {
+8:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+8:   cinn_buffer_t* _A1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+8:   cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+8:   cinn_buffer_malloc((void*)(0), _A1);
+8:   cinn_buffer_malloc((void*)(0), _B);
+8:   const float* A = ((const float*)(_A->memory));
+8:   float* A1 = ((float*)(_A1->memory));
+8:   float* B = ((float*)(_B->memory));
+8:   for (int32_t po0 = 0; po0 < 2; po0 += 1) {
+8:     for (int32_t po1 = 0; po1 < 16; po1 += 1) {
+8:       if (((((po1 >= 0) && (((16 * po0) + po1) >= 0)) && (po1 <= 15)) && (((16 * po0) + po1) <= 31))) {
+8:         for (int32_t i = 0; i < 3; i += 1) {
+8:           for (int32_t j = 0; j < 32; j += 1) {
+8:             A1[((32 * i) + j)] = A[((64 * i) + ((1024 * po0) + ((64 * po1) + j)))];
+8:           };
+8:         };
+8:       };
+8:       for (int32_t i = 0; i < 32; i += 1) {
+8:         B[((512 * po0) + ((32 * po1) + i))] = (A1[i] + (A1[(32 + i)] + A1[(64 + i)]));
+8:       };
+8:     };
+8:   };
+8:   cinn_buffer_free((void*)(0), _A1);
+8:   cinn_buffer_free((void*)(0), _B);
+8: }
+8: 
+8: [       OK ] ComputeAt.simple (60 ms)
+8: [ RUN      ] ComputeAt.Before1
+8: E0924 13:32:18.536700 27796 stage.cc:195] ComputeAt: transformed has no access to cache, skipped it
+8: I0924 13:32:18.545419 27796 stage_test.cc:373] fn:
+8: function fn (_A, _cache, _transformed)
+8: {
+8:   for (i, 100)
+8:   {
+8:     for (j, 200)
+8:     {
+8:       transformed[i, j] = 1
+8:     }
+8:   }
+8:   for (i, 100)
+8:   {
+8:     for (j, 200)
+8:     {
+8:       cache[i] = A[i, j]
+8:     }
+8:   }
+8: }
+8: E0924 13:32:18.546058 27796 stage.cc:195] ComputeAt: transformed has no access to cache, skipped it
+8: [       OK ] ComputeAt.Before1 (18 ms)
+8: [----------] 5 tests from ComputeAt (195 ms total)
+8: 
+8: [----------] 2 tests from Fuse
+8: [ RUN      ] Fuse.jit_precision_test
+8: I0924 13:32:18.560876 27796 stage_test.cc:444] fn:
+8: function fn (_A, _B, _C)
+8: {
+8:   for (i_j_fused, 1200)
+8:   {
+8:     C[(i_j_fused / 40), (i_j_fused % 40)] = (A[(i_j_fused / 40), (i_j_fused % 40)] + B[(i_j_fused / 40), (i_j_fused % 40)])
+8:   }
+8: }
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: I0924 13:32:18.581537 27796 stage_test.cc:466] 0.566904
+8: I0924 13:32:18.581550 27796 stage_test.cc:466] 0.818275
+8: I0924 13:32:18.581553 27796 stage_test.cc:466] 1.25823
+8: I0924 13:32:18.581555 27796 stage_test.cc:466] 0.186051
+8: [       OK ] Fuse.jit_precision_test (27 ms)
+8: [ RUN      ] Fuse.jit_precision_test2
+8: I0924 13:32:18.590202 27796 stage_test.cc:444] fn:
+8: function fn (_A, _B, _C)
+8: {
+8:   for (i_outer_i_inner_fused, 30)
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[i_outer_i_inner_fused, j] = (A[i_outer_i_inner_fused, j] + B[i_outer_i_inner_fused, j])
+8:     }
+8:   }
+8: }
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: I0924 13:32:18.610214 27796 stage_test.cc:466] 0.988877
+8: I0924 13:32:18.610225 27796 stage_test.cc:466] 0.858201
+8: I0924 13:32:18.610227 27796 stage_test.cc:466] 0.573442
+8: I0924 13:32:18.610230 27796 stage_test.cc:466] 0.873411
+8: [       OK ] Fuse.jit_precision_test2 (29 ms)
+8: [----------] 2 tests from Fuse (56 ms total)
+8: 
+8: [----------] 1 test from Tile
+8: [ RUN      ] Tile.jit_precision_test
+8: I0924 13:32:18.633318 27796 stage_test.cc:444] fn:
+8: function fn (_A, _B, _C)
+8: {
+8:   for (i_outer, 8)
+8:   {
+8:     for (i_inner, (1 + int32(cinn_min(3, (29 + (-4 * i_outer))))))
+8:     {
+8:       for (j_outer, 10)
+8:       {
+8:         for (j_inner, 4)
+8:         {
+8:           C[((4 * i_outer) + i_inner), ((4 * j_outer) + j_inner)] = (A[((4 * i_outer) + i_inner), ((4 * j_outer) + j_inner)] + B[((4 * i_outer) + i_inner), ((4 * j_outer) + j_inner)])
+8:         }
+8:       }
+8:     }
+8:   }
+8: }
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: I0924 13:32:18.655704 27796 stage_test.cc:466] 1.83821
+8: I0924 13:32:18.655715 27796 stage_test.cc:466] 0.631529
+8: I0924 13:32:18.655719 27796 stage_test.cc:466] 0.886526
+8: I0924 13:32:18.655721 27796 stage_test.cc:466] 1.68606
+8: [       OK ] Tile.jit_precision_test (45 ms)
+8: [----------] 1 test from Tile (45 ms total)
+8: 
+8: [----------] 1 test from Reorder
+8: [ RUN      ] Reorder.jit_precision_test
+8: I0924 13:32:18.662151 27796 stage_test.cc:444] fn:
+8: function fn (_A, _B, _C)
+8: {
+8:   for (j, 40)
+8:   {
+8:     for (i, 30)
+8:     {
+8:       C[i, j] = (A[i, j] + B[i, j])
+8:     }
+8:   }
+8: }
+15: [==========] Running 4 tests from 1 test case.
+15: [----------] Global test environment set-up.
+15: [----------] 4 tests from Tensor
+15: [ RUN      ] Tensor.inlined
+15: output: 
+15: function func_C (_A, _B, _D)
+15: {
+15:   for (i, 100)
+15:   {
+15:     for (j, 20)
+15:     {
+15:       D[i, j] = (1 + ((2 * A[i, j]) + (2 * B[i, j])))
+15:     }
+15:   }
+15: }
+15: [       OK ] Tensor.inlined (8 ms)
+15: [ RUN      ] Tensor.IsDependOnStatement
+15: [       OK ] Tensor.IsDependOnStatement (0 ms)
+15: [ RUN      ] Tensor.Reshape
+15: WARNING: Logging before InitGoogleLogging() is written to STDERR
+15: I0924 13:32:18.375077 27803 tensor_test.cc:85] source:
+15: #include <cinn_runtime.h>
+15: #include <stdio.h>
+15: 
+15: void fn(void* _args, int32_t num_args)
+15: {
+15:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+15:   cinn_buffer_t* _tensor = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+15:   cinn_buffer_malloc((void*)(0), _tensor);
+15:   const float* A_reshape = ((const float*)(_A->memory));
+15:   float* tensor = ((float*)(_tensor->memory));
+15:   for (int32_t i = 0; i < 10; i += 1) {
+15:     for (int32_t j = 0; j < 10; j += 1) {
+15:       for (int32_t k = 0; k < 100; k += 1) {
+15:         tensor[((1000 * i) + ((100 * j) + k))] = (2 * A_reshape[((1000 * i) + ((100 * j) + k))]);
+15:       };
+15:     };
+15:   };
+15:   cinn_buffer_free((void*)(0), _tensor);
+15: }
+15: 
+15: [       OK ] Tensor.Reshape (12 ms)
+15: [ RUN      ] Tensor.ReshapeCopied
+15: I0924 13:32:18.391335 27803 tensor_test.cc:131] source:
+15: #include <cinn_runtime.h>
+15: #include <stdio.h>
+15: 
+15: void fn(void* _args, int32_t num_args)
+15: {
+15:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+15:   cinn_buffer_t* _tensor = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+15:   cinn_buffer_malloc((void*)(0), _tensor);
+15:   const float* A = ((const float*)(_A->memory));
+15:   float* A_copied = ((float*)(_A_copied_reshape->memory));
+15:   const float* A_copied_reshape = ((const float*)(_A_copied_reshape->memory));
+15:   float* tensor = ((float*)(_tensor->memory));
+15:   for (int32_t i = 0; i < 100; i += 1) {
+15:     for (int32_t j = 0; j < 100; j += 1) {
+15:       A_copied[((100 * i) + j)] = A[((100 * i) + j)];
+15:     };
+15:   };
+15:   for (int32_t i = 0; i < 10; i += 1) {
+15:     for (int32_t j = 0; j < 10; j += 1) {
+15:       for (int32_t k = 0; k < 100; k += 1) {
+15:         tensor[((1000 * i) + ((100 * j) + k))] = (2 * A_copied_reshape[((1000 * i) + ((100 * j) + k))]);
+15:       };
+15:     };
+15:   };
+15:   cinn_buffer_free((void*)(0), _tensor);
+15: }
+15: 
+15: [       OK ] Tensor.ReshapeCopied (16 ms)
+15: [----------] 4 tests from Tensor (36 ms total)
+15: 
+15: [----------] Global test environment tear-down
+15: [==========] 4 tests from 1 test case ran. (36 ms total)
+15: [  PASSED  ] 4 tests.
+16: [==========] Running 6 tests from 1 test case.
+16: [----------] Global test environment set-up.
+16: [----------] 6 tests from CodeGenC
+16: [ RUN      ] CodeGenC.module
+16: WARNING: Logging before InitGoogleLogging() is written to STDERR
+16: I0924 13:32:18.357357 27804 codegen_c_test.cc:46] C.body: (A[i, j] + B[i, j])
+16: codegen C:
+16: #include <cinn_runtime.h>
+16: #include <stdio.h>
+16: 
+16: cinn_buffer_t* _C = cinn_buffer_t::new_((cinn_device_kind_t)(0)/*target*/, cinn_float32_t(), { 100, 20 }, 32/*align*/);
+16: void add1(void* _args, int32_t num_args)
+16: {
+16:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+16:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+16:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+16:   cinn_buffer_malloc((void*)(0), _C);
+16:   const float* A = ((const float*)(_A->memory));
+16:   const float* B = ((const float*)(_B->memory));
+16:   float* C = ((float*)(_C->memory));
+16:   for (int32_t i = 0; i < 100; i += 1) {
+16:     for (int32_t j = 0; j < 20; j += 1) {
+16:       C[((20 * i) + j)] = (A[((20 * i) + j)] + B[((20 * i) + j)]);
+16:     };
+16:   };
+16:   cinn_buffer_free((void*)(0), _C);
+16: }
+16: 
+16: 
+16: header:
+16: #ifndef _MODULE1_CINN_H_
+16: #define _MODULE1_CINN_H_
+16: 
+16: #include <cinn_runtime.h>
+16: #include <stdio.h>
+16: 
+16: void add1(void* _args, int32_t num_args);
+16: 
+16: 
+16: #endif  // _MODULE1_CINN_H_
+16: 
+16: W0924 13:32:18.365751 27804 codegen_c.cc:24] Output C header to file ./generated_module1.h
+16: W0924 13:32:18.366384 27804 codegen_c.cc:33] Output C source to file ./generated_module1.cc
+16: [       OK ] CodeGenC.module (9 ms)
+16: [ RUN      ] CodeGenC.module_with_transform
+16: I0924 13:32:18.403774 27804 codegen_c_test.cc:150] func:
+16: function add1 (_A, _B, _C, _D)
+16: {
+16:   for (i_outer, 25)
+16:   {
+16:     for (i_inner, 4)
+16:     {
+16:       for (j, 20)
+16:       {
+16:         C[((4 * i_outer) + i_inner), j] = (1 + ((3 * A[((4 * i_outer) + i_inner), j]) + B[((4 * i_outer) + i_inner), j]))
+16:       }
+16:     }
+16:   }
+16:   for (i_outer, 25)
+16:   {
+16:     for (i_inner, 4)
+16:     {
+16:       for (j_outer, 2)
+16:       {
+16:         for (j_inner, (1 + int32(cinn_min(15, (19 + (-16 * j_outer))))))
+16:         {
+16:           D[((4 * i_outer) + i_inner), ((16 * j_outer) + j_inner)] = ((2 * C[((4 * i_outer) + i_inner), ((16 * j_outer) + j_inner)]) + (4 * (C[((4 * i_outer) + i_inner), ((16 * j_outer) + j_inner)] * A[((4 * i_outer) + i_inner), ((16 * j_outer) + j_inner)])))
+16:         }
+16:       }
+16:     }
+16:   }
+16: }
+16: codegen C:
+16: #include <cinn_runtime.h>
+16: #include <stdio.h>
+16: 
+16: cinn_buffer_t* _C = cinn_buffer_t::new_((cinn_device_kind_t)(0)/*target*/, cinn_float32_t(), { 100, 20 }, 32/*align*/);
+16: void add1(void* _args, int32_t num_args)
+16: {
+16:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+16:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+16:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+16:   cinn_buffer_t* _D = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+16:   cinn_buffer_malloc((void*)(0), _C);
+16:   cinn_buffer_malloc((void*)(0), _D);
+16:   const float* A = ((const float*)(_A->memory));
+16:   const float* B = ((const float*)(_B->memory));
+16:   float* C = ((float*)(_C->memory));
+16:   float* D = ((float*)(_D->memory));
+16:   for (int32_t i_outer = 0; i_outer < 25; i_outer += 1) {
+16:     for (int32_t i_inner = 0; i_inner < 4; i_inner += 1) {
+16:       for (int32_t j = 0; j < 20; j += 1) {
+16:         C[((20 * i_inner) + ((80 * i_outer) + j))] = (1 + ((3 * A[((20 * i_inner) + ((80 * i_outer) + j))]) + B[((20 * i_inner) + ((80 * i_outer) + j))]));
+16:       };
+16:     };
+16:   };
+16:   for (int32_t i_outer = 0; i_outer < 25; i_outer += 1) {
+16:     for (int32_t i_inner = 0; i_inner < 4; i_inner += 1) {
+16:       for (int32_t j_outer = 0; j_outer < 2; j_outer += 1) {
+16:         for (int32_t j_inner = 0; j_inner < (1 + ((int32_t)(cinn_min(15, (19 + (-16 * j_outer)))))); j_inner += 1) {
+16:           D[((20 * i_inner) + ((80 * i_outer) + ((16 * j_outer) + j_inner)))] = ((2 * C[((20 * i_inner) + ((80 * i_outer) + ((16 * j_outer) + j_inner)))]) + (4 * (C[((20 * i_inner) + ((80 * i_outer) + ((16 * j_outer) + j_inner)))] * A[((20 * i_inner) + ((80 * i_outer) + ((16 * j_outer) + j_inner)))])));
+16:         };
+16:       };
+16:     };
+16:   };
+16:   cinn_buffer_free((void*)(0), _C);
+16:   cinn_buffer_free((void*)(0), _D);
+16: }
+16: 
+16: 
+16: [       OK ] CodeGenC.module_with_transform (47 ms)
+16: [ RUN      ] CodeGenC.matmul
+16: I0924 13:32:18.429903 27804 codegen_c_test.cc:238] C.body: matmul(Tensor(A, [100,20]), Tensor(B, [20,50]), Tensor(C, [100,50]))
+16: f
+16: function main (_A, _B, _C)
+16: {
+16:   matmul(_Buffer_<float32: 100,20>(_A), _Buffer_<float32: 20,50>(_B), _Buffer_<float32: 100,50>(_C))
+16: }
+16: codegen C:
+16: #include <cinn_runtime.h>
+16: #include <stdio.h>
+16: 
+16: cinn_buffer_t* _C = cinn_buffer_t::new_((cinn_device_kind_t)(0)/*target*/, cinn_float32_t(), { 100, 50 });
+16: void matmul(void* _args, int32_t num_args)
+16: {
+16:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+16:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+16:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+16:   cinn_buffer_malloc((void*)(0), _C);
+16:   const float* A = ((const float*)(_A->memory));
+16:   const float* B = ((const float*)(_B->memory));
+16:   float* C = ((float*)(_C->memory));
+16:   float* C_init = ((float*)(_C->memory));
+16:   for (int32_t i = 0; i < 100; i += 1) {
+16:     for (int32_t j = 0; j < 50; j += 1) {
+16:       C_init[((50 * i) + j)] = 0;
+16:       for (int32_t k0 = 0; k0 < 20; k0 += 1) {
+16:         C[((50 * i) + j)] = (C[((50 * i) + j)] + (A[((20 * i) + k0)] * B[((50 * k0) + j)]));
+16:       };
+16:     };
+16:   };
+16:   cinn_buffer_free((void*)(0), _C);
+16: }
+16: 
+16: void main(void* _args, int32_t num_args)
+16: {
+16:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+16:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+16:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+16:   cinn_buffer_malloc((void*)(0), _C);
+16:   const float* A = ((const float*)(_A->memory));
+16:   const float* B = ((const float*)(_B->memory));
+16:   float* C = ((float*)(_C->memory));
+16:   {
+16:     cinn_pod_value_t _pod_val_;
+16:     buffer_p_to_cinn_pod_value(_A, &_pod_val_);
+16:     cinn_pod_value_t _pod_val__0;
+16:     buffer_p_to_cinn_pod_value(_B, &_pod_val__0);
+16:     cinn_pod_value_t _pod_val__1;
+16:     buffer_p_to_cinn_pod_value(_C, &_pod_val__1);
+16:     cinn_pod_value_t _pod_arr[3];
+16:     cinn_args_construct(_pod_arr, 3, &_pod_val_, &_pod_val__0, &_pod_val__1);
+16:     matmul(_pod_arr, 5);
+16:   };
+16:   cinn_buffer_free((void*)(0), _C);
+16: }
+16: 
+16: 
+16: [       OK ] CodeGenC.matmul (20 ms)
+16: [ RUN      ] CodeGenC.matmul_tile
+16: codegen C:
+16: #include <cinn_runtime.h>
+16: #include <stdio.h>
+16: 
+16: cinn_buffer_t* _C = cinn_buffer_t::new_((cinn_device_kind_t)(0)/*target*/, cinn_float32_t(), { 100, 500 }, 32/*align*/);
+16: void matmul(void* _args, int32_t num_args)
+16: {
+16:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+16:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+16:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+16:   cinn_buffer_malloc((void*)(0), _C);
+16:   const float* A = ((const float*)(_A->memory));
+16:   const float* B = ((const float*)(_B->memory));
+16:   float* C = ((float*)(_C->memory));
+16:   float* C_init = ((float*)(_C->memory));
+16:   for (int32_t i_outer = 0; i_outer < 4; i_outer += 1) {
+16:     for (int32_t j_outer = 0; j_outer < 16; j_outer += 1) {
+16:       for (int32_t i_inner = 0; i_inner < (1 + ((int32_t)(cinn_min(31, (99 + (-32 * i_outer)))))); i_inner += 1) {
+16:         for (int32_t j_inner = 0; j_inner < (1 + ((int32_t)(cinn_min(31, (499 + (-32 * j_outer)))))); j_inner += 1) {
+16:           C_init[((500 * i_inner) + ((16000 * i_outer) + ((32 * j_outer) + j_inner)))] = 0;
+16:           for (int32_t k0_outer = 0; k0_outer < 50; k0_outer += 1) {
+16:             for (int32_t k0_inner = 0; k0_inner < 4; k0_inner += 1) {
+16:               C[((500 * i_inner) + ((16000 * i_outer) + ((32 * j_outer) + j_inner)))] = (C[((500 * i_inner) + ((16000 * i_outer) + ((32 * j_outer) + j_inner)))] + (A[((200 * i_inner) + ((6400 * i_outer) + ((4 * k0_outer) + k0_inner)))] * B[((32 * j_outer) + ((500 * k0_inner) + ((2000 * k0_outer) + j_inner)))]));
+16:             };
+16:           };
+16:         };
+16:       };
+16:     };
+16:   };
+16:   cinn_buffer_free((void*)(0), _C);
+16: }
+16: 
+16: 
+16: [       OK ] CodeGenC.matmul_tile (116 ms)
+16: [ RUN      ] CodeGenC.matmul_packed
+16: codegen C:
+16: #include <cinn_runtime.h>
+16: #include <stdio.h>
+16: 
+16: cinn_buffer_t* _C = cinn_buffer_t::new_((cinn_device_kind_t)(0)/*target*/, cinn_float32_t(), { 100, 500 }, 32/*align*/);
+16: cinn_buffer_t* _PackedB = cinn_buffer_t::new_((cinn_device_kind_t)(0)/*target*/, cinn_float32_t(), { 15, 200, 32 }, 32/*align*/);
+16: void matmul_with_packing(void* _args, int32_t num_args)
+16: {
+16:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+16:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+16:   cinn_buffer_t* _PackedB = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+16:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+16:   cinn_buffer_malloc((void*)(0), _PackedB);
+16:   cinn_buffer_malloc((void*)(0), _C);
+16:   const float* A = ((const float*)(_A->memory));
+16:   const float* B = ((const float*)(_B->memory));
+16:   float* C = ((float*)(_C->memory));
+16:   float* PackedB = ((float*)(_PackedB->memory));
+16:   for (int32_t i = 0; i < 15; i += 1) {
+16:     for (int32_t j = 0; j < 200; j += 1) {
+16:       for (int32_t k = 0; k < 32; k += 1) {
+16:         PackedB[((6400 * i) + ((32 * j) + k))] = B[((32 * i) + ((500 * j) + k))];
+16:       };
+16:     };
+16:   };
+16:   for (int32_t i_outer = 0; i_outer < 4; i_outer += 1) {
+16:     for (int32_t j_outer = 0; j_outer < 16; j_outer += 1) {
+16:       for (int32_t i_inner = 0; i_inner < (1 + ((int32_t)(cinn_min(31, (99 + (-32 * i_outer)))))); i_inner += 1) {
+16:         for (int32_t j_inner = 0; j_inner < (1 + ((int32_t)(cinn_min(31, (499 + (-32 * j_outer)))))); j_inner += 1) {
+16:           for (int32_t k0_outer = 0; k0_outer < 50; k0_outer += 1) {
+16:             for (int32_t k0_inner = 0; k0_inner < 4; k0_inner += 1) {
+16:               C[((500 * i_inner) + ((16000 * i_outer) + ((32 * j_outer) + j_inner)))] = (A[((200 * i_inner) + ((6400 * i_outer) + ((4 * k0_outer) + k0_inner)))] * PackedB[((j_inner % 32) + ((6400 * (j_inner / 32)) + ((6400 * j_outer) + ((32 * k0_inner) + (128 * k0_outer)))))]);
+16:             };
+16:           };
+16:         };
+16:       };
+16:     };
+16:   };
+16:   cinn_buffer_free((void*)(0), _PackedB);
+16:   cinn_buffer_free((void*)(0), _C);
+16: }
+16: 
+16: 
+16: [       OK ] CodeGenC.matmul_packed (66 ms)
+16: [ RUN      ] CodeGenC.call_extern
+16: codegen C:
+16: #include <cinn_runtime.h>
+16: #include <stdio.h>
+16: 
+16: void yy(void* _args, int32_t num_args)
+16: {
+16:   cinn_buffer_t* _y = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+16:   cinn_buffer_malloc((void*)(0), _y);
+16:   cinn_buffer_malloc((void*)(0), _x);
+16:   const float* x = ((const float*)(_x->memory));
+16:   float* y = ((float*)(_y->memory));
+16:   for (int32_t i = 0; i < 100; i += 1) {
+16:     y[i] = cinn_cpu_tanh_fp32(x[i]);
+16:   };
+16:   cinn_buffer_free((void*)(0), _y);
+16: }
+16: 
+16: 
+16: [       OK ] CodeGenC.call_extern (3 ms)
+16: [----------] 6 tests from CodeGenC (261 ms total)
+16: 
+16: [----------] Global test environment tear-down
+16: [==========] 6 tests from 1 test case ran. (261 ms total)
+16: [  PASSED  ] 6 tests.
+10/68 Test #14: test_buffer ............................   Passed    0.03 sec
+11/68 Test #13: test_collect_ir_nodes ..................   Passed    0.13 sec
+12/68 Test #12: test_host_intrinsics ...................   Passed    0.13 sec
+13/68 Test #10: test_cinn_runtime ......................   Passed    0.23 sec
+14/68 Test  #9: test_compute_at_transform ..............   Passed    0.23 sec
+15/68 Test #15: test_tensor ............................   Passed    0.33 sec
+16/68 Test #16: test_codegen_c .........................   Passed    0.33 sec
+test 17
+      Start 17: test_codegen_c_x86
+
+17: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/backends/test_codegen_c_x86 "--cinn_x86_builtin_code_root=/home/wangyue50/CINN-my/CINN/cinn/backends"
+17: Test timeout computed to be: 600
+test 18
+      Start 18: test_generated1
+
+18: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/backends/test_generated1 ""
+18: Test timeout computed to be: 600
+test 19
+      Start 19: test_compiler
+
+19: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/backends/test_compiler ""
+19: Test timeout computed to be: 600
+test 20
+      Start 20: test_codegen_llvm
+
+20: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/backends/llvm/test_codegen_llvm ""
+20: Test timeout computed to be: 600
+test 21
+      Start 21: test_execution_engine
+
+21: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/backends/llvm/test_execution_engine ""
+21: Test timeout computed to be: 600
+test 22
+      Start 22: test_codegen_x86
+
+22: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/backends/llvm/test_codegen_x86 ""
+22: Test timeout computed to be: 600
+test 23
+      Start 23: test_compute
+
+23: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/lang/test_compute ""
+23: Test timeout computed to be: 600
+53: W0924 13:32:18.485375 27786 codegen_c.cc:24] Output C header to file ./test02_matmul_array_packing.h
+53: W0924 13:32:18.496706 27786 codegen_c.cc:33] Output C source to file ./test02_matmul_array_packing.cc
+53: [       OK ] matmul.ArrayPacking (68 ms)
+53: [ RUN      ] matmul.varient_shape
+53: W0924 13:32:18.512287 27786 codegen_c.cc:24] Output C header to file ./test02_matmul_varient_shape.h
+53: W0924 13:32:18.513252 27786 codegen_c.cc:33] Output C source to file ./test02_matmul_varient_shape.cc
+53: I0924 13:32:18.549512 27786 test02_matmul_main.cc:203] func function matmul_dynamic_shape_tile (M, _A, _B, _C)
+53: {
+53:   for (i, int32(M))
+53:   {
+53:     for (j, 1024)
+53:     {
+53:       C_init_4[i, j] = 0
+53:     }
+53:   }
+53:   for (i_outer, (1 + int32(((-1 + M) / 32))))
+53:   {
+53:     for (i_inner, (1 + int32(cinn_min(31, (-1 + (M + (-32 * i_outer)))))))
+53:     {
+53:       for (j_outer, 32)
+53:       {
+53:         for (j_inner, 32)
+53:         {
+53:           for (k0, 1024)
+53:           {
+53:             C[((32 * i_outer) + i_inner), ((32 * j_outer) + j_inner)] = (C[((32 * i_outer) + i_inner), ((32 * j_outer) + j_inner)] + (A[((32 * i_outer) + i_inner), k0] * B[k0, ((32 * j_outer) + j_inner)]))
+53:           }
+53:         }
+53:       }
+53:     }
+53:   }
+53: }
+53: W0924 13:32:18.550587 27786 codegen_c.cc:24] Output C header to file ./test02_matmul_varient_shape_tile.h
+53: W0924 13:32:18.555127 27786 codegen_c.cc:33] Output C source to file ./test02_matmul_varient_shape_tile.cc
+53: [       OK ] matmul.varient_shape (59 ms)
+53: [ RUN      ] matmul.ArrayPacking_dynamic_shape
+53: W0924 13:32:18.622040 27786 codegen_c.cc:24] Output C header to file ./test02_matmul_array_packing_dynamic_shape.h
+53: W0924 13:32:18.633481 27786 codegen_c.cc:33] Output C source to file ./test02_matmul_array_packing_dynamic_shape.cc
+53: [       OK ] matmul.ArrayPacking_dynamic_shape (78 ms)
+53: [ RUN      ] matmul.call
+53: I0924 13:32:18.647380 27786 test02_matmul_main.cc:284] stage domain: { C[i, j, k0] : 0 <= i <= 1023 and 0 <= j <= 1023 and 0 <= k0 <= 1023 }
+53: W0924 13:32:18.659453 27786 codegen_c.cc:24] Output C header to file ./test02_matmul_call.h
+53: W0924 13:32:18.660598 27786 codegen_c.cc:33] Output C source to file ./test02_matmul_call.cc
+53: [       OK ] matmul.call (27 ms)
+53: [----------] 8 tests from matmul (449 ms total)
+53: 
+53: [----------] Global test environment tear-down
+53: [==========] 9 tests from 2 test cases ran. (501 ms total)
+53: [  PASSED  ] 9 tests.
+22: [==========] Running 1 test from 1 test case.
+22: [----------] Global test environment set-up.
+22: [----------] 1 test from Vectorize
+22: [ RUN      ] Vectorize.basic
+22: WARNING: Logging before InitGoogleLogging() is written to STDERR
+22: I0924 13:32:18.694113 27810 codegen_x86_test.cc:25] fn: function fn (_A, _B, _tensor)
+22: {
+22:   for (i, 128)
+22:   {
+22:     tensor[Ramp((8 * i),1,8)] = (A[Ramp((8 * i),1,8)] + B[Ramp((8 * i),1,8)])
+22:   }
+22: }
+22: I0924 13:32:18.694416 27810 codegen_x86_test.cc:32] 
+22: function fn (_A, _B, _tensor)
+22: {
+22:   for (i, 128)
+22:   {
+22:     tensor[Ramp((8 * i),1,8)] = (A[Ramp((8 * i),1,8)] + B[Ramp((8 * i),1,8)])
+22:   }
+22: }
+22: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+22: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+22: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+22: [       OK ] Vectorize.basic (61 ms)
+22: [----------] 1 test from Vectorize (61 ms total)
+22: 
+22: [----------] Global test environment tear-down
+22: [==========] 1 test from 1 test case ran. (62 ms total)
+22: [  PASSED  ] 1 test.
+21: [==========] Running 4 tests from 3 test cases.
+21: [----------] Global test environment set-up.
+21: [----------] 1 test from llvm_test01
+21: [ RUN      ] llvm_test01.elementwise_add
+21: [       OK ] llvm_test01.elementwise_add (0 ms)
+21: [----------] 1 test from llvm_test01 (0 ms total)
+21: 
+21: [----------] 1 test from llvm
+21: [ RUN      ] llvm.module_call_lowered_func
+21: WARNING: Logging before InitGoogleLogging() is written to STDERR
+21: I0924 13:32:18.698501 27809 execution_engine_test.cc:165] module:
+21: #include <cinn_runtime.h>
+21: #include <stdio.h>
+21: 
+21: void elementwise_add(void* _args, int32_t num_args)
+21: {
+21:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+21:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+21:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+21:   cinn_buffer_malloc((void*)(0), _C);
+21:   const float* A = ((const float*)(_A->memory));
+21:   const float* B = ((const float*)(_B->memory));
+21:   float* C = ((float*)(_C->memory));
+21:   for (int32_t i = 0; i < 100; i += 1) {
+21:     for (int32_t j = 0; j < 32; j += 1) {
+21:       C[((32 * i) + j)] = (A[((32 * i) + j)] + B[((32 * i) + j)]);
+21:     };
+21:   };
+21:   cinn_buffer_free((void*)(0), _C);
+21: }
+21: 
+21: void main(void* _args, int32_t num_args)
+21: {
+21:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+21:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+21:   cinn_buffer_t* _c_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+21:   cinn_buffer_malloc((void*)(0), _c_out);
+21:   const float* A = ((const float*)(_A->memory));
+21:   const float* B = ((const float*)(_B->memory));
+21:   float* c_out = ((float*)(_c_out->memory));
+21:   {
+21:     cinn_pod_value_t _pod_val_;
+21:     buffer_p_to_cinn_pod_value(_A, &_pod_val_);
+21:     cinn_pod_value_t _pod_val__0;
+21:     buffer_p_to_cinn_pod_value(_B, &_pod_val__0);
+21:     cinn_pod_value_t _pod_val__1;
+21:     buffer_p_to_cinn_pod_value(_c_out, &_pod_val__1);
+21:     cinn_pod_value_t _pod_arr[3];
+21:     cinn_args_construct(_pod_arr, 3, &_pod_val_, &_pod_val__0, &_pod_val__1);
+21:     elementwise_add(_pod_arr, 5);
+21:   };
+21:   cinn_buffer_free((void*)(0), _c_out);
+21: }
+21: 
+21: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+21: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+21: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+21: I0924 13:32:18.700588 27809 execution_engine_test.cc:172] JIT Link the module
+21: I0924 13:32:18.709628 27809 execution_engine_test.cc:175] => LLVM JIT cos(0) = 1
+21: I0924 13:32:18.726971 27809 execution_engine_test.cc:178] JIT get elementwise_add_addr
+21: [       OK ] llvm.module_call_lowered_func (39 ms)
+21: [----------] 1 test from llvm (39 ms total)
+21: 
+21: [----------] 2 tests from ExecutionEngine
+21: [ RUN      ] ExecutionEngine.custom_runtime_symbols
+21: [       OK ] ExecutionEngine.custom_runtime_symbols (1 ms)
+21: [ RUN      ] ExecutionEngine.call_extern
+21: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+21: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+21: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+21: [       OK ] ExecutionEngine.call_extern (31 ms)
+21: [----------] 2 tests from ExecutionEngine (32 ms total)
+21: 
+21: [----------] Global test environment tear-down
+21: [==========] 4 tests from 3 test cases ran. (71 ms total)
+21: [  PASSED  ] 4 tests.
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.ceil (28 ms)
+11: [ RUN      ] mkl_math.round
+11: I0924 13:32:18.596024 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_round_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_round_fp32_out[i, j] = cinn_cpu_round_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.round (29 ms)
+11: [ RUN      ] mkl_math.trunc
+11: I0924 13:32:18.624581 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_trunc_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_trunc_fp32_out[i, j] = cinn_cpu_trunc_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.trunc (29 ms)
+11: [ RUN      ] mkl_math.cos
+11: I0924 13:32:18.653442 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_cos_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_cos_fp32_out[i, j] = cinn_cpu_cos_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.cos (28 ms)
+11: [ RUN      ] mkl_math.cosh
+11: I0924 13:32:18.682230 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_cosh_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_cosh_fp32_out[i, j] = cinn_cpu_cosh_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.cosh (30 ms)
+11: [ RUN      ] mkl_math.tan
+11: I0924 13:32:18.711864 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_tan_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_tan_fp32_out[i, j] = cinn_cpu_tan_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.tan (29 ms)
+11: [ RUN      ] mkl_math.sin
+11: I0924 13:32:18.740679 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_sin_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_sin_fp32_out[i, j] = cinn_cpu_sin_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.sin (29 ms)
+11: [ RUN      ] mkl_math.sinh
+11: I0924 13:32:18.769448 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_sinh_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_sinh_fp32_out[i, j] = cinn_cpu_sinh_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.sinh (28 ms)
+11: [ RUN      ] mkl_math.acos
+11: I0924 13:32:18.798184 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_acos_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_acos_fp32_out[i, j] = cinn_cpu_acos_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.acos (29 ms)
+11: [ RUN      ] mkl_math.acosh
+11: I0924 13:32:18.826864 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_acosh_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_acosh_fp32_out[i, j] = cinn_cpu_acosh_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.acosh (29 ms)
+11: [ RUN      ] mkl_math.asin
+11: I0924 13:32:18.855567 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_asin_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_asin_fp32_out[i, j] = cinn_cpu_asin_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+20: [==========] Running 5 tests from 2 test cases.
+20: [----------] Global test environment set-up.
+20: [----------] 4 tests from CodeGenLLVM
+20: [ RUN      ] CodeGenLLVM.Imm
+20: [       OK ] CodeGenLLVM.Imm (0 ms)
+20: [ RUN      ] CodeGenLLVM.Expr
+20: [       OK ] CodeGenLLVM.Expr (0 ms)
+20: [ RUN      ] CodeGenLLVM.Statement
+20: [       OK ] CodeGenLLVM.Statement (0 ms)
+20: [ RUN      ] CodeGenLLVM.LowerFunc
+20: WARNING: Logging before InitGoogleLogging() is written to STDERR
+20: I0924 13:32:18.699189 27808 codegen_llvm_test.cc:457] ir function: function add1 (_a, _b, _c)
+20: {
+20:   for (i, 3)
+20:   {
+20:     for (j, 2)
+20:     {
+20:       c[i, j] = (a[i, j] + b[i, j])
+20:     }
+20:   }
+20: }
+20: [       OK ] CodeGenLLVM.LowerFunc (11 ms)
+20: [----------] 4 tests from CodeGenLLVM (11 ms total)
+20: 
+20: [----------] 1 test from SymbolTable
+20: [ RUN      ] SymbolTable.test
+20: [       OK ] SymbolTable.test (0 ms)
+20: [----------] 1 test from SymbolTable (0 ms total)
+20: 
+20: [----------] Global test environment tear-down
+20: [==========] 5 tests from 2 test cases ran. (11 ms total)
+20: [  PASSED  ] 5 tests.
+19: [==========] Running 2 tests from 1 test case.
+19: [----------] Global test environment set-up.
+19: [----------] 2 tests from Compiler
+19: [ RUN      ] Compiler.x86
+19: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 800
+19: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 800
+19: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 800
+19: [       OK ] Compiler.x86 (34 ms)
+19: [ RUN      ] Compiler.sqrt
+19: WARNING: Logging before InitGoogleLogging() is written to STDERR
+19: I0924 13:32:18.722045 27807 nn.cc:477] kernel_size length is: 2
+19: I0924 13:32:18.722067 27807 nn.cc:478] kernel_size is: 3
+19: I0924 13:32:18.722069 27807 nn.cc:479] padding_size length is: 4
+19: I0924 13:32:18.722071 27807 nn.cc:480] padding_size is: 1
+19: [       OK ] Compiler.sqrt (71 ms)
+19: [----------] 2 tests from Compiler (105 ms total)
+19: 
+19: [----------] Global test environment tear-down
+19: [==========] 2 tests from 1 test case ran. (105 ms total)
+19: [  PASSED  ] 2 tests.
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: I0924 13:32:18.682731 27796 stage_test.cc:466] 1.42271
+8: I0924 13:32:18.682749 27796 stage_test.cc:466] 0.919059
+8: I0924 13:32:18.682751 27796 stage_test.cc:466] 0.962657
+8: I0924 13:32:18.682754 27796 stage_test.cc:466] 1.50734
+8: [       OK ] Reorder.jit_precision_test (28 ms)
+8: [----------] 1 test from Reorder (28 ms total)
+8: 
+8: [----------] 2 tests from Unroll
+8: [ RUN      ] Unroll.jit_precision_test
+8: I0924 13:32:18.713403 27796 stage_test.cc:444] fn:
+8: function fn (_A, _B, _C)
+8: {
+8:   for (i, 30)
+8:   {
+8:     {
+8:       C[i, 0] = (A[i, 0] + B[i, 0])
+8:     }
+8:     {
+8:       C[i, 1] = (A[i, 1] + B[i, 1])
+8:     }
+8:     {
+8:       C[i, 2] = (A[i, 2] + B[i, 2])
+8:     }
+8:     {
+8:       C[i, 3] = (A[i, 3] + B[i, 3])
+8:     }
+8:     {
+8:       C[i, 4] = (A[i, 4] + B[i, 4])
+8:     }
+8:     {
+8:       C[i, 5] = (A[i, 5] + B[i, 5])
+8:     }
+8:     {
+8:       C[i, 6] = (A[i, 6] + B[i, 6])
+8:     }
+8:     {
+8:       C[i, 7] = (A[i, 7] + B[i, 7])
+8:     }
+8:     {
+8:       C[i, 8] = (A[i, 8] + B[i, 8])
+8:     }
+8:     {
+8:       C[i, 9] = (A[i, 9] + B[i, 9])
+8:     }
+8:     {
+8:       C[i, 10] = (A[i, 10] + B[i, 10])
+8:     }
+8:     {
+8:       C[i, 11] = (A[i, 11] + B[i, 11])
+8:     }
+8:     {
+8:       C[i, 12] = (A[i, 12] + B[i, 12])
+8:     }
+8:     {
+8:       C[i, 13] = (A[i, 13] + B[i, 13])
+8:     }
+8:     {
+8:       C[i, 14] = (A[i, 14] + B[i, 14])
+8:     }
+8:     {
+8:       C[i, 15] = (A[i, 15] + B[i, 15])
+8:     }
+8:     {
+8:       C[i, 16] = (A[i, 16] + B[i, 16])
+8:     }
+8:     {
+8:       C[i, 17] = (A[i, 17] + B[i, 17])
+8:     }
+8:     {
+8:       C[i, 18] = (A[i, 18] + B[i, 18])
+8:     }
+8:     {
+8:       C[i, 19] = (A[i, 19] + B[i, 19])
+8:     }
+8:     {
+8:       C[i, 20] = (A[i, 20] + B[i, 20])
+8:     }
+8:     {
+8:       C[i, 21] = (A[i, 21] + B[i, 21])
+8:     }
+8:     {
+8:       C[i, 22] = (A[i, 22] + B[i, 22])
+8:     }
+8:     {
+8:       C[i, 23] = (A[i, 23] + B[i, 23])
+8:     }
+8:     {
+8:       C[i, 24] = (A[i, 24] + B[i, 24])
+8:     }
+8:     {
+8:       C[i, 25] = (A[i, 25] + B[i, 25])
+8:     }
+8:     {
+8:       C[i, 26] = (A[i, 26] + B[i, 26])
+8:     }
+8:     {
+8:       C[i, 27] = (A[i, 27] + B[i, 27])
+8:     }
+8:     {
+8:       C[i, 28] = (A[i, 28] + B[i, 28])
+8:     }
+8:     {
+8:       C[i, 29] = (A[i, 29] + B[i, 29])
+8:     }
+8:     {
+8:       C[i, 30] = (A[i, 30] + B[i, 30])
+8:     }
+8:     {
+8:       C[i, 31] = (A[i, 31] + B[i, 31])
+8:     }
+8:     {
+8:       C[i, 32] = (A[i, 32] + B[i, 32])
+8:     }
+8:     {
+8:       C[i, 33] = (A[i, 33] + B[i, 33])
+8:     }
+8:     {
+8:       C[i, 34] = (A[i, 34] + B[i, 34])
+8:     }
+8:     {
+8:       C[i, 35] = (A[i, 35] + B[i, 35])
+8:     }
+8:     {
+8:       C[i, 36] = (A[i, 36] + B[i, 36])
+8:     }
+8:     {
+8:       C[i, 37] = (A[i, 37] + B[i, 37])
+8:     }
+8:     {
+8:       C[i, 38] = (A[i, 38] + B[i, 38])
+8:     }
+8:     {
+8:       C[i, 39] = (A[i, 39] + B[i, 39])
+8:     }
+8:   }
+8: }
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: I0924 13:32:18.747979 27796 stage_test.cc:466] 1.14587
+8: I0924 13:32:18.747992 27796 stage_test.cc:466] 0.950917
+8: I0924 13:32:18.747995 27796 stage_test.cc:466] 1.1457
+8: I0924 13:32:18.747997 27796 stage_test.cc:466] 0.782843
+8: [       OK ] Unroll.jit_precision_test (65 ms)
+8: [ RUN      ] Unroll.jit_precision_test1
+8: I0924 13:32:18.773546 27796 stage_test.cc:444] fn:
+8: function fn (_A, _B, _C)
+8: {
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[0, j] = (A[0, j] + B[0, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[1, j] = (A[1, j] + B[1, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[2, j] = (A[2, j] + B[2, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[3, j] = (A[3, j] + B[3, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[4, j] = (A[4, j] + B[4, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[5, j] = (A[5, j] + B[5, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[6, j] = (A[6, j] + B[6, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[7, j] = (A[7, j] + B[7, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[8, j] = (A[8, j] + B[8, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[9, j] = (A[9, j] + B[9, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[10, j] = (A[10, j] + B[10, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[11, j] = (A[11, j] + B[11, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[12, j] = (A[12, j] + B[12, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[13, j] = (A[13, j] + B[13, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[14, j] = (A[14, j] + B[14, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[15, j] = (A[15, j] + B[15, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[16, j] = (A[16, j] + B[16, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[17, j] = (A[17, j] + B[17, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[18, j] = (A[18, j] + B[18, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[19, j] = (A[19, j] + B[19, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[20, j] = (A[20, j] + B[20, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[21, j] = (A[21, j] + B[21, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[22, j] = (A[22, j] + B[22, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[23, j] = (A[23, j] + B[23, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[24, j] = (A[24, j] + B[24, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[25, j] = (A[25, j] + B[25, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[26, j] = (A[26, j] + B[26, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[27, j] = (A[27, j] + B[27, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[28, j] = (A[28, j] + B[28, j])
+8:     }
+8:   }
+8:   {
+8:     for (j, 40)
+8:     {
+8:       C[29, j] = (A[29, j] + B[29, j])
+8:     }
+8:   }
+8: }
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+8: I0924 13:32:18.803447 27796 stage_test.cc:466] 0.52927
+8: I0924 13:32:18.803457 27796 stage_test.cc:466] 1.21315
+8: I0924 13:32:18.803459 27796 stage_test.cc:466] 1.52707
+8: I0924 13:32:18.803462 27796 stage_test.cc:466] 0.940553
+8: [       OK ] Unroll.jit_precision_test1 (55 ms)
+8: [----------] 2 tests from Unroll (120 ms total)
+8: 
+8: [----------] 2 tests from ComputeInline
+8: [ RUN      ] ComputeInline.basic
+8: I0924 13:32:18.811084 27796 stage_test.cc:531] fn:
+8: function fn (_A, _C)
+8: {
+8:   for (i, 100)
+8:   {
+8:     for (j, 200)
+8:     {
+8:       C[i, j] = (6 + (2 * A[i, j]))
+8:     }
+8:   }
+8: }
+8: [       OK ] ComputeInline.basic (8 ms)
+8: [ RUN      ] ComputeInline.complex_graph
+8: I0924 13:32:18.828562 27796 stage_test.cc:575] fn:
+8: function fn (_A, _C, _C1, _C2)
+8: {
+8:   for (i, 100)
+8:   {
+8:     for (j, 200)
+8:     {
+8:       C2[i, j] = (6 + (2 * A[i, j]))
+8:     }
+8:   }
+8:   for (i, 100)
+8:   {
+8:     for (j, 200)
+8:     {
+8:       C1[i, j] = (4 + (2 * A[i, j]))
+8:     }
+8:   }
+8:   for (i, 100)
+8:   {
+8:     for (j, 200)
+8:     {
+8:       C[i, j] = (2 + (2 * A[i, j]))
+8:     }
+8:   }
+8: }
+8: [       OK ] ComputeInline.complex_graph (17 ms)
+8: [----------] 2 tests from ComputeInline (25 ms total)
+8: 
+8: [----------] 1 test from ShareBufferWith
+8: [ RUN      ] ShareBufferWith.basic
+8: I0924 13:32:18.839054 27796 stage_test.cc:622] fn:
+8: function fn (_A, _B)
+8: {
+8:   for (i, 100)
+8:   {
+8:     for (j, 200)
+8:     {
+8:       B[i, j] = (1 + A[i, j])
+8:     }
+8:   }
+8:   for (i, 100)
+8:   {
+8:     for (j, 200)
+8:     {
+8:       B1[i, j] = (1 + B[i, j])
+8:     }
+8:   }
+8: }
+8: I0924 13:32:18.839084 27796 stage_test.cc:630] 
+8: #include <cinn_runtime.h>
+8: #include <stdio.h>
+8: 
+8: void fn(void* _args, int32_t num_args)
+8: {
+8:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+8:   cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+8:   cinn_buffer_malloc((void*)(0), _B);
+8:   const float* A = ((const float*)(_A->memory));
+8:   float* B = ((float*)(_B->memory));
+8:   float* B1 = ((float*)(_B->memory));
+8:   for (int32_t i = 0; i < 100; i += 1) {
+8:     for (int32_t j = 0; j < 200; j += 1) {
+8:       B[((200 * i) + j)] = (1 + A[((200 * i) + j)]);
+8:     };
+8:   };
+8:   for (int32_t i = 0; i < 100; i += 1) {
+8:     for (int32_t j = 0; j < 200; j += 1) {
+8:       B1[((200 * i) + j)] = (1 + B[((200 * i) + j)]);
+8:     };
+8:   };
+8:   cinn_buffer_free((void*)(0), _B);
+8: }
+8: 
+8: [       OK ] ShareBufferWith.basic (12 ms)
+8: [----------] 1 test from ShareBufferWith (12 ms total)
+8: 
+8: [----------] 2 tests from isl
+8: [ RUN      ] isl.test
+8: I0924 13:32:18.840884 27796 stage_test.cc:641] schedule_intersected: [p0, p1] -> { p[i, j] -> p[t0, t1, t2 = j] : p0 = 0 and 2t1 = i and (t0) mod 2 = 0 and 0 <= p1 <= 2 and 4p1 <= i <= 1 + 4p1 and 0 <= j <= 9 + 4p1 - i and 0 <= t0 <= 1 }
+8: I0924 13:32:18.841017 27796 stage_test.cc:644] space: [p0, p1] -> {  :  }
+8: I0924 13:32:18.843125 27796 stage_test.cc:648] code:
+8: if (p0 == 0 && p1 >= 0 && p1 <= 2)
+8:   for (int c2 = 0; c2 <= 9; c2 += 1)
+8:     p(4 * p1, c2);
+8: [       OK ] isl.test (3 ms)
+8: [ RUN      ] isl.test1
+8: I0924 13:32:18.860146 27796 stage_test.cc:687] code:
+8: 
+8: for (int c1 = 0; c1 <= 2; c1 += 1)
+8:   for (int c2 = 0; c2 <= 9; c2 += 1)
+8:     for (int c3 = 0; c3 <= 9; c3 += 1) {
+8:       if (p0 == 0 && p1 <= 2 && c2 == 0 && c3 == 2 * p1)
+8:         for (int c5 = 0; c5 <= 9; c5 += 1)
+8:           p(4 * p1, c5);
+8:       c(4 * c1, c2, c3);
+8:     }
+8: [       OK ] isl.test1 (17 ms)
+8: [----------] 2 tests from isl (20 ms total)
+8: 
+8: [----------] Global test environment tear-down
+8: [==========] 23 tests from 10 test cases ran. (508 ms total)
+8: [  PASSED  ] 23 tests.
+18: [==========] Running 0 tests from 0 test cases.
+18: [==========] 0 tests from 0 test cases ran. (0 ms total)
+18: [  PASSED  ] 0 tests.
+17: [==========] Running 1 test from 1 test case.
+17: [----------] Global test environment set-up.
+17: [----------] 1 test from CodeGenCX86
+17: [ RUN      ] CodeGenCX86.basic
+17: before optim
+17: {
+17:   for (i, 100)
+17:   {
+17:     for (j, 500)
+17:     {
+17:       D[i, j] = (A[i, j] * B[i, j])
+17:     }
+17:   }
+17:   for (i, 100)
+17:   {
+17:     {
+17:       C[Broadcast(i,16), Ramp(0,1,16)] = (A[i, Ramp(0,1,16)] * B[i, Ramp(0,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(16,1,16)] = (A[i, Ramp(16,1,16)] * B[i, Ramp(16,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(32,1,16)] = (A[i, Ramp(32,1,16)] * B[i, Ramp(32,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(48,1,16)] = (A[i, Ramp(48,1,16)] * B[i, Ramp(48,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(64,1,16)] = (A[i, Ramp(64,1,16)] * B[i, Ramp(64,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(80,1,16)] = (A[i, Ramp(80,1,16)] * B[i, Ramp(80,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(96,1,16)] = (A[i, Ramp(96,1,16)] * B[i, Ramp(96,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(112,1,16)] = (A[i, Ramp(112,1,16)] * B[i, Ramp(112,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(128,1,16)] = (A[i, Ramp(128,1,16)] * B[i, Ramp(128,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(144,1,16)] = (A[i, Ramp(144,1,16)] * B[i, Ramp(144,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(160,1,16)] = (A[i, Ramp(160,1,16)] * B[i, Ramp(160,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(176,1,16)] = (A[i, Ramp(176,1,16)] * B[i, Ramp(176,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(192,1,16)] = (A[i, Ramp(192,1,16)] * B[i, Ramp(192,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(208,1,16)] = (A[i, Ramp(208,1,16)] * B[i, Ramp(208,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(224,1,16)] = (A[i, Ramp(224,1,16)] * B[i, Ramp(224,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(240,1,16)] = (A[i, Ramp(240,1,16)] * B[i, Ramp(240,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(256,1,16)] = (A[i, Ramp(256,1,16)] * B[i, Ramp(256,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(272,1,16)] = (A[i, Ramp(272,1,16)] * B[i, Ramp(272,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(288,1,16)] = (A[i, Ramp(288,1,16)] * B[i, Ramp(288,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(304,1,16)] = (A[i, Ramp(304,1,16)] * B[i, Ramp(304,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(320,1,16)] = (A[i, Ramp(320,1,16)] * B[i, Ramp(320,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(336,1,16)] = (A[i, Ramp(336,1,16)] * B[i, Ramp(336,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(352,1,16)] = (A[i, Ramp(352,1,16)] * B[i, Ramp(352,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(368,1,16)] = (A[i, Ramp(368,1,16)] * B[i, Ramp(368,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(384,1,16)] = (A[i, Ramp(384,1,16)] * B[i, Ramp(384,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(400,1,16)] = (A[i, Ramp(400,1,16)] * B[i, Ramp(400,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(416,1,16)] = (A[i, Ramp(416,1,16)] * B[i, Ramp(416,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(432,1,16)] = (A[i, Ramp(432,1,16)] * B[i, Ramp(432,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(448,1,16)] = (A[i, Ramp(448,1,16)] * B[i, Ramp(448,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(464,1,16)] = (A[i, Ramp(464,1,16)] * B[i, Ramp(464,1,16)])
+17:     }
+17:     {
+17:       C[Broadcast(i,16), Ramp(480,1,16)] = (A[i, Ramp(480,1,16)] * B[i, Ramp(480,1,16)])
+17:     }
+17:   }
+17: }
+17: out:
+17: #include <cinn_runtime.h>
+17: #include <stdio.h>
+17: 
+17: void matmul(void* _args, int32_t num_args)
+17: {
+17:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+17:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+17:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+17:   cinn_buffer_t* _D = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+17:   cinn_buffer_malloc((void*)(0), _C);
+17:   cinn_buffer_malloc((void*)(0), _D);
+17:   const float* A = ((const float*)(_A->memory));
+17:   const float* B = ((const float*)(_B->memory));
+17:   float* C = ((float*)(_C->memory));
+17:   float* D = ((float*)(_D->memory));
+17:   for (int32_t i = 0; i < 100; i += 1) {
+17:     for (int32_t j = 0; j < 500; j += 1) {
+17:       D[((500 * i) + j)] = (A[((500 * i) + j)] * B[((500 * i) + j)]);
+17:     };
+17:   };
+17:   for (int32_t i = 0; i < 100; i += 1) {
+17:     cinn_avx512_store(C + (500 * i), cinn_avx512_mul(cinn_avx512_load(A + (500 * i)), cinn_avx512_load(B + (500 * i))));
+17:     cinn_avx512_store(C + (16 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (16 + (500 * i))), cinn_avx512_load(B + (16 + (500 * i)))));
+17:     cinn_avx512_store(C + (32 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (32 + (500 * i))), cinn_avx512_load(B + (32 + (500 * i)))));
+17:     cinn_avx512_store(C + (48 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (48 + (500 * i))), cinn_avx512_load(B + (48 + (500 * i)))));
+17:     cinn_avx512_store(C + (64 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (64 + (500 * i))), cinn_avx512_load(B + (64 + (500 * i)))));
+17:     cinn_avx512_store(C + (80 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (80 + (500 * i))), cinn_avx512_load(B + (80 + (500 * i)))));
+17:     cinn_avx512_store(C + (96 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (96 + (500 * i))), cinn_avx512_load(B + (96 + (500 * i)))));
+17:     cinn_avx512_store(C + (112 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (112 + (500 * i))), cinn_avx512_load(B + (112 + (500 * i)))));
+17:     cinn_avx512_store(C + (128 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (128 + (500 * i))), cinn_avx512_load(B + (128 + (500 * i)))));
+17:     cinn_avx512_store(C + (144 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (144 + (500 * i))), cinn_avx512_load(B + (144 + (500 * i)))));
+17:     cinn_avx512_store(C + (160 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (160 + (500 * i))), cinn_avx512_load(B + (160 + (500 * i)))));
+17:     cinn_avx512_store(C + (176 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (176 + (500 * i))), cinn_avx512_load(B + (176 + (500 * i)))));
+17:     cinn_avx512_store(C + (192 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (192 + (500 * i))), cinn_avx512_load(B + (192 + (500 * i)))));
+17:     cinn_avx512_store(C + (208 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (208 + (500 * i))), cinn_avx512_load(B + (208 + (500 * i)))));
+17:     cinn_avx512_store(C + (224 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (224 + (500 * i))), cinn_avx512_load(B + (224 + (500 * i)))));
+17:     cinn_avx512_store(C + (240 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (240 + (500 * i))), cinn_avx512_load(B + (240 + (500 * i)))));
+17:     cinn_avx512_store(C + (256 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (256 + (500 * i))), cinn_avx512_load(B + (256 + (500 * i)))));
+17:     cinn_avx512_store(C + (272 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (272 + (500 * i))), cinn_avx512_load(B + (272 + (500 * i)))));
+17:     cinn_avx512_store(C + (288 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (288 + (500 * i))), cinn_avx512_load(B + (288 + (500 * i)))));
+17:     cinn_avx512_store(C + (304 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (304 + (500 * i))), cinn_avx512_load(B + (304 + (500 * i)))));
+17:     cinn_avx512_store(C + (320 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (320 + (500 * i))), cinn_avx512_load(B + (320 + (500 * i)))));
+17:     cinn_avx512_store(C + (336 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (336 + (500 * i))), cinn_avx512_load(B + (336 + (500 * i)))));
+17:     cinn_avx512_store(C + (352 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (352 + (500 * i))), cinn_avx512_load(B + (352 + (500 * i)))));
+17:     cinn_avx512_store(C + (368 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (368 + (500 * i))), cinn_avx512_load(B + (368 + (500 * i)))));
+17:     cinn_avx512_store(C + (384 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (384 + (500 * i))), cinn_avx512_load(B + (384 + (500 * i)))));
+17:     cinn_avx512_store(C + (400 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (400 + (500 * i))), cinn_avx512_load(B + (400 + (500 * i)))));
+17:     cinn_avx512_store(C + (416 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (416 + (500 * i))), cinn_avx512_load(B + (416 + (500 * i)))));
+17:     cinn_avx512_store(C + (432 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (432 + (500 * i))), cinn_avx512_load(B + (432 + (500 * i)))));
+17:     cinn_avx512_store(C + (448 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (448 + (500 * i))), cinn_avx512_load(B + (448 + (500 * i)))));
+17:     cinn_avx512_store(C + (464 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (464 + (500 * i))), cinn_avx512_load(B + (464 + (500 * i)))));
+17:     cinn_avx512_store(C + (480 + (500 * i)), cinn_avx512_mul(cinn_avx512_load(A + (480 + (500 * i))), cinn_avx512_load(B + (480 + (500 * i)))));
+17:   };
+17:   cinn_buffer_free((void*)(0), _C);
+17:   cinn_buffer_free((void*)(0), _D);
+17: }
+17: 
+17: [       OK ] CodeGenCX86.basic (80 ms)
+17: [----------] 1 test from CodeGenCX86 (80 ms total)
+17: 
+17: [----------] Global test environment tear-down
+17: [==========] 1 test from 1 test case ran. (80 ms total)
+17: [  PASSED  ] 1 test.
+23: [==========] Running 1 test from 1 test case.
+23: [----------] Global test environment set-up.
+23: [----------] 1 test from Call
+23: [ RUN      ] Call.basic
+23: [       OK ] Call.basic (0 ms)
+23: [----------] 1 test from Call (0 ms total)
+23: 
+23: [----------] Global test environment tear-down
+23: [==========] 1 test from 1 test case ran. (0 ms total)
+23: [  PASSED  ] 1 test.
+17/68 Test #53: test02_matmul_main .....................   Passed    0.53 sec
+18/68 Test #22: test_codegen_x86 .......................   Passed    0.08 sec
+19/68 Test #21: test_execution_engine ..................   Passed    0.09 sec
+20/68 Test #20: test_codegen_llvm ......................   Passed    0.19 sec
+21/68 Test #19: test_compiler ..........................   Passed    0.19 sec
+22/68 Test  #8: test_stage .............................   Passed    0.53 sec
+23/68 Test #18: test_generated1 ........................   Passed    0.19 sec
+24/68 Test #17: test_codegen_c_x86 .....................   Passed    0.19 sec
+25/68 Test #23: test_compute ...........................   Passed    0.19 sec
+test 24
+      Start 24: test_placeholder
+
+24: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/lang/test_placeholder ""
+24: Test timeout computed to be: 600
+test 25
+      Start 25: test_lower
+
+25: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/lang/test_lower ""
+25: Test timeout computed to be: 600
+test 26
+      Start 26: test_lower_impl
+
+26: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/lang/test_lower_impl ""
+26: Test timeout computed to be: 600
+test 27
+      Start 27: test_packed_func
+
+27: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/lang/test_packed_func ""
+27: Test timeout computed to be: 600
+test 28
+      Start 28: test_remove_nested_block
+
+28: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/optim/test_remove_nested_block ""
+28: Test timeout computed to be: 600
+test 29
+      Start 29: test_ir_copy
+
+29: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/optim/test_ir_copy ""
+29: Test timeout computed to be: 600
+test 30
+      Start 30: test_ir_simplify
+
+30: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/optim/test_ir_simplify ""
+30: Test timeout computed to be: 600
+test 31
+      Start 31: test_replace_call_with_expr
+
+31: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/optim/test_replace_call_with_expr ""
+31: Test timeout computed to be: 600
+test 32
+      Start 32: test_vectorize_loops
+
+32: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/optim/test_vectorize_loops "--cinn_x86_builtin_code_root=/home/wangyue50/CINN-my/CINN/cinn/backends"
+32: Test timeout computed to be: 600
+31: [==========] Running 0 tests from 0 test cases.
+31: [==========] 0 tests from 0 test cases ran. (0 ms total)
+31: [  PASSED  ] 0 tests.
+30: [==========] Running 2 tests from 2 test cases.
+30: [----------] Global test environment set-up.
+30: [----------] 1 test from IrSimplify
+30: [ RUN      ] IrSimplify.basic
+30: WARNING: Logging before InitGoogleLogging() is written to STDERR
+30: I0924 13:32:18.883857 27818 ir_simplify_test.cc:26] B (((C[i, 0] + 0) + 100) + 24.5)
+30: I0924 13:32:18.884227 27818 ir_simplify_test.cc:29] simplified: (124.5 + C[i, 0])
+30: I0924 13:32:18.892341 27818 ir_simplify_test.cc:50] original body:
+30: {
+30:   for (i, 100)
+30:   {
+30:     for (j, 20)
+30:     {
+30:       B[i, j] = (125 + (X[i, j] + y[i, 0]))
+30:     }
+30:   }
+30: }
+30: I0924 13:32:18.900732 27818 ir_simplify_test.cc:83] original body:
+30: {
+30:   for (i, 100)
+30:   {
+30:     for (j, 20)
+30:     {
+30:       B[i, j] = (125 + (X[(1000 * i), 0] + (0.333333 * y[i, 0])))
+30:     }
+30:   }
+30: }
+30: [       OK ] IrSimplify.basic (18 ms)
+30: [----------] 1 test from IrSimplify (18 ms total)
+30: 
+30: [----------] 1 test from reverse
+30: [ RUN      ] reverse.prod
+30: I0924 13:32:18.906757 27818 ir_simplify_test.cc:108] fn:
+30: function fn (_A, _tensor)
+30: {
+30:   for (i, 100)
+30:   {
+30:     for (j, 20)
+30:     {
+30:       tensor[i, j] = (1 / A[i, j])
+30:     }
+30:   }
+30: }
+30: [       OK ] reverse.prod (5 ms)
+30: [----------] 1 test from reverse (5 ms total)
+30: 
+30: [----------] Global test environment tear-down
+30: [==========] 2 tests from 2 test cases ran. (23 ms total)
+30: [  PASSED  ] 2 tests.
+29: [==========] Running 1 test from 1 test case.
+29: [----------] Global test environment set-up.
+29: [----------] 1 test from IrCopy
+29: [ RUN      ] IrCopy.basic
+29: WARNING: Logging before InitGoogleLogging() is written to STDERR
+29: I0924 13:32:18.883689 27817 ir_copy_test.cc:13] aa 1
+29: [       OK ] IrCopy.basic (0 ms)
+29: [----------] 1 test from IrCopy (0 ms total)
+29: 
+29: [----------] Global test environment tear-down
+29: [==========] 1 test from 1 test case ran. (0 ms total)
+29: [  PASSED  ] 1 test.
+28: [==========] Running 1 test from 1 test case.
+28: [----------] Global test environment set-up.
+28: [----------] 1 test from RemoveNestedBlock
+28: [ RUN      ] RemoveNestedBlock.basic
+28: origin:
+28: {
+28:   {
+28:     1
+28:     1
+28:   }
+28: }
+28: e:
+28: {
+28:   1
+28:   1
+28: }
+28: [       OK ] RemoveNestedBlock.basic (0 ms)
+28: [----------] 1 test from RemoveNestedBlock (0 ms total)
+28: 
+28: [----------] Global test environment tear-down
+28: [==========] 1 test from 1 test case ran. (0 ms total)
+28: [  PASSED  ] 1 test.
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.asin (29 ms)
+11: [ RUN      ] mkl_math.asinh
+11: I0924 13:32:18.885082 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_asinh_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_asinh_fp32_out[i, j] = cinn_cpu_asinh_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.asinh (30 ms)
+11: [ RUN      ] mkl_math.atan
+11: I0924 13:32:18.914476 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_atan_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_atan_fp32_out[i, j] = cinn_cpu_atan_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.atan (28 ms)
+11: [ RUN      ] mkl_math.atanh
+11: I0924 13:32:18.943289 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_atanh_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_atanh_fp32_out[i, j] = cinn_cpu_atanh_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.atanh (29 ms)
+11: [ RUN      ] mkl_math.isnan
+11: I0924 13:32:18.971868 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_isnan_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_isnan_fp32_out[i, j] = cinn_cpu_isnan_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.isnan (29 ms)
+11: [ RUN      ] mkl_math.tanh
+11: I0924 13:32:19.000425 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_tanh_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_tanh_fp32_out[i, j] = cinn_cpu_tanh_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+27: [==========] Running 4 tests from 1 test case.
+27: [----------] Global test environment set-up.
+27: [----------] 4 tests from Function
+27: [ RUN      ] Function.test
+27: WARNING: Logging before InitGoogleLogging() is written to STDERR
+27: I0924 13:32:18.881062 27815 packed_func_test.cc:21] c 3
+27: [       OK ] Function.test (0 ms)
+27: [ RUN      ] Function.test1
+27: I0924 13:32:18.881088 27815 packed_func_test.cc:33] hello world
+27: [       OK ] Function.test1 (0 ms)
+27: [ RUN      ] Function.Expr
+27: [       OK ] Function.Expr (0 ms)
+27: [ RUN      ] Function.ReturnMultiValue
+27: [       OK ] Function.ReturnMultiValue (0 ms)
+27: [----------] 4 tests from Function (0 ms total)
+27: 
+27: [----------] Global test environment tear-down
+27: [==========] 4 tests from 1 test case ran. (0 ms total)
+27: [  PASSED  ] 4 tests.
+26: [==========] Running 5 tests from 1 test case.
+26: [----------] Global test environment set-up.
+26: [----------] 5 tests from CreateCompGraph
+26: [ RUN      ] CreateCompGraph.single_layer
+26: WARNING: Logging before InitGoogleLogging() is written to STDERR
+26: I0924 13:32:18.881656 27814 lower_impl_test.cc:25] 5
+26: I0924 13:32:18.881683 27814 lower_impl_test.cc:27] e: (A[i, j] + B[i, j])
+26: I0924 13:32:18.881696 27814 lower_impl_test.cc:27] e: 100
+26: I0924 13:32:18.881700 27814 lower_impl_test.cc:27] e: 200
+26: I0924 13:32:18.881702 27814 lower_impl_test.cc:27] e: 100
+26: I0924 13:32:18.881705 27814 lower_impl_test.cc:27] e: 200
+26: I0924 13:32:18.882537 27814 lower_impl_test.cc:33] graph:
+26: digraph G {
+26:    node_0[label="A"]
+26:    node_1[label="B"]
+26:    node_2[label="C"]
+26:    node_0->node_2
+26:    node_1->node_2
+26: } // end G
+26: [       OK ] CreateCompGraph.single_layer (1 ms)
+26: [ RUN      ] CreateCompGraph.multi_layers
+26: I0924 13:32:18.883672 27814 lower_impl_test.cc:80] graph:
+26: digraph G {
+26:    node_3[label="A"]
+26:    node_4[label="B"]
+26:    node_6[label="C"]
+26:    node_7[label="D"]
+26:    node_5[label="E"]
+26:    node_3->node_5
+26:    node_3->node_6
+26:    node_4->node_5
+26:    node_4->node_7
+26:    node_4->node_6
+26:    node_6->node_5
+26:    node_6->node_7
+26:    node_7->node_5
+26: } // end G
+26: [       OK ] CreateCompGraph.multi_layers (1 ms)
+26: [ RUN      ] CreateCompGraph.multi_layers_with_extra_deps
+26: I0924 13:32:18.885052 27814 lower_impl_test.cc:150] graph:
+26: digraph G {
+26:    node_8[label="A"]
+26:    node_9[label="B"]
+26:    node_11[label="C"]
+26:    node_12[label="D"]
+26:    node_13[label="E"]
+26:    node_10[label="F"]
+26:    node_8->node_13
+26:    node_8->node_11
+26:    node_9->node_12
+26:    node_11->node_10
+26:    node_11->node_13
+26:    node_11->node_12
+26:    node_12->node_10
+26:    node_13->node_10
+26: } // end G
+26: [       OK ] CreateCompGraph.multi_layers_with_extra_deps (2 ms)
+26: [ RUN      ] CreateCompGraph.inline_compatible
+26: I0924 13:32:18.886207 27814 lower_impl_test.cc:216] graph:
+26: digraph G {
+26:    node_14[label="A"]
+26:    node_15[label="B"]
+26:    node_17[label="C"]
+26:    node_16[label="E"]
+26:    node_14->node_16
+26:    node_14->node_17
+26:    node_15->node_16
+26:    node_15->node_17
+26:    node_17->node_16
+26: } // end G
+26: [       OK ] CreateCompGraph.inline_compatible (1 ms)
+26: [ RUN      ] CreateCompGraph.inline_compatible1
+26: I0924 13:32:18.887346 27814 lower_impl_test.cc:275] graph:
+26: digraph G {
+26:    node_18[label="A"]
+26:    node_19[label="B"]
+26:    node_21[label="D"]
+26:    node_20[label="E"]
+26:    node_18->node_20
+26:    node_18->node_21
+26:    node_19->node_20
+26:    node_19->node_21
+26:    node_21->node_20
+26: } // end G
+26: [       OK ] CreateCompGraph.inline_compatible1 (1 ms)
+26: [----------] 5 tests from CreateCompGraph (6 ms total)
+26: 
+26: [----------] Global test environment tear-down
+26: [==========] 5 tests from 1 test case ran. (6 ms total)
+26: [  PASSED  ] 5 tests.
+25: [==========] Running 5 tests from 1 test case.
+25: [----------] Global test environment set-up.
+25: [----------] 5 tests from lower
+25: [ RUN      ] lower.basic
+25: WARNING: Logging before InitGoogleLogging() is written to STDERR
+25: I0924 13:32:18.889024 27813 lower_test.cc:27] lower_size function cal_B (_A, _B)
+25: {
+25:   for (i, 100)
+25:   {
+25:     for (j, 15)
+25:     {
+25:       B[i, j] = (1 + A[i, j])
+25:     }
+25:   }
+25: }
+25: 
+25: {
+25:   for (i, 100)
+25:   {
+25:     for (j, 15)
+25:     {
+25:       B[i, j] = (1 + A[i, j])
+25:     }
+25:   }
+25: }
+25: [       OK ] lower.basic (7 ms)
+25: [ RUN      ] lower.more_complex
+25: func:
+25: function cal_C (_A, _B, _C)
+25: {
+25:   for (i, 100)
+25:   {
+25:     for (j, 15)
+25:     {
+25:       for (k, 200)
+25:       {
+25:         C[i, j, k] = (A[i, j] * B[j, k])
+25:       }
+25:     }
+25:   }
+25: }
+25: [       OK ] lower.more_complex (9 ms)
+25: [ RUN      ] lower.dynamic_shape
+25: func:
+25: function cal_C (_X, _W, _C)
+25: {
+25:   for (i, int32(B))
+25:   {
+25:     for (j, 15)
+25:     {
+25:       for (k, 200)
+25:       {
+25:         C[i, j, k] = (X[i, j] * W[j, k])
+25:       }
+25:     }
+25:   }
+25: }
+25: [       OK ] lower.dynamic_shape (11 ms)
+25: [ RUN      ] lower.lowered_call
+25: I0924 13:32:18.910259 27813 lower_test.cc:101] call_op: lowered_fun0(Tensor(X, [B,15]), Tensor(Y, [B,15]), Tensor(Z, [B,15]), Tensor(C, [B,15]))
+25: [       OK ] lower.lowered_call (9 ms)
+25: [ RUN      ] lower.temp_buffer_collects
+25: [       OK ] lower.temp_buffer_collects (9 ms)
+25: [----------] 5 tests from lower (45 ms total)
+25: 
+25: [----------] Global test environment tear-down
+25: [==========] 5 tests from 1 test case ran. (45 ms total)
+25: [  PASSED  ] 5 tests.
+24: [==========] Running 2 tests from 1 test case.
+24: [----------] Global test environment set-up.
+24: [----------] 2 tests from placeholder
+24: [ RUN      ] placeholder.basic
+24: WARNING: Logging before InitGoogleLogging() is written to STDERR
+24: I0924 13:32:18.880643 27812 placeholder_test.cc:20] slice x[i, j]
+24: [       OK ] placeholder.basic (0 ms)
+24: [ RUN      ] placeholder.dynamic_shape
+24: [       OK ] placeholder.dynamic_shape (0 ms)
+24: [----------] 2 tests from placeholder (0 ms total)
+24: 
+24: [----------] Global test environment tear-down
+24: [==========] 2 tests from 1 test case ran. (0 ms total)
+24: [  PASSED  ] 2 tests.
+32: [==========] Running 4 tests from 1 test case.
+32: [----------] Global test environment set-up.
+32: [----------] 4 tests from Vectorize
+32: [ RUN      ] Vectorize.replace_var
+32: [       OK ] Vectorize.replace_var (11 ms)
+32: [ RUN      ] Vectorize.TestMarkVectorize
+32: before optim
+32: {
+32:   for (i, 100)
+32:   {
+32:     for (j, 500)
+32:     {
+32:       D[i, j] = (A[i, j] * B[i, j])
+32:     }
+32:   }
+32:   for (i, 100)
+32:   {
+32:     for (j, 31)
+32:     {
+32:       C[Broadcast(i,16), Ramp((16 * j),1,16)] = (A[i, Ramp((16 * j),1,16)] * B[i, Ramp((16 * j),1,16)])
+32:     }
+32:   }
+32: }
+32: out:
+32: #include <cinn_runtime.h>
+32: #include <stdio.h>
+32: 
+32: void matmul(void* _args, int32_t num_args)
+32: {
+32:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+32:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+32:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+32:   cinn_buffer_malloc((void*)(0), _C);
+32:   const float* A = ((const float*)(_A->memory));
+32:   const float* B = ((const float*)(_B->memory));
+32:   float* C = ((float*)(_C->memory));
+32:   float* D = ((float*)(_C->memory));
+32:   for (int32_t i = 0; i < 100; i += 1) {
+32:     for (int32_t j = 0; j < 500; j += 1) {
+32:       D[((500 * i) + j)] = (A[((500 * i) + j)] * B[((500 * i) + j)]);
+32:     };
+32:   };
+32:   for (int32_t i = 0; i < 100; i += 1) {
+32:     for (int32_t j = 0; j < 31; j += 1) {
+32:       C[StackVec<16,int32_t>::Ramp(((500 * i) + (16 * j)), 1, 16)] = (StackedVec<float,16>::Load(A,((500 * i) + (16 * j))) * StackedVec<float,16>::Load(B,((500 * i) + (16 * j))));
+32:     };
+32:   };
+32:   cinn_buffer_free((void*)(0), _C);
+32: }
+32: 
+32: [       OK ] Vectorize.TestMarkVectorize (16 ms)
+32: [ RUN      ] Vectorize.vectorize
+32: [       OK ] Vectorize.vectorize (0 ms)
+32: [ RUN      ] Vectorize.single_for
+32: WARNING: Logging before InitGoogleLogging() is written to STDERR
+32: I0924 13:32:18.913712 27820 vectorize_loops_test.cc:216] Forloop
+32: for (k0, 1)
+32: {
+32:   C[Ramp((16 * k0),1,16)] = (A[Ramp((16 * k0),1,16)] + B[Ramp((16 * k0),1,16)])
+32: }
+32: [       OK ] Vectorize.single_for (1 ms)
+32: [----------] 4 tests from Vectorize (28 ms total)
+32: 
+32: [----------] Global test environment tear-down
+32: [==========] 4 tests from 1 test case ran. (28 ms total)
+32: [  PASSED  ] 4 tests.
+26/68 Test #31: test_replace_call_with_expr ............   Passed    0.01 sec
+27/68 Test #30: test_ir_simplify .......................   Passed    0.04 sec
+28/68 Test #29: test_ir_copy ...........................   Passed    0.04 sec
+29/68 Test #28: test_remove_nested_block ...............   Passed    0.04 sec
+30/68 Test #27: test_packed_func .......................   Passed    0.14 sec
+31/68 Test #26: test_lower_impl ........................   Passed    0.14 sec
+32/68 Test #25: test_lower .............................   Passed    0.14 sec
+33/68 Test #24: test_placeholder .......................   Passed    0.14 sec
+34/68 Test #32: test_vectorize_loops ...................   Passed    0.14 sec
+test 33
+      Start 33: test_transform_polyfor_to_for
+
+33: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/optim/test_transform_polyfor_to_for "--cinn_x86_builtin_code_root=/home/wangyue50/CINN-my/CINN/cinn/backends"
+33: Test timeout computed to be: 600
+test 34
+      Start 34: test_optimize
+
+34: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/optim/test_optimize ""
+34: Test timeout computed to be: 600
+test 35
+      Start 35: test_cache_read_write_replace
+
+35: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/optim/test_cache_read_write_replace ""
+35: Test timeout computed to be: 600
+test 36
+      Start 36: test_hlir_framework_buffer
+
+36: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/framework/test_hlir_framework_buffer ""
+36: Test timeout computed to be: 600
+test 37
+      Start 37: test_hlir_framework_infershape_pass
+
+37: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/framework/test_hlir_framework_infershape_pass ""
+37: Test timeout computed to be: 600
+test 38
+      Start 38: test_hlir_framework_tensor
+
+38: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/framework/test_hlir_framework_tensor ""
+38: Test timeout computed to be: 600
+test 39
+      Start 39: test_hlir_framework_scope
+
+39: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/framework/test_hlir_framework_scope ""
+39: Test timeout computed to be: 600
+test 40
+      Start 40: test_hlir_framework_instruction
+
+40: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/framework/test_hlir_framework_instruction ""
+40: Test timeout computed to be: 600
+test 41
+      Start 41: test_hlir_framework_op
+
+41: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/framework/test_hlir_framework_op ""
+41: Test timeout computed to be: 600
+39: [==========] Running 1 test from 1 test case.
+39: [----------] Global test environment set-up.
+39: [----------] 1 test from Scope
+39: [ RUN      ] Scope.basic
+39: [       OK ] Scope.basic (0 ms)
+39: [----------] 1 test from Scope (0 ms total)
+39: 
+39: [----------] Global test environment tear-down
+39: [==========] 1 test from 1 test case ran. (0 ms total)
+39: [  PASSED  ] 1 test.
+38: [==========] Running 1 test from 1 test case.
+38: [----------] Global test environment set-up.
+38: [----------] 1 test from Tensor
+38: [ RUN      ] Tensor.basic
+38: [       OK ] Tensor.basic (0 ms)
+38: [----------] 1 test from Tensor (0 ms total)
+38: 
+38: [----------] Global test environment tear-down
+38: [==========] 1 test from 1 test case ran. (0 ms total)
+38: [  PASSED  ] 1 test.
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register relu
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register relu6
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register conv2d
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register depthwise_conv2d
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register batchnorm
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register pool1d
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register pool2d
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register pool3d
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register sigmoid
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register softmax
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register slice
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register depthwise_conv2d
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register elementwise_add
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register elementwise_mul
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register scale
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register matmul
+37: I0100 00:00:00.000000 27825 registry.h:89] RAW: Register mul
+37: [==========] Running 1 test from 1 test case.
+37: [----------] Global test environment set-up.
+37: [----------] 1 test from Operator
+37: [ RUN      ] Operator.GetAttrs
+37: WARNING: Logging before InitGoogleLogging() is written to STDERR
+37: I0924 13:32:19.052171 27825 graph_compiler.cc:39] [Debug] C Code is:
+37: #include <cinn_runtime.h>
+37: #include <stdio.h>
+37: 
+37: void fn_elementwise_add_0(void* _args, int32_t num_args)
+37: {
+37:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+37:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+37:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+37:   cinn_buffer_malloc((void*)(0), _C);
+37:   const float* A = ((const float*)(_A->memory));
+37:   const float* B = ((const float*)(_B->memory));
+37:   float* C = ((float*)(_C->memory));
+37:   for (int32_t i = 0; i < 100; i += 1) {
+37:     for (int32_t j = 0; j < 32; j += 1) {
+37:       C[((32 * i) + j)] = (A[((32 * i) + j)] + B[((32 * i) + j)]);
+37:     };
+37:   };
+37:   cinn_buffer_free((void*)(0), _C);
+37: }
+37: 
+37: void fn_elementwise_add_1(void* _args, int32_t num_args)
+37: {
+37:   const cinn_buffer_t* _var = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+37:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+37:   cinn_buffer_t* _C_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+37:   cinn_buffer_malloc((void*)(0), _C_0);
+37:   const float* B = ((const float*)(_B->memory));
+37:   float* C_0 = ((float*)(_C_0->memory));
+37:   const float* var = ((const float*)(_var->memory));
+37:   for (int32_t i = 0; i < 100; i += 1) {
+37:     for (int32_t j = 0; j < 32; j += 1) {
+37:       C_0[((32 * i) + j)] = (var[((32 * i) + j)] + B[((32 * i) + j)]);
+37:     };
+37:   };
+37:   cinn_buffer_free((void*)(0), _C_0);
+37: }
+37: 
+37: void fn_elementwise_add_2(void* _args, int32_t num_args)
+37: {
+37:   const cinn_buffer_t* _var = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+37:   const cinn_buffer_t* _var_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+37:   cinn_buffer_t* _C_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+37:   cinn_buffer_malloc((void*)(0), _C_1);
+37:   float* C_1 = ((float*)(_C_1->memory));
+37:   const float* var = ((const float*)(_var->memory));
+37:   const float* var_0 = ((const float*)(_var_0->memory));
+37:   for (int32_t i = 0; i < 100; i += 1) {
+37:     for (int32_t j = 0; j < 32; j += 1) {
+37:       C_1[((32 * i) + j)] = (var[((32 * i) + j)] + var_0[((32 * i) + j)]);
+37:     };
+37:   };
+37:   cinn_buffer_free((void*)(0), _C_1);
+37: }
+37: 
+37: I0924 13:32:19.080960 27825 infershape_pass_test.cc:68] data: 1.5662 + 1.20472 = 2.77092
+37: I0924 13:32:19.080993 27825 infershape_pass_test.cc:68] data: 1.59688 + 1.46827 = 3.06515
+37: I0924 13:32:19.080997 27825 infershape_pass_test.cc:68] data: 1.82329 + 1.40903 = 3.23232
+37: [       OK ] Operator.GetAttrs (53 ms)
+37: [----------] 1 test from Operator (53 ms total)
+37: 
+37: [----------] Global test environment tear-down
+37: [==========] 1 test from 1 test case ran. (53 ms total)
+37: [  PASSED  ] 1 test.
+36: [==========] Running 1 test from 1 test case.
+36: [----------] Global test environment set-up.
+36: [----------] 1 test from Buffer
+36: [ RUN      ] Buffer.basic
+36: [       OK ] Buffer.basic (0 ms)
+36: [----------] 1 test from Buffer (0 ms total)
+36: 
+36: [----------] Global test environment tear-down
+36: [==========] 1 test from 1 test case ran. (0 ms total)
+36: [  PASSED  ] 1 test.
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.tanh (29 ms)
+11: [ RUN      ] mkl_math.isfinite
+11: I0924 13:32:19.029877 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_isfinite_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_isfinite_fp32_out[i, j] = cinn_cpu_isfinite_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.isfinite (30 ms)
+11: [ RUN      ] mkl_math.isinf
+11: I0924 13:32:19.059433 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _cinn_cpu_isinf_fp32_out)
+11: {
+11:   for (i, 10)
+11:   {
+11:     for (j, 10)
+11:     {
+11:       cinn_cpu_isinf_fp32_out[i, j] = cinn_cpu_isinf_fp32(x[i, j])
+11:     }
+11:   }
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.isinf (28 ms)
+11: [ RUN      ] mkl_math.tanh_v_fp32
+11: E0924 13:32:19.084836 27799 lower_impl.cc:283] tensor [cinn_mkl_tanh_v_fp32_out] buffer is null
+11: I0924 13:32:19.085230 27799 mkl_math_test.cc:59] func:
+11: function fn (_x, _tuple_cinn_mkl_tanh_v_fp32_out0__0)
+11: {
+11:   cinn_mkl_tanh_v_fp32(Tensor(x, [10,10]), Tensor(tuple_cinn_mkl_tanh_v_fp32_out0__0, [10,10]))
+11: }
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 400
+11: [       OK ] mkl_math.tanh_v_fp32 (58 ms)
+11: [----------] 26 tests from mkl_math (783 ms total)
+11: 
+11: [----------] 1 test from cinn_cpu_mkl_gemm_fp32
+11: [ RUN      ] cinn_cpu_mkl_gemm_fp32.test
+11: E0924 13:32:19.142848 27799 lower_impl.cc:283] tensor [extern_call] buffer is null
+11: I0924 13:32:19.143520 27799 mkl_math_test.cc:166] func:
+11: function fn (_A, _B, _tuple_cinn_cpu_mkl_gemm_fp32_out0__0)
+11: {
+11:   cinn_cpu_mkl_gemm_fp32(1, 30, 20, 40, 0, 0, 30, 40, 30, 0, Tensor(A, [30,40]), Tensor(B, [40,20]), Tensor(tuple_cinn_cpu_mkl_gemm_fp32_out0__0, [30,20]))
+11: }
+35: [==========] Running 2 tests from 1 test case.
+35: [----------] Global test environment set-up.
+35: [----------] 2 tests from CacheReadWriteReplace
+35: [ RUN      ] CacheReadWriteReplace.basic
+35: WARNING: Logging before InitGoogleLogging() is written to STDERR
+35: I0924 13:32:19.045128 27823 cache_read_write_replace_test.cc:30] fn:
+35: function fn (_A, _B, _C)
+35: {
+35:   for (i, 100)
+35:   {
+35:     for (j, 20)
+35:     {
+35:       A_read_cache[i, j] = A[i, j]
+35:     }
+35:   }
+35:   __syncthreads()
+35:   for (i, 100)
+35:   {
+35:     for (j, 20)
+35:     {
+35:       C[i, j] = (A_read_cache[i, j] + B[i, j])
+35:     }
+35:   }
+35:   for (i, 100)
+35:   {
+35:     for (j, 20)
+35:     {
+35:       C_cache_write_out[i, j] = C[i, j]
+35:     }
+35:   }
+35: }
+35: [       OK ] CacheReadWriteReplace.basic (18 ms)
+35: [ RUN      ] CacheReadWriteReplace.cache_write
+35: I0924 13:32:19.056802 27823 cache_read_write_replace_test.cc:87] 
+35: function fn (_A, _B, _C1_cache_write_out)
+35: {
+35:   for (i, 100)
+35:   {
+35:     for (j, 100)
+35:     {
+35:       C1[i, j] = (3 + A[i, j])
+35:     }
+35:   }
+35:   for (i, 100)
+35:   {
+35:     for (j, 100)
+35:     {
+35:       C1_cache_write_out[i, j] = C1[i, j]
+35:     }
+35:   }
+35: }
+35: [       OK ] CacheReadWriteReplace.cache_write (11 ms)
+35: [----------] 2 tests from CacheReadWriteReplace (29 ms total)
+35: 
+35: [----------] Global test environment tear-down
+35: [==========] 2 tests from 1 test case ran. (29 ms total)
+35: [  PASSED  ] 2 tests.
+34: [==========] Running 1 test from 1 test case.
+34: [----------] Global test environment set-up.
+34: [----------] 1 test from Optimize
+34: [ RUN      ] Optimize.Unroll
+34: [       OK ] Optimize.Unroll (16 ms)
+34: [----------] 1 test from Optimize (16 ms total)
+34: 
+34: [----------] Global test environment tear-down
+34: [==========] 1 test from 1 test case ran. (16 ms total)
+34: [  PASSED  ] 1 test.
+33: [==========] Running 1 test from 1 test case.
+33: [----------] Global test environment set-up.
+33: [----------] 1 test from Expr
+33: [ RUN      ] Expr.basic
+33: out:
+33: #include <cinn_runtime.h>
+33: #include <stdio.h>
+33: 
+33: void matmul(void* _args, int32_t num_args)
+33: {
+33:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+33:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+33:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+33:   cinn_buffer_malloc((void*)(0), _C);
+33:   const float* A = ((const float*)(_A->memory));
+33:   const float* B = ((const float*)(_B->memory));
+33:   float* C = ((float*)(_C->memory));
+33:   for (int32_t i_outer = 0; i_outer < 64; i_outer += 1) {
+33:     for (int32_t i_inner = 0; i_inner < 8; i_inner += 1) {
+33:       for (int32_t j_outer = 0; j_outer < 63; j_outer += 1) {
+33:         for (int32_t j_inner = 0; j_inner < (1 + ((int32_t)(cinn_min(7, (499 + (-8 * j_outer)))))); j_inner += 1) {
+33:           for (int32_t k0 = 0; k0 < 200; k0 += 1) {
+33:             C[((500 * i_inner) + ((4000 * i_outer) + ((8 * j_outer) + j_inner)))] = (C[((500 * i_inner) + ((4000 * i_outer) + ((8 * j_outer) + j_inner)))] + (A[((200 * i_inner) + ((1600 * i_outer) + k0))] * B[((8 * j_outer) + ((500 * k0) + j_inner))]));
+33:           };
+33:         };
+33:       };
+33:     };
+33:   };
+33:   cinn_buffer_free((void*)(0), _C);
+33: }
+33: 
+33: out:
+33: #include <cinn_runtime.h>
+33: #include <stdio.h>
+33: 
+33: void matmul(void* _args, int32_t num_args)
+33: {
+33:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+33:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+33:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+33:   cinn_buffer_malloc((void*)(0), _C);
+33:   const float* A = ((const float*)(_A->memory));
+33:   const float* B = ((const float*)(_B->memory));
+33:   float* C = ((float*)(_C->memory));
+33:   for (int32_t i_outer = 0; i_outer < 64; i_outer += 1) {
+33:     for (int32_t i_inner = 0; i_inner < 8; i_inner += 1) {
+33:       for (int32_t j_outer = 0; j_outer < 63; j_outer += 1) {
+33:         for (int32_t j_inner = 0; j_inner < (1 + ((int32_t)(cinn_min(7, (499 + (-8 * j_outer)))))); j_inner += 1) {
+33:           for (int32_t k0 = 0; k0 < 200; k0 += 1) {
+33:             C[((500 * i_inner) + ((4000 * i_outer) + ((8 * j_outer) + j_inner)))] = (C[((500 * i_inner) + ((4000 * i_outer) + ((8 * j_outer) + j_inner)))] + (A[((200 * i_inner) + ((1600 * i_outer) + k0))] * B[((8 * j_outer) + ((500 * k0) + j_inner))]));
+33:           };
+33:         };
+33:       };
+33:     };
+33:   };
+33:   cinn_buffer_free((void*)(0), _C);
+33: }
+33: 
+33: [       OK ] Expr.basic (43 ms)
+33: [----------] 1 test from Expr (43 ms total)
+33: 
+33: [----------] Global test environment tear-down
+33: [==========] 1 test from 1 test case ran. (43 ms total)
+33: [  PASSED  ] 1 test.
+40: [==========] Running 1 test from 1 test case.
+40: [----------] Global test environment set-up.
+40: [----------] 1 test from Instruction
+40: [ RUN      ] Instruction.basic
+40: WARNING: Logging before InitGoogleLogging() is written to STDERR
+40: I0924 13:32:19.092864 27828 instruction_test.cc:74] data: 0.783099 + 0.903366 = 1.68647
+40: I0924 13:32:19.092911 27828 instruction_test.cc:74] data: 0.79844 + 0.983596 = 1.78204
+40: I0924 13:32:19.092916 27828 instruction_test.cc:74] data: 0.911647 + 0.66688 = 1.57853
+40: [       OK ] Instruction.basic (64 ms)
+40: [----------] 1 test from Instruction (64 ms total)
+40: 
+40: [----------] Global test environment tear-down
+40: [==========] 1 test from 1 test case ran. (64 ms total)
+40: [  PASSED  ] 1 test.
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register relu
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register relu6
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register conv2d
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register depthwise_conv2d
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register batchnorm
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register pool1d
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register pool2d
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register pool3d
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register sigmoid
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register softmax
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register slice
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register depthwise_conv2d
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register elementwise_add
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register elementwise_mul
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register scale
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register matmul
+41: I0100 00:00:00.000000 27829 registry.h:89] RAW: Register mul
+41: [==========] Running 1 test from 1 test case.
+41: [----------] Global test environment set-up.
+41: [----------] 1 test from Operator
+41: [ RUN      ] Operator.GetAttrs
+41: WARNING: Logging before InitGoogleLogging() is written to STDERR
+41: I0924 13:32:19.037526 27829 op_test.cc:46] Test Strategy Codegen:
+41: function add1 (_A, _B, _C)
+41: {
+41:   for (i, 100)
+41:   {
+41:     for (j, 32)
+41:     {
+41:       C[i, j] = (A[i, j] + B[i, j])
+41:     }
+41:   }
+41: }
+41: [       OK ] Operator.GetAttrs (7 ms)
+41: [----------] 1 test from Operator (7 ms total)
+41: 
+41: [----------] Global test environment tear-down
+41: [==========] 1 test from 1 test case ran. (7 ms total)
+41: [  PASSED  ] 1 test.
+35/68 Test #39: test_hlir_framework_scope ..............   Passed    0.01 sec
+36/68 Test #38: test_hlir_framework_tensor .............   Passed    0.02 sec
+37/68 Test #37: test_hlir_framework_infershape_pass ....   Passed    0.07 sec
+38/68 Test #36: test_hlir_framework_buffer .............   Passed    0.07 sec
+39/68 Test #35: test_cache_read_write_replace ..........   Passed    0.17 sec
+40/68 Test #34: test_optimize ..........................   Passed    0.17 sec
+41/68 Test #33: test_transform_polyfor_to_for ..........   Passed    0.17 sec
+42/68 Test #40: test_hlir_framework_instruction ........   Passed    0.17 sec
+43/68 Test #41: test_hlir_framework_op .................   Passed    0.17 sec
+test 42
+      Start 42: test_hlir_framework_print_graph_pass
+
+42: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/framework/test_hlir_framework_print_graph_pass ""
+42: Test timeout computed to be: 600
+test 43
+      Start 43: test_pe_elementwise
+
+43: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/pe/test_pe_elementwise ""
+43: Test timeout computed to be: 600
+test 44
+      Start 44: test_pe_broadcast
+
+44: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/pe/test_pe_broadcast ""
+44: Test timeout computed to be: 600
+test 45
+      Start 45: test_pe_transform
+
+45: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/pe/test_pe_transform ""
+45: Test timeout computed to be: 600
+test 46
+      Start 46: test_op_broadcast
+
+46: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/op/test_op_broadcast ""
+46: Test timeout computed to be: 600
+test 47
+      Start 47: test_op_nn
+
+47: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/hlir/op/test_op_nn ""
+47: Test timeout computed to be: 600
+test 48
+      Start 48: test_frontend_syntax
+
+48: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/frontend/test_frontend_syntax "--model_dir=/home/wangyue50/CINN-my/CINN/build/thirds/naive_mul_model"
+48: Test timeout computed to be: 600
+test 49
+      Start 49: test_frontend_executor
+
+49: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/frontend/test_frontend_executor "--model_dir=/home/wangyue50/CINN-my/CINN/build/thirds/naive_mul_model"
+49: Test timeout computed to be: 600
+test 50
+      Start 50: test_model_parser
+
+50: Test command: /home/wangyue50/CINN-my/CINN/build/cinn/frontend/paddle/test_model_parser "--model_dir=/home/wangyue50/CINN-my/CINN/build/thirds/model/lite_naive_model"
+50: Test timeout computed to be: 600
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register relu
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register relu6
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register conv2d
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register depthwise_conv2d
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register batchnorm
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register pool1d
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register pool2d
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register pool3d
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register sigmoid
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register softmax
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register slice
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register depthwise_conv2d
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register elementwise_add
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register elementwise_mul
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register scale
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register matmul
+48: I0100 00:00:00.000000 27836 registry.h:89] RAW: Register mul
+48: [==========] Running 5 tests from 2 test cases.
+48: [----------] Global test environment set-up.
+48: [----------] 4 tests from syntax
+48: [ RUN      ] syntax.basic
+48: WARNING: Logging before InitGoogleLogging() is written to STDERR
+48: I0924 13:32:19.206811 27836 syntax_test.cc:43] instruction: var_1 = elementwise_add(placeholder, placeholder_0)
+48: I0924 13:32:19.206851 27836 syntax_test.cc:43] instruction: var_2 = elementwise_add(placeholder, var_1)
+48: [       OK ] syntax.basic (0 ms)
+48: [ RUN      ] syntax.program_execute_multi_elementwise_add
+48: I0924 13:32:19.206964 27836 syntax_test.cc:57] graph:
+48: digraph G {
+48:    node_3[label="elementwise_add_0"]
+48:    node_5[label="elementwise_add_1"]
+48:    node_0[label="placeholder_1"]
+48:    node_1[label="placeholder_2"]
+48:    node_2[label="var_5"]
+48:    node_4[label="var_6"]
+48:    node_0->node_5
+48:    node_0->node_3
+48:    node_1->node_3
+48:    node_2->node_5
+48:    node_3->node_2
+48:    node_5->node_4
+48: } // end G
+48: I0924 13:32:19.223773 27836 graph_compiler.cc:39] [Debug] C Code is:
+48: #include <cinn_runtime.h>
+48: #include <stdio.h>
+48: 
+48: void fn_elementwise_add_0(void* _args, int32_t num_args)
+48: {
+48:   const cinn_buffer_t* _placeholder_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+48:   const cinn_buffer_t* _placeholder_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+48:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+48:   cinn_buffer_malloc((void*)(0), _C);
+48:   float* C = ((float*)(_C->memory));
+48:   const float* placeholder_1 = ((const float*)(_placeholder_1->memory));
+48:   const float* placeholder_2 = ((const float*)(_placeholder_2->memory));
+48:   for (int32_t i = 0; i < 32; i += 1) {
+48:     for (int32_t j = 0; j < 24; j += 1) {
+48:       C[((24 * i) + j)] = (placeholder_1[((24 * i) + j)] + placeholder_2[((24 * i) + j)]);
+48:     };
+48:   };
+48:   cinn_buffer_free((void*)(0), _C);
+48: }
+48: 
+48: void fn_elementwise_add_1(void* _args, int32_t num_args)
+48: {
+48:   const cinn_buffer_t* _placeholder_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+48:   const cinn_buffer_t* _var_5 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+48:   cinn_buffer_t* _C_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+48:   cinn_buffer_malloc((void*)(0), _C_0);
+48:   float* C_0 = ((float*)(_C_0->memory));
+48:   const float* placeholder_1 = ((const float*)(_placeholder_1->memory));
+48:   const float* var_5 = ((const float*)(_var_5->memory));
+48:   for (int32_t i = 0; i < 32; i += 1) {
+48:     for (int32_t j = 0; j < 24; j += 1) {
+48:       C_0[((24 * i) + j)] = (placeholder_1[((24 * i) + j)] + var_5[((24 * i) + j)]);
+48:     };
+48:   };
+48:   cinn_buffer_free((void*)(0), _C_0);
+48: }
+48: 
+48: [       OK ] syntax.program_execute_multi_elementwise_add (44 ms)
+48: [ RUN      ] syntax.program_execute_multi_elementwise_add2
+48: I0924 13:32:19.250627 27836 syntax_test.cc:80] graph:
+48: digraph G {
+48:    node_9[label="elementwise_add_0"]
+48:    node_11[label="elementwise_add_1"]
+48:    node_6[label="placeholder_3"]
+48:    node_7[label="placeholder_4"]
+48:    node_10[label="var_10"]
+48:    node_8[label="var_9"]
+48:    node_6->node_11
+48:    node_6->node_9
+48:    node_7->node_9
+48:    node_8->node_11
+48:    node_9->node_8
+48:    node_11->node_10
+48: } // end G
+48: I0924 13:32:19.265573 27836 graph_compiler.cc:39] [Debug] C Code is:
+48: #include <cinn_runtime.h>
+48: #include <stdio.h>
+48: 
+48: void fn_elementwise_add_0(void* _args, int32_t num_args)
+48: {
+48:   const cinn_buffer_t* _placeholder_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+48:   const cinn_buffer_t* _placeholder_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+48:   cinn_buffer_t* _C_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+48:   cinn_buffer_malloc((void*)(0), _C_1);
+48:   float* C_1 = ((float*)(_C_1->memory));
+48:   const float* placeholder_3 = ((const float*)(_placeholder_3->memory));
+48:   const float* placeholder_4 = ((const float*)(_placeholder_4->memory));
+48:   for (int32_t i = 0; i < 32; i += 1) {
+48:     for (int32_t j = 0; j < 24; j += 1) {
+48:       C_1[((24 * i) + j)] = (placeholder_3[((24 * i) + j)] + placeholder_4[((24 * i) + j)]);
+48:     };
+48:   };
+48:   cinn_buffer_free((void*)(0), _C_1);
+48: }
+48: 
+48: void fn_elementwise_add_1(void* _args, int32_t num_args)
+48: {
+48:   const cinn_buffer_t* _placeholder_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+48:   const cinn_buffer_t* _var_9 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+48:   cinn_buffer_t* _C_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+48:   cinn_buffer_malloc((void*)(0), _C_2);
+48:   float* C_2 = ((float*)(_C_2->memory));
+48:   const float* placeholder_3 = ((const float*)(_placeholder_3->memory));
+48:   const float* var_9 = ((const float*)(_var_9->memory));
+48:   for (int32_t i = 0; i < 32; i += 1) {
+48:     for (int32_t j = 0; j < 24; j += 1) {
+48:       C_2[((24 * i) + j)] = (placeholder_3[((24 * i) + j)] + var_9[((24 * i) + j)]);
+48:     };
+48:   };
+48:   cinn_buffer_free((void*)(0), _C_2);
+48: }
+48: 
+48: [       OK ] syntax.program_execute_multi_elementwise_add2 (40 ms)
+48: [ RUN      ] syntax.program_execute_fc
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register relu
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register relu6
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register conv2d
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register depthwise_conv2d
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register batchnorm
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register pool1d
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register pool2d
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register pool3d
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register sigmoid
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register softmax
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register slice
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register depthwise_conv2d
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register elementwise_add
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register elementwise_mul
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register scale
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register matmul
+47: I0100 00:00:00.000000 27835 registry.h:89] RAW: Register mul
+47: [==========] Running 5 tests from 1 test case.
+47: [----------] Global test environment set-up.
+47: [----------] 5 tests from Operator
+47: [ RUN      ] Operator.Operator_Pool2d_Test0
+47: WARNING: Logging before InitGoogleLogging() is written to STDERR
+47: I0924 13:32:19.206691 27835 nn.cc:477] kernel_size length is: 2
+47: I0924 13:32:19.206710 27835 nn.cc:478] kernel_size is: 2
+47: I0924 13:32:19.206713 27835 nn.cc:479] padding_size length is: 4
+47: I0924 13:32:19.206717 27835 nn.cc:480] padding_size is: 1
+47: I0924 13:32:19.254041 27835 op_nn_test.cc:53] Test Strategy Codegen:
+47: function pool2d (_A, _pad_temp_0, _T_Pool2d_out_0)
+47: {
+47:   for (j, 3)
+47:   {
+47:     for (k, 5)
+47:     {
+47:       for (a, 5)
+47:       {
+47:         T_Pool2d_out_0_init[0, j, k, a] = 0
+47:       }
+47:     }
+47:   }
+47:   for (j, 3)
+47:   {
+47:     for (k, 10)
+47:     {
+47:       for (a, 10)
+47:       {
+47:         pad_temp_0[0, j, k, a] = select(((a < 9) and ((a >= 1) and ((k < 9) and (k >= 1)))), A[0, j, (-1 + k), (-1 + a)], -3.40282e+38)
+47:       }
+47:     }
+47:   }
+47:   for (j, 3)
+47:   {
+47:     for (k, 5)
+47:     {
+47:       for (a, 5)
+47:       {
+47:         for (kernel_idx, 2)
+47:         {
+47:           for (kernel_idx_0, 2)
+47:           {
+47:             T_Pool2d_out_0[0, j, k, a] = cinn_max(T_Pool2d_out_0[0, j, k, a], pad_temp_0[0, j, ((2 * k) + kernel_idx), ((2 * a) + kernel_idx_0)])
+47:           }
+47:         }
+47:       }
+47:     }
+47:   }
+47: }
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 768
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 1200
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 300
+47: [       OK ] Operator.Operator_Pool2d_Test0 (79 ms)
+47: [ RUN      ] Operator.Operator_Pool2d_Test1
+47: I0924 13:32:19.285432 27835 nn.cc:477] kernel_size length is: 2
+47: I0924 13:32:19.285440 27835 nn.cc:478] kernel_size is: 2
+47: I0924 13:32:19.285442 27835 nn.cc:479] padding_size length is: 4
+47: I0924 13:32:19.285444 27835 nn.cc:480] padding_size is: 1
+47: I0924 13:32:19.332312 27835 op_nn_test.cc:109] Test Strategy Codegen:
+47: function pool2d (_A, _pad_temp_1, _T_Pool2d_out_1_0)
+47: {
+47:   for (j, 3)
+47:   {
+47:     for (k, 5)
+47:     {
+47:       for (a, 5)
+47:       {
+47:         T_Pool2d_out_1_0_init[0, j, k, a] = 0
+47:       }
+47:     }
+47:   }
+47:   for (j, 3)
+47:   {
+47:     for (k, 11)
+47:     {
+47:       for (a, 11)
+47:       {
+47:         pad_temp_1[0, j, k, a] = select(((a < 9) and ((a >= 1) and ((k < 9) and (k >= 1)))), A[0, j, (-1 + k), (-1 + a)], 0)
+47:       }
+47:     }
+47:   }
+47:   for (j, 3)
+47:   {
+47:     for (k, 5)
+47:     {
+47:       for (a, 5)
+47:       {
+47:         for (kernel_idx_1, 2)
+47:         {
+47:           for (kernel_idx_2, 2)
+47:           {
+47:             T_Pool2d_out_1_0[0, j, k, a] = (T_Pool2d_out_1_0[0, j, k, a] + (pad_temp_1[0, j, ((2 * k) + kernel_idx_1), ((2 * a) + kernel_idx_2)] * (1 / float32(((1 * 2) * 2)))))
+47:           }
+47:         }
+47:       }
+47:     }
+47:   }
+47: }
+47: I0924 13:32:19.340415 27835 codegen_llvm.cc:344] instr: i32 4
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 768
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 1452
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 300
+47: [       OK ] Operator.Operator_Pool2d_Test1 (76 ms)
+47: [ RUN      ] Operator.Operator_Pool2d_Test2
+47: I0924 13:32:19.361708 27835 nn.cc:477] kernel_size length is: 2
+47: I0924 13:32:19.361716 27835 nn.cc:478] kernel_size is: 2
+47: I0924 13:32:19.361719 27835 nn.cc:479] padding_size length is: 4
+47: I0924 13:32:19.361722 27835 nn.cc:480] padding_size is: 1
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register relu
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register relu6
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register conv2d
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register depthwise_conv2d
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register batchnorm
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register pool1d
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register pool2d
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register pool3d
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register sigmoid
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register softmax
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register slice
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register depthwise_conv2d
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register elementwise_add
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register elementwise_mul
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register scale
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register matmul
+46: I0100 00:00:00.000000 27834 registry.h:89] RAW: Register mul
+46: [==========] Running 2 tests from 1 test case.
+46: [----------] Global test environment set-up.
+46: [----------] 2 tests from Operator
+46: [ RUN      ] Operator.Operator_ElementWise_Add_Test0
+46: WARNING: Logging before InitGoogleLogging() is written to STDERR
+46: I0924 13:32:19.210886 27834 op_broadcast_test.cc:44] Test Strategy Codegen:
+46: function add1 (_A, _B, _C)
+46: {
+46:   for (i, 100)
+46:   {
+46:     for (j, 32)
+46:     {
+46:       C[i, j] = (A[i, j] + B[i, j])
+46:     }
+46:   }
+46: }
+46: [       OK ] Operator.Operator_ElementWise_Add_Test0 (7 ms)
+46: [ RUN      ] Operator.Operator_ElementWise_Add_Test1
+46: I0924 13:32:19.217064 27834 op_broadcast_test.cc:76] Test Strategy Codegen:
+46: function add1 (_A, _B, _C_0)
+46: {
+46:   for (i, 100)
+46:   {
+46:     for (j, 32)
+46:     {
+46:       C_0[i, j] = (A[i, j] + B[j])
+46:     }
+46:   }
+46: }
+46: function add1 (_A, _B, _C_0)
+46: {
+46:   for (i, 100)
+46:   {
+46:     for (j, 32)
+46:     {
+46:       C_0[i, j] = (A[i, j] + B[j])
+46:     }
+46:   }
+46: }[       OK ] Operator.Operator_ElementWise_Add_Test1 (7 ms)
+46: [----------] 2 tests from Operator (14 ms total)
+46: 
+46: [----------] Global test environment tear-down
+46: [==========] 2 tests from 1 test case ran. (14 ms total)
+46: [  PASSED  ] 2 tests.
+45: [==========] Running 1 test from 1 test case.
+45: [----------] Global test environment set-up.
+45: [----------] 1 test from MatmulPE
+45: [ RUN      ] MatmulPE.PE_Matmul_Test0
+45: WARNING: Logging before InitGoogleLogging() is written to STDERR
+45: I0924 13:32:19.210407 27833 pe_transform_test.cc:32] func:
+45: function fn (_A, _B, _C)
+45: {
+45:   for (i, 100)
+45:   {
+45:     for (j, 32)
+45:     {
+45:       for (kk, 16)
+45:       {
+45:         C[i, j] = (C[i, j] + (A[i, kk] * B[kk, j]))
+45:       }
+45:     }
+45:   }
+45: }
+45: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 6400
+45: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 2048
+45: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+45: [       OK ] MatmulPE.PE_Matmul_Test0 (38 ms)
+45: [----------] 1 test from MatmulPE (38 ms total)
+45: 
+45: [----------] Global test environment tear-down
+45: [==========] 1 test from 1 test case ran. (38 ms total)
+45: [  PASSED  ] 1 test.
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4800
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 3200
+11: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 2400
+11: [       OK ] cinn_cpu_mkl_gemm_fp32.test (49 ms)
+11: [----------] 1 test from cinn_cpu_mkl_gemm_fp32 (49 ms total)
+11: 
+11: [----------] Global test environment tear-down
+11: [==========] 27 tests from 2 test cases ran. (832 ms total)
+11: [  PASSED  ] 27 tests.
+44: [==========] Running 6 tests from 3 test cases.
+44: [----------] Global test environment set-up.
+44: [----------] 2 tests from broadcast_pe
+44: [ RUN      ] broadcast_pe.Add
+44: WARNING: Logging before InitGoogleLogging() is written to STDERR
+44: I0924 13:32:19.207803 27832 pe_broadcast_test.cc:33] func:
+44: function fn (_A, _B, _C)
+44: {
+44:   for (i, 100)
+44:   {
+44:     for (j, 32)
+44:     {
+44:       C[i, j] = (A[i, j] + B[i, j])
+44:     }
+44:   }
+44: }
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+44: [       OK ] broadcast_pe.Add (34 ms)
+44: [ RUN      ] broadcast_pe.Multiply
+44: I0924 13:32:19.240713 27832 pe_broadcast_test.cc:33] func:
+44: function fn (_A, _B, _C)
+44: {
+44:   for (i, 100)
+44:   {
+44:     for (j, 32)
+44:     {
+44:       C[i, j] = (A[i, j] * B[i, j])
+44:     }
+44:   }
+44: }
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+44: [       OK ] broadcast_pe.Multiply (30 ms)
+44: [----------] 2 tests from broadcast_pe (64 ms total)
+44: 
+44: [----------] 2 tests from broadcast_pe1
+44: [ RUN      ] broadcast_pe1.Add
+44: I0924 13:32:19.274186 27832 pe_broadcast_test.cc:80] func:
+44: function fn (_A, _B, _C)
+44: {
+44:   for (i, 100)
+44:   {
+44:     for (j, 32)
+44:     {
+44:       for (k, 10)
+44:       {
+44:         C[i, j, k] = (A[i, j, k] + B[j])
+44:       }
+44:     }
+44:   }
+44: }
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128000
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128000
+44: [       OK ] broadcast_pe1.Add (35 ms)
+44: [ RUN      ] broadcast_pe1.Multiply
+44: I0924 13:32:19.309664 27832 pe_broadcast_test.cc:80] func:
+44: function fn (_A, _B, _C)
+44: {
+44:   for (i, 100)
+44:   {
+44:     for (j, 32)
+44:     {
+44:       for (k, 10)
+44:       {
+44:         C[i, j, k] = (A[i, j, k] * B[j])
+44:       }
+44:     }
+44:   }
+44: }
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128000
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128000
+44: [       OK ] broadcast_pe1.Multiply (36 ms)
+44: [----------] 2 tests from broadcast_pe1 (71 ms total)
+44: 
+44: [----------] 2 tests from broadcast_pe2
+44: [ RUN      ] broadcast_pe2.Add
+44: I0924 13:32:19.348361 27832 pe_broadcast_test.cc:127] func:
+44: function fn (_A, _B, _C)
+44: {
+44:   for (i, 100)
+44:   {
+44:     for (j, 32)
+44:     {
+44:       for (k, 10)
+44:       {
+44:         C[i, j, k, 0] = (A[i, j, k, 0] + B[j, k])
+44:       }
+44:     }
+44:   }
+44: }
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128000
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 1280
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128000
+44: [       OK ] broadcast_pe2.Add (40 ms)
+44: [ RUN      ] broadcast_pe2.Multiply
+44: I0924 13:32:19.388643 27832 pe_broadcast_test.cc:127] func:
+44: function fn (_A, _B, _C)
+44: {
+44:   for (i, 100)
+44:   {
+44:     for (j, 32)
+44:     {
+44:       for (k, 10)
+44:       {
+44:         C[i, j, k, 0] = (A[i, j, k, 0] * B[j, k])
+44:       }
+44:     }
+44:   }
+44: }
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128000
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 1280
+44: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128000
+44: [       OK ] broadcast_pe2.Multiply (40 ms)
+44: [----------] 2 tests from broadcast_pe2 (80 ms total)
+44: 
+44: [----------] Global test environment tear-down
+44: [==========] 6 tests from 3 test cases ran. (215 ms total)
+44: [  PASSED  ] 6 tests.
+43: [==========] Running 25 tests from 1 test case.
+43: [----------] Global test environment set-up.
+43: [----------] 25 tests from elementwise_pe
+43: [ RUN      ] elementwise_pe.exp
+43: WARNING: Logging before InitGoogleLogging() is written to STDERR
+43: I0924 13:32:19.206897 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_exp_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_exp_fp32_out[i, j] = cinn_cpu_exp_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.exp (33 ms)
+43: [ RUN      ] elementwise_pe.erf
+43: I0924 13:32:19.239110 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_erf_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_erf_fp32_out[i, j] = cinn_cpu_erf_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.erf (29 ms)
+43: [ RUN      ] elementwise_pe.sqrt
+43: I0924 13:32:19.268132 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_sqrt_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_sqrt_fp32_out[i, j] = cinn_cpu_sqrt_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.sqrt (29 ms)
+43: [ RUN      ] elementwise_pe.log
+43: I0924 13:32:19.296869 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_log_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_log_fp32_out[i, j] = cinn_cpu_log_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.log (29 ms)
+43: [ RUN      ] elementwise_pe.log2
+43: I0924 13:32:19.325762 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_log2_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_log2_fp32_out[i, j] = cinn_cpu_log2_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.log2 (29 ms)
+43: [ RUN      ] elementwise_pe.log10
+43: I0924 13:32:19.354607 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_log10_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_log10_fp32_out[i, j] = cinn_cpu_log10_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.log10 (29 ms)
+43: [ RUN      ] elementwise_pe.floor
+43: I0924 13:32:19.383556 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_floor_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_floor_fp32_out[i, j] = cinn_cpu_floor_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.floor (29 ms)
+43: [ RUN      ] elementwise_pe.ceil
+43: I0924 13:32:19.412417 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_ceil_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_ceil_fp32_out[i, j] = cinn_cpu_ceil_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.ceil (29 ms)
+43: [ RUN      ] elementwise_pe.round
+43: I0924 13:32:19.441426 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_round_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_round_fp32_out[i, j] = cinn_cpu_round_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.round (28 ms)
+43: [ RUN      ] elementwise_pe.trunc
+43: I0924 13:32:19.470185 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_trunc_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_trunc_fp32_out[i, j] = cinn_cpu_trunc_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.trunc (29 ms)
+43: [ RUN      ] elementwise_pe.cos
+43: I0924 13:32:19.498968 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_cos_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_cos_fp32_out[i, j] = cinn_cpu_cos_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register relu
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register relu6
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register conv2d
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register depthwise_conv2d
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register batchnorm
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register pool1d
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register pool2d
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register pool3d
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register sigmoid
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register softmax
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register slice
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register depthwise_conv2d
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register elementwise_add
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register elementwise_mul
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register scale
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register matmul
+42: I0100 00:00:00.000000 27830 registry.h:89] RAW: Register mul
+42: [==========] Running 1 test from 1 test case.
+42: [----------] Global test environment set-up.
+42: [----------] 1 test from Operator
+42: [ RUN      ] Operator.GetAttrs
+42: WARNING: Logging before InitGoogleLogging() is written to STDERR
+42: I0924 13:32:19.200210 27830 print_graph_pass_test.cc:53] 0:elementwise_add(elementwise_add_0)
+42: 1:elementwise_add(elementwise_add_1)
+42: 2:elementwise_add(elementwise_add_2)
+42: [       OK ] Operator.GetAttrs (1 ms)
+42: [----------] 1 test from Operator (1 ms total)
+42: 
+42: [----------] Global test environment tear-down
+42: [==========] 1 test from 1 test case ran. (1 ms total)
+42: [  PASSED  ] 1 test.
+50: [==========] Running 1 test from 1 test case.
+50: [----------] Global test environment set-up.
+50: [----------] 1 test from LoadModelPb
+50: [ RUN      ] LoadModelPb.naive_model
+50: WARNING: Logging before InitGoogleLogging() is written to STDERR
+50: I0924 13:32:19.206583 27838 model_parser.cc:188] model_dir is: /home/wangyue50/CINN-my/CINN/build/thirds/model/lite_naive_model
+50: I0924 13:32:19.206604 27838 model_parser.cc:189] model_file is: __model__
+50: I0924 13:32:19.206606 27838 model_parser.cc:190] param_file is: 
+50: I0924 13:32:19.207155 27838 model_parser_test.cc:21] feed
+50: I0924 13:32:19.207161 27838 model_parser_test.cc:21] mul
+50: I0924 13:32:19.207163 27838 model_parser_test.cc:21] scale
+50: I0924 13:32:19.207165 27838 model_parser_test.cc:21] fetch
+50: [       OK ] LoadModelPb.naive_model (1 ms)
+50: [----------] 1 test from LoadModelPb (1 ms total)
+50: 
+50: [----------] Global test environment tear-down
+50: [==========] 1 test from 1 test case ran. (1 ms total)
+50: [  PASSED  ] 1 test.
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register relu
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register relu6
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register conv2d
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register depthwise_conv2d
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register batchnorm
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register pool1d
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register pool2d
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register pool3d
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register sigmoid
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register softmax
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register slice
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register depthwise_conv2d
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register elementwise_add
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register elementwise_mul
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register scale
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register matmul
+49: I0100 00:00:00.000000 27837 registry.h:89] RAW: Register mul
+49: [==========] Running 1 test from 1 test case.
+49: [----------] Global test environment set-up.
+49: [----------] 1 test from Executor
+49: [ RUN      ] Executor.basic
+49: WARNING: Logging before InitGoogleLogging() is written to STDERR
+49: I0924 13:32:19.207593 27837 syntax.cc:130] Loading Paddle model from /home/wangyue50/CINN-my/CINN/build/thirds/naive_mul_model
+49: I0924 13:32:19.207660 27837 model_parser.cc:188] model_dir is: /home/wangyue50/CINN-my/CINN/build/thirds/naive_mul_model
+49: I0924 13:32:19.207664 27837 model_parser.cc:189] model_file is: __model__
+49: I0924 13:32:19.207665 27837 model_parser.cc:190] param_file is: 
+49: I0924 13:32:19.208698 27837 paddle_model_to_program.cc:26] detect model output: [save_infer_model/scale_0.tmp_0]
+49: I0924 13:32:19.208746 27837 executor.cc:49] Program:
+49: Program {
+49: var_3 = mul(A, fc_0__w_0, y_num_col_dims=1, x_num_col_dims=1)
+49: var_7 = elementwise_add(var_3, fc_0__b_0, axis=1)
+49: var_9 = relu(var_7)
+49: var_11 = scale(var_9, scale=1)
+49: }
+49: I0924 13:32:19.239845 27837 graph_compiler.cc:39] [Debug] C Code is:
+49: #include <cinn_runtime.h>
+49: #include <stdio.h>
+49: 
+49: void fn_mul_0(void* _args, int32_t num_args)
+49: {
+49:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+49:   const cinn_buffer_t* _fc_0__w_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+49:   cinn_buffer_t* _Mul_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+49:   cinn_buffer_malloc((void*)(0), _Mul_out);
+49:   const float* A_reshape = ((const float*)(_A->memory));
+49:   float* Mul_out = ((float*)(_Mul_out->memory));
+49:   float* Mul_out_init = ((float*)(_Mul_out->memory));
+49:   const float* fc_0__w_0_reshape = ((const float*)(_fc_0__w_0->memory));
+49:   for (int32_t j = 0; j < 30; j += 1) {
+49:     Mul_out_init[j] = 0;
+49:   };
+49:   for (int32_t j = 0; j < 30; j += 1) {
+49:     for (int32_t axis_k = 0; axis_k < 30; axis_k += 1) {
+49:       Mul_out[j] = (Mul_out[j] + (A_reshape[axis_k] * fc_0__w_0_reshape[((30 * axis_k) + j)]));
+49:     };
+49:   };
+49:   cinn_buffer_free((void*)(0), _Mul_out);
+49: }
+49: 
+49: void fn_elementwise_add_1(void* _args, int32_t num_args)
+49: {
+49:   const cinn_buffer_t* _var_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+49:   const cinn_buffer_t* _fc_0__b_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+49:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+49:   cinn_buffer_malloc((void*)(0), _C);
+49:   float* C = ((float*)(_C->memory));
+49:   const float* fc_0__b_0 = ((const float*)(_fc_0__b_0->memory));
+49:   const float* var_3 = ((const float*)(_var_3->memory));
+49:   for (int32_t j = 0; j < 30; j += 1) {
+49:     C[j] = (var_3[j] + fc_0__b_0[j]);
+49:   };
+49:   cinn_buffer_free((void*)(0), _C);
+49: }
+49: 
+49: void fn_relu_2(void* _args, int32_t num_args)
+49: {
+49:   const cinn_buffer_t* _var_7 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+49:   cinn_buffer_t* _Relu_output = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+49:   cinn_buffer_malloc((void*)(0), _Relu_output);
+49:   float* Relu_output = ((float*)(_Relu_output->memory));
+49:   const float* var_7 = ((const float*)(_var_7->memory));
+49:   for (int32_t j = 0; j < 30; j += 1) {
+49:     Relu_output[j] = cinn_max(var_7[j], 0);
+49:   };
+49:   cinn_buffer_free((void*)(0), _Relu_output);
+49: }
+49: 
+49: void fn_scale_3(void* _args, int32_t num_args)
+49: {
+49:   const cinn_buffer_t* _var_9 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+49:   cinn_buffer_t* _Scale_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+49:   cinn_buffer_malloc((void*)(0), _Scale_out);
+49:   float* Scale_out = ((float*)(_Scale_out->memory));
+49:   const float* var_9 = ((const float*)(_var_9->memory));
+49:   for (int32_t j = 0; j < 30; j += 1) {
+49:     Scale_out[j] = var_9[j];
+49:   };
+49:   cinn_buffer_free((void*)(0), _Scale_out);
+49: }
+49: 
+49: [       OK ] Executor.basic (61 ms)
+49: [----------] 1 test from Executor (61 ms total)
+49: 
+49: [----------] Global test environment tear-down
+49: [==========] 1 test from 1 test case ran. (61 ms total)
+49: [  PASSED  ] 1 test.
+44/68 Test #46: test_op_broadcast ......................   Passed    0.21 sec
+45/68 Test #45: test_pe_transform ......................   Passed    0.21 sec
+46/68 Test #11: test_mkl_math ..........................   Passed    1.06 sec
+47/68 Test #44: test_pe_broadcast ......................   Passed    0.23 sec
+48/68 Test #42: test_hlir_framework_print_graph_pass ...   Passed    0.33 sec
+49/68 Test #50: test_model_parser ......................   Passed    0.32 sec
+50/68 Test #49: test_frontend_executor .................   Passed    0.32 sec
+test 52
+      Start 52: test01_elementwise_add_case
+
+52: Test command: /home/wangyue50/CINN-my/CINN/build/tests/test01_elementwise_add_case ""
+52: Test timeout computed to be: 600
+test 54
+      Start 54: test02_matmul_case
+
+54: Test command: /home/wangyue50/CINN-my/CINN/build/tests/test02_matmul_case ""
+54: Test timeout computed to be: 600
+test 56
+      Start 56: test03_conv_case
+
+56: Test command: /home/wangyue50/CINN-my/CINN/build/tests/test03_conv_case ""
+56: Test timeout computed to be: 600
+test 57
+      Start 57: test_cinn_python_api
+
+57: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_matmul.py"
+57: Test timeout computed to be: 9.99988e+06
+test 58
+      Start 58: test_cinn_common
+
+58: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_common.py"
+58: Test timeout computed to be: 9.99988e+06
+test 59
+      Start 59: test_cinn_packed_func
+
+59: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_packed_func.py"
+59: Test timeout computed to be: 9.99988e+06
+test 60
+      Start 60: test_cinn_pe_elementwise
+
+60: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_pe_elementwise.py"
+60: Test timeout computed to be: 9.99988e+06
+48: I0924 13:32:19.314476 27836 graph_compiler.cc:39] [Debug] C Code is:
+48: #include <cinn_runtime.h>
+48: #include <stdio.h>
+48: 
+48: void fn_mul_0(void* _args, int32_t num_args)
+48: {
+48:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+48:   const cinn_buffer_t* _W = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+48:   cinn_buffer_t* _Mul_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+48:   cinn_buffer_malloc((void*)(0), _Mul_out);
+48:   const float* A_reshape = ((const float*)(_A->memory));
+48:   float* Mul_out = ((float*)(_Mul_out->memory));
+48:   float* Mul_out_init = ((float*)(_Mul_out->memory));
+48:   const float* W_reshape = ((const float*)(_W->memory));
+48:   for (int32_t i = 0; i < 320; i += 1) {
+48:     for (int32_t j = 0; j < 24; j += 1) {
+48:       Mul_out_init[((24 * i) + j)] = 0;
+48:     };
+48:   };
+48:   for (int32_t i = 0; i < 320; i += 1) {
+48:     for (int32_t j = 0; j < 24; j += 1) {
+48:       for (int32_t axis_k = 0; axis_k < 18; axis_k += 1) {
+48:         Mul_out[((24 * i) + j)] = (Mul_out[((24 * i) + j)] + (A_reshape[((18 * i) + axis_k)] * W_reshape[((24 * axis_k) + j)]));
+48:       };
+48:     };
+48:   };
+48:   cinn_buffer_free((void*)(0), _Mul_out);
+48: }
+48: 
+48: void fn_elementwise_add_1(void* _args, int32_t num_args)
+48: {
+48:   const cinn_buffer_t* _var_14 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+48:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+48:   cinn_buffer_t* _C_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+48:   cinn_buffer_malloc((void*)(0), _C_3);
+48:   const float* B = ((const float*)(_B->memory));
+48:   float* C_3 = ((float*)(_C_3->memory));
+48:   const float* var_14 = ((const float*)(_var_14->memory));
+48:   for (int32_t i = 0; i < 320; i += 1) {
+48:     for (int32_t j = 0; j < 24; j += 1) {
+48:       C_3[((24 * i) + j)] = (var_14[((24 * i) + j)] + B[j]);
+48:     };
+48:   };
+48:   cinn_buffer_free((void*)(0), _C_3);
+48: }
+48: 
+48: [       OK ] syntax.program_execute_fc (50 ms)
+48: [----------] 4 tests from syntax (134 ms total)
+48: 
+48: [----------] 1 test from load_paddle_model
+48: [ RUN      ] load_paddle_model.fc_execute
+48: I0924 13:32:19.340806 27836 syntax.cc:130] Loading Paddle model from /home/wangyue50/CINN-my/CINN/build/thirds/naive_mul_model
+48: I0924 13:32:19.340857 27836 model_parser.cc:188] model_dir is: /home/wangyue50/CINN-my/CINN/build/thirds/naive_mul_model
+48: I0924 13:32:19.340860 27836 model_parser.cc:189] model_file is: __model__
+48: I0924 13:32:19.340862 27836 model_parser.cc:190] param_file is: 
+48: I0924 13:32:19.341847 27836 paddle_model_to_program.cc:26] detect model output: [save_infer_model/scale_0.tmp_0]
+48: I0924 13:32:19.341898 27836 syntax_test.cc:152] program:
+48: Program {
+48: var_20 = mul(A, fc_0__w_0, y_num_col_dims=1, x_num_col_dims=1)
+48: var_24 = elementwise_add(var_20, fc_0__b_0, axis=1)
+48: var_26 = relu(var_24)
+48: var_28 = scale(var_26, scale=1)
+48: }
+48: I0924 13:32:19.371641 27836 graph_compiler.cc:39] [Debug] C Code is:
+48: #include <cinn_runtime.h>
+48: #include <stdio.h>
+48: 
+48: void fn_mul_0(void* _args, int32_t num_args)
+48: {
+48:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+48:   const cinn_buffer_t* _fc_0__w_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+48:   cinn_buffer_t* _Mul_out_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+48:   cinn_buffer_malloc((void*)(0), _Mul_out_0);
+48:   const float* A_reshape_0 = ((const float*)(_A->memory));
+48:   float* Mul_out_0 = ((float*)(_Mul_out_0->memory));
+48:   float* Mul_out_0_init = ((float*)(_Mul_out_0->memory));
+48:   const float* fc_0__w_0_reshape = ((const float*)(_fc_0__w_0->memory));
+48:   for (int32_t j = 0; j < 30; j += 1) {
+48:     Mul_out_0_init[j] = 0;
+48:   };
+48:   for (int32_t j = 0; j < 30; j += 1) {
+48:     for (int32_t axis_k_0 = 0; axis_k_0 < 30; axis_k_0 += 1) {
+48:       Mul_out_0[j] = (Mul_out_0[j] + (A_reshape_0[axis_k_0] * fc_0__w_0_reshape[((30 * axis_k_0) + j)]));
+48:     };
+48:   };
+48:   cinn_buffer_free((void*)(0), _Mul_out_0);
+48: }
+48: 
+48: void fn_elementwise_add_1(void* _args, int32_t num_args)
+48: {
+48:   const cinn_buffer_t* _var_20 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+48:   const cinn_buffer_t* _fc_0__b_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+48:   cinn_buffer_t* _C_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+48:   cinn_buffer_malloc((void*)(0), _C_4);
+48:   float* C_4 = ((float*)(_C_4->memory));
+48:   const float* fc_0__b_0 = ((const float*)(_fc_0__b_0->memory));
+48:   const float* var_20 = ((const float*)(_var_20->memory));
+48:   for (int32_t j = 0; j < 30; j += 1) {
+48:     C_4[j] = (var_20[j] + fc_0__b_0[j]);
+48:   };
+48:   cinn_buffer_free((void*)(0), _C_4);
+48: }
+48: 
+48: void fn_relu_2(void* _args, int32_t num_args)
+48: {
+48:   const cinn_buffer_t* _var_24 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+48:   cinn_buffer_t* _Relu_output = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+48:   cinn_buffer_malloc((void*)(0), _Relu_output);
+48:   float* Relu_output = ((float*)(_Relu_output->memory));
+48:   const float* var_24 = ((const float*)(_var_24->memory));
+48:   for (int32_t j = 0; j < 30; j += 1) {
+48:     Relu_output[j] = cinn_max(var_24[j], 0);
+48:   };
+48:   cinn_buffer_free((void*)(0), _Relu_output);
+48: }
+48: 
+48: void fn_scale_3(void* _args, int32_t num_args)
+48: {
+48:   const cinn_buffer_t* _var_26 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+48:   cinn_buffer_t* _Scale_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+48:   cinn_buffer_malloc((void*)(0), _Scale_out);
+48:   float* Scale_out = ((float*)(_Scale_out->memory));
+48:   const float* var_26 = ((const float*)(_var_26->memory));
+48:   for (int32_t j = 0; j < 30; j += 1) {
+48:     Scale_out[j] = var_26[j];
+48:   };
+48:   cinn_buffer_free((void*)(0), _Scale_out);
+48: }
+48: 
+48: I0924 13:32:19.398069 27836 syntax_test.cc:169] scope.names: var_26,fc_0__w_0,fc_0__b_0,var_28,A,var_20,var_24
+48: I0924 13:32:19.398090 27836 syntax_test.cc:173] tensor.shape: 1,30
+48: I0924 13:32:19.398094 27836 syntax_test.cc:175] data: 1.1576
+48: I0924 13:32:19.398100 27836 syntax_test.cc:175] data: -0.79889
+48: I0924 13:32:19.398105 27836 syntax_test.cc:175] data: 0.125167
+48: I0924 13:32:19.398108 27836 syntax_test.cc:175] data: 0.0934057
+48: I0924 13:32:19.398110 27836 syntax_test.cc:175] data: 0.0106909
+48: I0924 13:32:19.398113 27836 syntax_test.cc:175] data: 0.60541
+48: I0924 13:32:19.398114 27836 syntax_test.cc:175] data: -0.777485
+48: I0924 13:32:19.398116 27836 syntax_test.cc:175] data: 0.546453
+48: I0924 13:32:19.398119 27836 syntax_test.cc:175] data: -0.624406
+48: I0924 13:32:19.398121 27836 syntax_test.cc:175] data: 1.4859
+48: [       OK ] load_paddle_model.fc_execute (58 ms)
+48: [----------] 1 test from load_paddle_model (58 ms total)
+48: 
+48: [----------] Global test environment tear-down
+48: [==========] 5 tests from 2 test cases ran. (192 ms total)
+48: [  PASSED  ] 5 tests.
+47: I0924 13:32:19.415700 27835 op_nn_test.cc:167] Test Strategy Codegen:
+47: function pool2d (_A, _pad_temp_2, _T_Pool2d_out_2_0)
+47: {
+47:   for (j, 5)
+47:   {
+47:     for (k, 5)
+47:     {
+47:       for (a, 3)
+47:       {
+47:         T_Pool2d_out_2_0_init[0, j, k, a] = 0
+47:       }
+47:     }
+47:   }
+47:   for (j, 11)
+47:   {
+47:     for (k, 11)
+47:     {
+47:       for (a, 3)
+47:       {
+47:         pad_temp_2[0, j, k, a] = select(((k < 9) and ((k >= 1) and ((j < 9) and (j >= 1)))), A[0, (-1 + j), (-1 + k), a], 0)
+47:       }
+47:     }
+47:   }
+47:   for (j, 5)
+47:   {
+47:     for (k, 5)
+47:     {
+47:       for (a, 3)
+47:       {
+47:         for (kernel_idx_3, 2)
+47:         {
+47:           for (kernel_idx_4, 2)
+47:           {
+47:             T_Pool2d_out_2_0[0, j, k, a] = (T_Pool2d_out_2_0[0, j, k, a] + (pad_temp_2[0, ((2 * j) + kernel_idx_3), ((2 * k) + kernel_idx_4), a] * (1 / float32(cinn_max(((1 * (cinn_min(((-1 + (2 * j)) + 2), 8) - cinn_max((-1 + (2 * j)), 0))) * (cinn_min(((-1 + (2 * k)) + 2), 8) - cinn_max((-1 + (2 * k)), 0))), 1)))))
+47:           }
+47:         }
+47:       }
+47:     }
+47:   }
+47: }
+47: I0924 13:32:19.424890 27835 codegen_llvm.cc:344] instr:   %89 = select i1 %88, i32 %87, i32 1
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 768
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 1452
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 300
+47: [       OK ] Operator.Operator_Pool2d_Test2 (85 ms)
+47: [ RUN      ] Operator.Operator_Pool3d_Test0
+47: I0924 13:32:19.446368 27835 nn.cc:477] kernel_size length is: 3
+47: I0924 13:32:19.446377 27835 nn.cc:478] kernel_size is: 2
+47: I0924 13:32:19.446380 27835 nn.cc:479] padding_size length is: 6
+47: I0924 13:32:19.446382 27835 nn.cc:480] padding_size is: 1
+47: I0924 13:32:19.519640 27835 op_nn_test.cc:226] Test Strategy Codegen:
+47: function pool3d (_A, _pad_temp_3, _T_Pool3d_out_0)
+47: {
+47:   for (j, 5)
+47:   {
+47:     for (k, 5)
+47:     {
+47:       for (a, 5)
+47:       {
+47:         for (b, 3)
+47:         {
+47:           T_Pool3d_out_0_init[0, j, k, a, b] = 0
+47:         }
+47:       }
+47:     }
+47:   }
+47:   for (j, 10)
+47:   {
+47:     for (k, 10)
+47:     {
+47:       for (a, 10)
+47:       {
+47:         for (b, 3)
+47:         {
+47:           pad_temp_3[0, j, k, a, b] = select(((a < 9) and ((a >= 1) and ((k < 9) and ((k >= 1) and ((j < 9) and (j >= 1)))))), A[0, (-1 + j), (-1 + k), (-1 + a), b], -3.40282e+38)
+47:         }
+47:       }
+47:     }
+47:   }
+47:   for (j, 5)
+47:   {
+47:     for (k, 5)
+47:     {
+47:       for (a, 5)
+47:       {
+47:         for (b, 3)
+47:         {
+47:           for (kernel_idx_5, 2)
+47:           {
+47:             for (kernel_idx_6, 2)
+47:             {
+47:               for (kernel_idx_7, 2)
+47:               {
+47:                 T_Pool3d_out_0[0, j, k, a, b] = cinn_max(T_Pool3d_out_0[0, j, k, a, b], pad_temp_3[0, ((2 * j) + kernel_idx_5), ((2 * k) + kernel_idx_6), ((2 * a) + kernel_idx_7), b])
+47:               }
+47:             }
+47:           }
+47:         }
+47:       }
+47:     }
+47:   }
+47: }
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 6144
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 15972
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 1500
+47: [       OK ] Operator.Operator_Pool3d_Test0 (111 ms)
+47: [ RUN      ] Operator.Operator_Pool1d_Test0
+47: I0924 13:32:19.558034 27835 nn.cc:477] kernel_size length is: 1
+47: I0924 13:32:19.558044 27835 nn.cc:478] kernel_size is: 2
+47: I0924 13:32:19.558048 27835 nn.cc:479] padding_size length is: 2
+47: I0924 13:32:19.558050 27835 nn.cc:480] padding_size is: 1
+47: I0924 13:32:19.587836 27835 op_nn_test.cc:284] Test Strategy Codegen:
+47: function pool1d (_A, _pad_temp_4, _T_Pool1d_out_0)
+47: {
+47:   for (j, 5)
+47:   {
+47:     for (k, 3)
+47:     {
+47:       T_Pool1d_out_0_init[0, j, k] = 0
+47:     }
+47:   }
+47:   for (j, 10)
+47:   {
+47:     for (k, 3)
+47:     {
+47:       pad_temp_4[0, j, k] = select(((j < 9) and (j >= 1)), A[0, (-1 + j), k], -3.40282e+38)
+47:     }
+47:   }
+47:   for (j, 5)
+47:   {
+47:     for (k, 3)
+47:     {
+47:       for (kernel_idx_8, 2)
+47:       {
+47:         T_Pool1d_out_0[0, j, k] = cinn_max(T_Pool1d_out_0[0, j, k], pad_temp_4[0, ((2 * j) + kernel_idx_8), k])
+47:       }
+47:     }
+47:   }
+47: }
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 96
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 132
+47: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 60
+47: [       OK ] Operator.Operator_Pool1d_Test0 (70 ms)
+47: [----------] 5 tests from Operator (421 ms total)
+47: 
+47: [----------] Global test environment tear-down
+47: [==========] 5 tests from 1 test case ran. (421 ms total)
+47: [  PASSED  ] 5 tests.
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register relu
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register relu6
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register conv2d
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register depthwise_conv2d
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register batchnorm
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register pool1d
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register pool2d
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register pool3d
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register sigmoid
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register softmax
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register slice
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register depthwise_conv2d
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register elementwise_add
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register elementwise_mul
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register scale
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register matmul
+60: I0100 00:00:00.000000 27850 registry.h:89] RAW: Register mul
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register relu
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register relu6
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register conv2d
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register depthwise_conv2d
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register batchnorm
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register pool1d
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register pool2d
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register pool3d
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register sigmoid
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register softmax
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register slice
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register depthwise_conv2d
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register elementwise_add
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register elementwise_mul
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register scale
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register matmul
+59: I0100 00:00:00.000000 27849 registry.h:89] RAW: Register mul
+59: ....
+59: ----------------------------------------------------------------------
+59: Ran 4 tests in 0.002s
+59: 
+59: OK
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register relu
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register relu6
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register conv2d
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register depthwise_conv2d
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register batchnorm
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register pool1d
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register pool2d
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register pool3d
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register sigmoid
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register softmax
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register slice
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register depthwise_conv2d
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register elementwise_add
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register elementwise_mul
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register scale
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register matmul
+58: I0100 00:00:00.000000 27848 registry.h:89] RAW: Register mul
+58: ..
+58: ----------------------------------------------------------------------
+58: Ran 2 tests in 0.001s
+58: 
+58: OK
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register relu
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register relu6
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register conv2d
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register depthwise_conv2d
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register batchnorm
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register pool1d
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register pool2d
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register pool3d
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register sigmoid
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register softmax
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register slice
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register depthwise_conv2d
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register elementwise_add
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register elementwise_mul
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register scale
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register matmul
+57: I0100 00:00:00.000000 27847 registry.h:89] RAW: Register mul
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.cos (29 ms)
+43: [ RUN      ] elementwise_pe.cosh
+43: I0924 13:32:19.527927 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_cosh_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_cosh_fp32_out[i, j] = cinn_cpu_cosh_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.cosh (30 ms)
+43: [ RUN      ] elementwise_pe.tan
+43: I0924 13:32:19.557880 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_tan_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_tan_fp32_out[i, j] = cinn_cpu_tan_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.tan (32 ms)
+43: [ RUN      ] elementwise_pe.sin
+43: I0924 13:32:19.593026 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_sin_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_sin_fp32_out[i, j] = cinn_cpu_sin_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.sin (43 ms)
+43: [ RUN      ] elementwise_pe.sinh
+43: I0924 13:32:19.635660 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_sinh_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_sinh_fp32_out[i, j] = cinn_cpu_sinh_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.sinh (43 ms)
+43: [ RUN      ] elementwise_pe.acos
+43: I0924 13:32:19.678943 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_acos_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_acos_fp32_out[i, j] = cinn_cpu_acos_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.acos (39 ms)
+43: [ RUN      ] elementwise_pe.acosh
+43: I0924 13:32:19.714777 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_acosh_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_acosh_fp32_out[i, j] = cinn_cpu_acosh_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.acosh (30 ms)
+43: [ RUN      ] elementwise_pe.asin
+43: I0924 13:32:19.744695 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_asin_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_asin_fp32_out[i, j] = cinn_cpu_asin_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.asin (29 ms)
+43: [ RUN      ] elementwise_pe.asinh
+43: I0924 13:32:19.774551 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_asinh_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_asinh_fp32_out[i, j] = cinn_cpu_asinh_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.asinh (31 ms)
+43: [ RUN      ] elementwise_pe.atan
+43: I0924 13:32:19.804793 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_atan_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_atan_fp32_out[i, j] = cinn_cpu_atan_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.atan (30 ms)
+43: [ RUN      ] elementwise_pe.atanh
+43: I0924 13:32:19.834564 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_atanh_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_atanh_fp32_out[i, j] = cinn_cpu_atanh_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.atanh (29 ms)
+43: [ RUN      ] elementwise_pe.isnan
+43: I0924 13:32:19.864051 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_isnan_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_isnan_fp32_out[i, j] = cinn_cpu_isnan_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.isnan (29 ms)
+43: [ RUN      ] elementwise_pe.tanh
+43: I0924 13:32:19.893018 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_tanh_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_tanh_fp32_out[i, j] = cinn_cpu_tanh_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.tanh (29 ms)
+43: [ RUN      ] elementwise_pe.isfinite
+43: I0924 13:32:19.922298 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_isfinite_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_isfinite_fp32_out[i, j] = cinn_cpu_isfinite_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.isfinite (36 ms)
+43: [ RUN      ] elementwise_pe.isinf
+43: I0924 13:32:19.960331 27831 pe_elementwise_test.cc:33] func:
+43: function fn (_A, _PE_Elementwise_isinf_fp32_out)
+43: {
+43:   for (i, 100)
+43:   {
+43:     for (j, 32)
+43:     {
+43:       PE_Elementwise_isinf_fp32_out[i, j] = cinn_cpu_isinf_fp32(A[i, j])
+43:     }
+43:   }
+43: }
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+43: [       OK ] elementwise_pe.isinf (40 ms)
+43: [----------] 25 tests from elementwise_pe (792 ms total)
+43: 
+43: [----------] Global test environment tear-down
+43: [==========] 25 tests from 1 test case ran. (792 ms total)
+43: [  PASSED  ] 25 tests.
+56: [==========] Running 1 test from 1 test case.
+56: [----------] Global test environment set-up.
+56: [----------] 1 test from test03
+56: [ RUN      ] test03.basic
+56: [       OK ] test03.basic (0 ms)
+56: [----------] 1 test from test03 (0 ms total)
+56: 
+56: [----------] Global test environment tear-down
+56: [==========] 1 test from 1 test case ran. (0 ms total)
+56: [  PASSED  ] 1 test.
+54: [==========] Running 1 test from 1 test case.
+54: [----------] Global test environment set-up.
+54: [----------] 1 test from test02
+54: [ RUN      ] test02.basic
+54: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+54: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+54: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+54: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+54: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+52: [==========] Running 3 tests from 1 test case.
+52: [----------] Global test environment set-up.
+52: [----------] 3 tests from test01
+52: [ RUN      ] test01.basic
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+52: WARNING: Logging before InitGoogleLogging() is written to STDERR
+52: I0924 13:32:19.536278 27839 test01_elementwise_add_case.cc:30] test1 basic
+52: I0924 13:32:19.536424 27839 test01_elementwise_add_case.cc:34] test1 vectorize
+52: [       OK ] test01.basic (0 ms)
+52: [ RUN      ] test01.compute_at
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+52: I0924 13:32:19.536634 27839 test01_elementwise_add_case.cc:73] test1 basic
+52: I0924 13:32:19.536728 27839 test01_elementwise_add_case.cc:77] test1 vectorize
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128
+52: [       OK ] test01.compute_at (0 ms)
+52: [ RUN      ] test01.compute_at_level1
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+52: I0924 13:32:19.536936 27839 test01_elementwise_add_case.cc:120] test1 basic
+52: I0924 13:32:19.537029 27839 test01_elementwise_add_case.cc:124] test1 vectorize
+52: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4
+52: [       OK ] test01.compute_at_level1 (1 ms)
+52: [----------] 3 tests from test01 (1 ms total)
+52: 
+52: [----------] Global test environment tear-down
+52: [==========] 3 tests from 1 test case ran. (1 ms total)
+52: [  PASSED  ] 3 tests.
+51/68 Test #48: test_frontend_syntax ...................   Passed    0.34 sec
+52/68 Test #59: test_cinn_packed_func ..................   Passed    0.28 sec
+53/68 Test #58: test_cinn_common .......................   Passed    0.28 sec
+54/68 Test #43: test_pe_elementwise ....................   Passed    0.81 sec
+55/68 Test #56: test03_conv_case .......................   Passed    0.47 sec
+56/68 Test #52: test01_elementwise_add_case ............   Passed    0.58 sec
+test 61
+      Start 61: test_cinn_pe_reduction
+
+61: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_pe_reduction.py"
+61: Test timeout computed to be: 9.99988e+06
+test 62
+      Start 62: test_cinn_pe_transform
+
+62: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_pe_transform.py"
+62: Test timeout computed to be: 9.99988e+06
+test 63
+      Start 63: test_cinn_op_nn
+
+63: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_op_nn.py"
+63: Test timeout computed to be: 9.99988e+06
+test 64
+      Start 64: test_cinn_op_broadcast
+
+64: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_op_broadcast.py"
+64: Test timeout computed to be: 9.99988e+06
+test 65
+      Start 65: test_cinn_frontend
+
+65: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_frontend.py" "/home/wangyue50/CINN-my/CINN/build/thirds/naive_mul_model" "/home/wangyue50/CINN-my/CINN/build/thirds/multi_fc_model"
+65: Test timeout computed to be: 9.99988e+06
+test 66
+      Start 66: test_cinn_fake_resnet
+
+66: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_resnet.py" "/home/wangyue50/CINN-my/CINN/build/thirds/resnet_model"
+66: Test timeout computed to be: 9.99988e+06
+60: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+60: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+57: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+57: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+57: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+57: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+57: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register relu
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register relu6
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register conv2d
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register depthwise_conv2d
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register batchnorm
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register pool1d
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register pool2d
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register pool3d
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register sigmoid
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register softmax
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register slice
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register depthwise_conv2d
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register elementwise_add
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register elementwise_mul
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register scale
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register matmul
+62: I0100 00:00:00.000000 28051 registry.h:89] RAW: Register mul
+62: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 6400
+62: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 2048
+62: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+62: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 6400
+62: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 2048
+62: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+62: .
+62: ----------------------------------------------------------------------
+62: Ran 2 tests in 0.165s
+62: 
+62: OK
+62: function test_matmul (_x, _y, _T_Matmul_out)
+62: {
+62:   for (i, 100)
+62:   {
+62:     for (j, 32)
+62:     {
+62:       for (kk, 16)
+62:       {
+62:         T_Matmul_out[i, j] = (T_Matmul_out[i, j] + (x[i, kk] * y[kk, j]))
+62:       }
+62:     }
+62:   }
+62: }
+62: function test_matmul (_x, _y, _T_Matmul_out)
+62: {
+62:   for (i, 100)
+62:   {
+62:     for (j, 32)
+62:     {
+62:       for (kk_0, 16)
+62:       {
+62:         T_Matmul_out[i, j] = (T_Matmul_out[i, j] + (x[i, kk_0] * y[j, kk_0]))
+62:       }
+62:     }
+62:   }
+62: }
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register relu
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register relu6
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register conv2d
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register depthwise_conv2d
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register batchnorm
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register pool1d
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register pool2d
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register pool3d
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register sigmoid
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register softmax
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register slice
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register depthwise_conv2d
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register elementwise_add
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register elementwise_mul
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register scale
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register matmul
+61: I0100 00:00:00.000000 28046 registry.h:89] RAW: Register mul
+61: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+61: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4
+61: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+61: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4
+61: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+61: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128
+61: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+61: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128
+61: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+61: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128
+57/68 Test #47: test_op_nn .............................   Passed    1.01 sec
+58/68 Test #62: test_cinn_pe_transform .................   Passed    0.56 sec
+test 67
+      Start 67: test_cinn_real_resnet18
+
+67: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_resnet18.py" "/home/wangyue50/CINN-my/CINN/build/thirds/ResNet18"
+67: Test timeout computed to be: 9.99988e+06
+test 68
+      Start 68: test_cinn_real_mobilenetV2
+
+68: Test command: /usr/bin/cmake "-E" "env" "PYTHONPATH=/home/wangyue50/CINN-my/CINN/build/python:/home/wangyue50/CINN-my/CINN/python/tests/..:/home/wangyue50/CINN-my/CINN/build/python/cinn:/home/wangyue50/CINN-my/CINN" "python3" "/home/wangyue50/CINN-my/CINN/python/tests/test_mobilenetv2.py" "/home/wangyue50/CINN-my/CINN/build/thirds/MobileNetV2"
+68: Test timeout computed to be: 9.99988e+06
+60: .
+60: ----------------------------------------------------------------------
+60: Ran 1 test in 0.835s
+60: 
+60: OK
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register relu
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register relu6
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register conv2d
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register depthwise_conv2d
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register batchnorm
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register pool1d
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register pool2d
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register pool3d
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register sigmoid
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register softmax
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register slice
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register depthwise_conv2d
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register elementwise_add
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register elementwise_mul
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register scale
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register matmul
+64: I0100 00:00:00.000000 28048 registry.h:89] RAW: Register mul
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+64: WARNING:root:func:
+64: 
+64: function elementwise_add (_Var_1, _Var_2, _C)
+64: {
+64:   for (i, 100)
+64:   {
+64:     for (j, 32)
+64:     {
+64:       C[i, j] = (Var_1[i, j] + Var_2[i, j])
+64:     }
+64:   }
+64: }
+64: 
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+64: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 24
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 8
+64: WARNING:root:func:
+64: 
+64: function elementwise_add (_Var_1, _Var_2, _C_0)
+64: {
+64:   for (i, 3)
+64:   {
+64:     for (j, 2)
+64:     {
+64:       C_0[i, j] = (Var_1[i, j] + Var_2[j])
+64:     }
+64:   }
+64: }
+64: 
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 24
+64: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+64: WARNING:root:func:
+64: 
+64: function elementwise_mul (_Var_1, _Var_2, _C_1)
+64: {
+64:   for (i, 100)
+64:   {
+64:     for (j, 32)
+64:     {
+64:       C_1[i, j] = (Var_1[i, j] * Var_2[i, j])
+64:     }
+64:   }
+64: }
+64: 
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+64: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 24
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 8
+64: WARNING:root:func:
+64: 
+64: function elementwise_mul (_Var_1, _Var_2, _C_2)
+64: {
+64:   for (i, 3)
+64:   {
+64:     for (j, 2)
+64:     {
+64:       C_2[i, j] = (Var_1[i, j] * Var_2[j])
+64:     }
+64:   }
+64: }
+64: 
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 24
+64: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+64: WARNING:root:func:
+64: 
+64: function scale (_Var_1, _Scale_out)
+64: {
+64:   for (i, 100)
+64:   {
+64:     for (j, 32)
+64:     {
+64:       Scale_out[i, j] = (0.3 + (0.7 * Var_1[i, j]))
+64:     }
+64:   }
+64: }
+64: 
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+64: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+64: WARNING:root:func:
+64: 
+64: function scale (_Var_1, _Scale_out_0)
+64: {
+64:   for (i, 100)
+64:   {
+64:     for (j, 32)
+64:     {
+64:       Scale_out_0[i, j] = (0.24 + (0.6 * Var_1[i, j]))
+64:     }
+64:   }
+64: }
+64: 
+64: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 12800
+64: ..
+64: ----------------------------------------------------------------------
+64: Ran 7 tests in 0.302s
+64: 
+64: OK
+61: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+61: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128
+61: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+61: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128
+61: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4096
+61: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 128
+61: .
+61: ----------------------------------------------------------------------
+61: Ran 8 tests in 0.826s
+61: 
+61: OK
+61: function test_sum (_x, _T_sum_out)
+61: {
+61:   T_sum_out_init[0, 0] = 2
+61:   for (kk, 32)
+61:   {
+61:     for (kk_0, 32)
+61:     {
+61:       T_sum_out[0, 0] = (T_sum_out[0, 0] + x[kk, kk_0])
+61:     }
+61:   }
+61: }
+61: function test_prod (_x, _T_prod_out)
+61: {
+61:   T_prod_out_init[0, 0] = 2
+61:   for (kk_1, 32)
+61:   {
+61:     for (kk_2, 32)
+61:     {
+61:       T_prod_out[0, 0] = (T_prod_out[0, 0] * x[kk_1, kk_2])
+61:     }
+61:   }
+61: }
+61: function test_sum (_x, _T_sum_out)
+61: {
+61:   T_sum_out_init_0[0] = 2
+61:   for (kk_3, 32)
+61:   {
+61:     for (kk_4, 32)
+61:     {
+61:       T_sum_out[0] = (T_sum_out[0] + x[kk_3, kk_4])
+61:     }
+61:   }
+61: }
+61: function test_prod (_x, _T_prod_out)
+61: {
+61:   T_prod_out_init_0[0] = 2
+61:   for (kk_5, 32)
+61:   {
+61:     for (kk_6, 32)
+61:     {
+61:       T_prod_out[0] = (T_prod_out[0] * x[kk_5, kk_6])
+61:     }
+61:   }
+61: }
+61: function test_sum (_x, _T_sum_out)
+61: {
+61:   for (i, 32)
+61:   {
+61:     T_sum_out_init_1[i] = 2
+61:   }
+61:   for (i, 32)
+61:   {
+61:     for (kk_7, 32)
+61:     {
+61:       T_sum_out[i] = (T_sum_out[i] + x[kk_7, i])
+61:     }
+61:   }
+61: }
+61: function test_prod (_x, _T_prod_out)
+61: {
+61:   for (i, 32)
+61:   {
+61:     T_prod_out_init_1[i] = 2
+61:   }
+61:   for (i, 32)
+61:   {
+61:     for (kk_8, 32)
+61:     {
+61:       T_prod_out[i] = (T_prod_out[i] * x[kk_8, i])
+61:     }
+61:   }
+61: }
+61: function test_sum (_x, _T_sum_out)
+61: {
+61:   for (j, 32)
+61:   {
+61:     T_sum_out_init_2[0, j] = 2
+61:   }
+61:   for (j, 32)
+61:   {
+61:     for (kk_9, 32)
+61:     {
+61:       T_sum_out[0, j] = (T_sum_out[0, j] + x[kk_9, j])
+61:     }
+61:   }
+61: }
+61: function test_prod (_x, _T_prod_out)
+61: {
+61:   for (j, 32)
+61:   {
+61:     T_prod_out_init_2[0, j] = 2
+61:   }
+61:   for (j, 32)
+61:   {
+61:     for (kk_10, 32)
+61:     {
+61:       T_prod_out[0, j] = (T_prod_out[0, j] * x[kk_10, j])
+61:     }
+61:   }
+61: }
+61: function test_sum (_x, _T_sum_out)
+61: {
+61:   for (i, 32)
+61:   {
+61:     T_sum_out_init_3[i] = 2
+61:   }
+61:   for (i, 32)
+61:   {
+61:     for (kk_11, 32)
+61:     {
+61:       T_sum_out[i] = (T_sum_out[i] + x[i, kk_11])
+61:     }
+61:   }
+61: }
+61: function test_prod (_x, _T_prod_out)
+61: {
+61:   for (i, 32)
+61:   {
+61:     T_prod_out_init_3[i] = 2
+61:   }
+61:   for (i, 32)
+61:   {
+61:     for (kk_12, 32)
+61:     {
+61:       T_prod_out[i] = (T_prod_out[i] * x[i, kk_12])
+61:     }
+61:   }
+61: }
+61: function test_sum (_x, _T_sum_out)
+61: {
+61:   for (i, 32)
+61:   {
+61:     T_sum_out_init_4[i, 0] = 2
+61:   }
+61:   for (i, 32)
+61:   {
+61:     for (kk_13, 32)
+61:     {
+61:       T_sum_out[i, 0] = (T_sum_out[i, 0] + x[i, kk_13])
+61:     }
+61:   }
+61: }
+61: function test_prod (_x, _T_prod_out)
+61: {
+61:   for (i, 32)
+61:   {
+61:     T_prod_out_init_4[i, 0] = 2
+61:   }
+61:   for (i, 32)
+61:   {
+61:     for (kk_14, 32)
+61:     {
+61:       T_prod_out[i, 0] = (T_prod_out[i, 0] * x[i, kk_14])
+61:     }
+61:   }
+61: }
+61: function test_max (_x, _T_max_out)
+61: {
+61:   for (i, 32)
+61:   {
+61:     for (kk_15, 32)
+61:     {
+61:       T_max_out[i, 0] = cinn_max(T_max_out[i, 0], x[i, kk_15])
+61:     }
+61:   }
+61: }
+61: function test_min (_x, _T_min_out)
+61: {
+61:   for (i, 32)
+61:   {
+61:     for (kk_16, 32)
+61:     {
+61:       T_min_out[i, 0] = cinn_min(T_min_out[i, 0], x[i, kk_16])
+61:     }
+61:   }
+61: }
+61: function test_max (_x, _T_max_out)
+61: {
+61:   for (i, 32)
+61:   {
+61:     for (kk_17, 32)
+61:     {
+61:       T_max_out[i] = cinn_max(T_max_out[i], x[i, kk_17])
+61:     }
+61:   }
+61: }
+61: function test_min (_x, _T_min_out)
+61: {
+61:   for (i, 32)
+61:   {
+61:     for (kk_18, 32)
+61:     {
+61:       T_min_out[i] = cinn_min(T_min_out[i], x[i, kk_18])
+61:     }
+61:   }
+61: }
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register relu
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register relu6
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register conv2d
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register depthwise_conv2d
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register batchnorm
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register pool1d
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register pool2d
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register pool3d
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register sigmoid
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register softmax
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register slice
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register depthwise_conv2d
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register elementwise_add
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register elementwise_mul
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register scale
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register matmul
+66: I0100 00:00:00.000000 28050 registry.h:89] RAW: Register mul
+66: WARNING: Logging before InitGoogleLogging() is written to STDERR
+66: I0924 13:32:21.558060 28050 syntax.cc:130] Loading Paddle model from /home/wangyue50/CINN-my/CINN/build/thirds/resnet_model
+66: I0924 13:32:21.558151 28050 model_parser.cc:188] model_dir is: /home/wangyue50/CINN-my/CINN/build/thirds/resnet_model
+66: I0924 13:32:21.558156 28050 model_parser.cc:189] model_file is: __model__
+66: I0924 13:32:21.558158 28050 model_parser.cc:190] param_file is: 
+66: I0924 13:32:21.560164 28050 paddle_model_to_program.cc:26] detect model output: [save_infer_model/scale_0.tmp_0]
+66: I0924 13:32:21.560267 28050 executor.cc:49] Program:
+66: Program {
+66: var_3, var_4 = depthwise_conv2d(resnet_input, depthwise_conv2d_0__w_0, stride=[1,1], padding=[1,1], groups=32, dilation=[1,1])
+66: var_8 = elementwise_add(var_4, depthwise_conv2d_0__b_0, axis=1)
+66: var_10 = relu6(var_8)
+66: var_12 = relu(var_10)
+66: var_14 = scale(var_12, scale=1)
+66: }
+59/68 Test #60: test_cinn_pe_elementwise ...............   Passed    1.64 sec
+60/68 Test #64: test_cinn_op_broadcast .................   Passed    1.07 sec
+61/68 Test #61: test_cinn_pe_reduction .................   Passed    1.48 sec
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register relu
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register relu6
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register conv2d
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register depthwise_conv2d
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register batchnorm
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register pool1d
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register pool2d
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register pool3d
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register sigmoid
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register softmax
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register slice
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register depthwise_conv2d
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register elementwise_add
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register elementwise_mul
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register scale
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register matmul
+65: I0100 00:00:00.000000 28049 registry.h:89] RAW: Register mul
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register relu
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register relu6
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register conv2d
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register depthwise_conv2d
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register batchnorm
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register pool1d
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register pool2d
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register pool3d
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register sigmoid
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register softmax
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register slice
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register depthwise_conv2d
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register elementwise_add
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register elementwise_mul
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register scale
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register matmul
+63: I0100 00:00:00.000000 28047 registry.h:89] RAW: Register mul
+63: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 1200
+63: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 37632
+63: 2020-09-24 13:32:21,684-WARNING: func:
+63: 
+63: function conv2d (_Var_1, _Var_2, _T_pad_out_0, _weights_dilation, _Conv2d_nchw_out)
+63: {
+63:   for (j, 3)
+63:   {
+63:     for (k, 12)
+63:     {
+63:       for (a, 12)
+63:       {
+63:         T_pad_out_0[0, j, k, a] = select(((a < 11) and ((a >= 1) and ((k < 11) and (k >= 1)))), Var_1[0, j, (-1 + k), (-1 + a)], 0)
+63:       }
+63:     }
+63:   }
+63:   for (i, 64)
+63:   {
+63:     for (j, 3)
+63:     {
+63:       for (k, 13)
+63:       {
+63:         for (a, 13)
+63:         {
+63:           weights_dilation[i, j, k, a] = select((((a % 2) == 0) and ((k % 2) == 0)), Var_2[i, j, (k / 2), (a / 2)], 0)
+63:         }
+63:       }
+63:     }
+63:   }
+63:   for (j, 64)
+63:   {
+63:     for (fc, 3)
+63:     {
+63:       for (fy, 13)
+63:       {
+63:         for (fx, 13)
+63:         {
+63:           Conv2d_nchw_out[0, j, 0, 0] = (Conv2d_nchw_out[0, j, 0, 0] + (T_pad_out_0[0, fc, fy, fx] * weights_dilation[j, fc, fy, fx]))
+63:         }
+63:       }
+63:     }
+63:   }
+63: }
+63: 
+63: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 1728
+63: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 129792
+63: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 256
+63: ..
+63: ----------------------------------------------------------------------
+63: Ran 2 tests in 0.212s
+63: 
+63: OK
+63: output's shape is: (1, 1, 64, 1, 1)
+63: pad's shape is: [1, 3, 12, 12]
+66: I0924 13:32:21.720008 28050 graph_compiler.cc:39] [Debug] C Code is:
+66: #include <cinn_runtime.h>
+66: #include <stdio.h>
+66: 
+66: void fn_depthwise_conv2d_0(void* _args, int32_t num_args)
+66: {
+66:   const cinn_buffer_t* _resnet_input = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+66:   const cinn_buffer_t* _depthwise_conv2d_0__w_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+66:   cinn_buffer_t* _T_pad_out_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+66:   cinn_buffer_t* _T_depthwise_conv2d_nchw_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+66:   cinn_buffer_malloc((void*)(0), _T_pad_out_0);
+66:   cinn_buffer_malloc((void*)(0), _T_depthwise_conv2d_nchw_out);
+66:   float* T_depthwise_conv2d_nchw_out = ((float*)(_T_depthwise_conv2d_nchw_out->memory));
+66:   float* T_depthwise_conv2d_nchw_out_init = ((float*)(_T_depthwise_conv2d_nchw_out->memory));
+66:   float* T_pad_out_0 = ((float*)(_T_pad_out_0->memory));
+66:   const float* depthwise_conv2d_0__w_0 = ((const float*)(_depthwise_conv2d_0__w_0->memory));
+66:   const float* resnet_input = ((const float*)(_resnet_input->memory));
+66:   for (int32_t j = 0; j < 32; j += 1) {
+66:     for (int32_t k = 0; k < 112; k += 1) {
+66:       for (int32_t a = 0; a < 112; a += 1) {
+66:         T_depthwise_conv2d_nchw_out_init[((12544 * j) + ((112 * k) + a))] = 0;
+66:       };
+66:     };
+66:   };
+66:   for (int32_t j = 0; j < 32; j += 1) {
+66:     for (int32_t k = 0; k < 114; k += 1) {
+66:       for (int32_t a = 0; a < 114; a += 1) {
+66:         T_pad_out_0[((12996 * j) + ((114 * k) + a))] = ((((a < 113) && ((a >= 1) && ((k < 113) && (k >= 1))))) ? resnet_input[(-113 + ((12544 * j) + ((112 * k) + a)))] : 0);
+66:       };
+66:     };
+66:   };
+66:   for (int32_t j = 0; j < 32; j += 1) {
+66:     for (int32_t k = 0; k < 112; k += 1) {
+66:       for (int32_t a = 0; a < 112; a += 1) {
+66:         for (int32_t kh = 0; kh < 3; kh += 1) {
+66:           for (int32_t kw = 0; kw < 3; kw += 1) {
+66:             T_depthwise_conv2d_nchw_out[((12544 * j) + ((112 * k) + a))] = (T_depthwise_conv2d_nchw_out[((12544 * j) + ((112 * k) + a))] + (T_pad_out_0[(((j / 1) * 12996) + ((114 * k) + ((114 * kh) + (a + kw))))] * depthwise_conv2d_0__w_0[(((j / 1) * 9) + ((3 * kh) + kw))]));
+66:           };
+66:         };
+66:       };
+66:     };
+66:   };
+66:   cinn_buffer_free((void*)(0), _T_pad_out_0);
+66:   cinn_buffer_free((void*)(0), _T_depthwise_conv2d_nchw_out);
+66: }
+66: 
+66: void fn_elementwise_add_1(void* _args, int32_t num_args)
+66: {
+66:   const cinn_buffer_t* _var_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+66:   const cinn_buffer_t* _depthwise_conv2d_0__b_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+66:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+66:   cinn_buffer_malloc((void*)(0), _C);
+66:   float* C = ((float*)(_C->memory));
+66:   const float* depthwise_conv2d_0__b_0 = ((const float*)(_depthwise_conv2d_0__b_0->memory));
+66:   const float* var_4 = ((const float*)(_var_4->memory));
+66:   for (int32_t j = 0; j < 32; j += 1) {
+66:     for (int32_t k = 0; k < 112; k += 1) {
+66:       for (int32_t a = 0; a < 112; a += 1) {
+66:         C[((12544 * j) + ((112 * k) + a))] = (var_4[((12544 * j) + ((112 * k) + a))] + depthwise_conv2d_0__b_0[j]);
+66:       };
+66:     };
+66:   };
+66:   cinn_buffer_free((void*)(0), _C);
+66: }
+66: 
+66: void fn_relu6_2(void* _args, int32_t num_args)
+66: {
+66:   const cinn_buffer_t* _var_8 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+66:   cinn_buffer_t* _Relu6_output = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+66:   cinn_buffer_malloc((void*)(0), _Relu6_output);
+66:   float* Relu6_output = ((float*)(_Relu6_output->memory));
+66:   const float* var_8 = ((const float*)(_var_8->memory));
+66:   for (int32_t j = 0; j < 32; j += 1) {
+66:     for (int32_t k = 0; k < 112; k += 1) {
+66:       for (int32_t a = 0; a < 112; a += 1) {
+66:         Relu6_output[((12544 * j) + ((112 * k) + a))] = cinn_min(cinn_max(var_8[((12544 * j) + ((112 * k) + a))], 0), 6);
+66:       };
+66:     };
+66:   };
+66:   cinn_buffer_free((void*)(0), _Relu6_output);
+66: }
+66: 
+66: void fn_relu_3(void* _args, int32_t num_args)
+66: {
+66:   const cinn_buffer_t* _var_10 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+66:   cinn_buffer_t* _Relu_output = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+66:   cinn_buffer_malloc((void*)(0), _Relu_output);
+66:   float* Relu_output = ((float*)(_Relu_output->memory));
+66:   const float* var_10 = ((const float*)(_var_10->memory));
+66:   for (int32_t j = 0; j < 32; j += 1) {
+66:     for (int32_t k = 0; k < 112; k += 1) {
+66:       for (int32_t a = 0; a < 112; a += 1) {
+66:         Relu_output[((12544 * j) + ((112 * k) + a))] = cinn_max(var_10[((12544 * j) + ((112 * k) + a))], 0);
+66:       };
+66:     };
+66:   };
+66:   cinn_buffer_free((void*)(0), _Relu_output);
+66: }
+66: 
+66: void fn_scale_4(void* _args, int32_t num_args)
+66: {
+66:   const cinn_buffer_t* _var_12 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+66:   cinn_buffer_t* _Scale_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+66:   cinn_buffer_malloc((void*)(0), _Scale_out);
+66:   float* Scale_out = ((float*)(_Scale_out->memory));
+66:   const float* var_12 = ((const float*)(_var_12->memory));
+66:   for (int32_t j = 0; j < 32; j += 1) {
+66:     for (int32_t k = 0; k < 112; k += 1) {
+66:       for (int32_t a = 0; a < 112; a += 1) {
+66:         Scale_out[((12544 * j) + ((112 * k) + a))] = var_12[((12544 * j) + ((112 * k) + a))];
+66:       };
+66:     };
+66:   };
+66:   cinn_buffer_free((void*)(0), _Scale_out);
+66: }
+66: 
+66: .
+66: ----------------------------------------------------------------------
+66: Ran 1 test in 0.399s
+66: 
+66: OK
+66: result in test_model: 
+66: 
+62/68 Test #63: test_cinn_op_nn ........................   Passed    1.78 sec
+63/68 Test #66: test_cinn_fake_resnet ..................   Passed    2.08 sec
+65: WARNING: Logging before InitGoogleLogging() is written to STDERR
+65: I0924 13:32:21.836252 28049 graph_compiler.cc:39] [Debug] C Code is:
+65: #include <cinn_runtime.h>
+65: #include <stdio.h>
+65: 
+65: void fn_elementwise_add_0(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _C);
+65:   const float* A = ((const float*)(_A->memory));
+65:   const float* B = ((const float*)(_B->memory));
+65:   float* C = ((float*)(_C->memory));
+65:   for (int32_t j = 0; j < 24; j += 1) {
+65:     for (int32_t k = 0; k < 56; k += 1) {
+65:       for (int32_t a = 0; a < 56; a += 1) {
+65:         C[((3136 * j) + ((56 * k) + a))] = (A[((3136 * j) + ((56 * k) + a))] + B[((3136 * j) + ((56 * k) + a))]);
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _C);
+65: }
+65: 
+65: void fn_relu_1(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Relu_output = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Relu_output);
+65:   float* Relu_output = ((float*)(_Relu_output->memory));
+65:   const float* var = ((const float*)(_var->memory));
+65:   for (int32_t j = 0; j < 24; j += 1) {
+65:     for (int32_t k = 0; k < 56; k += 1) {
+65:       for (int32_t a = 0; a < 56; a += 1) {
+65:         Relu_output[((3136 * j) + ((56 * k) + a))] = cinn_max(var[((3136 * j) + ((56 * k) + a))], 0);
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Relu_output);
+65: }
+65: 
+65: void fn_conv2d_2(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _E = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _T_Identity_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_t* _weights_dilation = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+65:   cinn_buffer_t* _Conv2d_nchw_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+65:   cinn_buffer_malloc((void*)(0), _T_Identity_out);
+65:   cinn_buffer_malloc((void*)(0), _weights_dilation);
+65:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out);
+65:   float* Conv2d_nchw_out = ((float*)(_Conv2d_nchw_out->memory));
+65:   float* Conv2d_nchw_out_init = ((float*)(_Conv2d_nchw_out->memory));
+65:   const float* E = ((const float*)(_E->memory));
+65:   float* T_Identity_out = ((float*)(_T_Identity_out->memory));
+65:   const float* var_0 = ((const float*)(_var_0->memory));
+65:   float* weights_dilation = ((float*)(_weights_dilation->memory));
+65:   for (int32_t j = 0; j < 144; j += 1) {
+65:     for (int32_t k = 0; k < 56; k += 1) {
+65:       for (int32_t a = 0; a < 56; a += 1) {
+65:         Conv2d_nchw_out_init[((3136 * j) + ((56 * k) + a))] = 0;
+65:       };
+65:     };
+65:   };
+65:   for (int32_t i = 0; i < 144; i += 1) {
+65:     for (int32_t j = 0; j < 24; j += 1) {
+65:       weights_dilation[((24 * i) + j)] = (((((0 % 1) == 0) && ((0 % 1) == 0))) ? E[((24 * i) + j)] : 0);
+65:     };
+65:   };
+65:   for (int32_t j = 0; j < 24; j += 1) {
+65:     for (int32_t k = 0; k < 56; k += 1) {
+65:       for (int32_t a = 0; a < 56; a += 1) {
+65:         T_Identity_out[((3136 * j) + ((56 * k) + a))] = var_0[((3136 * j) + ((56 * k) + a))];
+65:       };
+65:     };
+65:   };
+65:   for (int32_t j = 0; j < 144; j += 1) {
+65:     for (int32_t k = 0; k < 56; k += 1) {
+65:       for (int32_t a = 0; a < 56; a += 1) {
+65:         for (int32_t fc = 0; fc < 24; fc += 1) {
+65:           Conv2d_nchw_out[((3136 * j) + ((56 * k) + a))] = (Conv2d_nchw_out[((3136 * j) + ((56 * k) + a))] + (T_Identity_out[((3136 * fc) + ((56 * k) + a))] * weights_dilation[((24 * j) + fc)]));
+65:         };
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _T_Identity_out);
+65:   cinn_buffer_free((void*)(0), _weights_dilation);
+65:   cinn_buffer_free((void*)(0), _Conv2d_nchw_out);
+65: }
+65: 
+65: void fn_scale_3(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Scale_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Scale_out);
+65:   float* Scale_out = ((float*)(_Scale_out->memory));
+65:   const float* var_3 = ((const float*)(_var_3->memory));
+65:   for (int32_t j = 0; j < 144; j += 1) {
+65:     for (int32_t k = 0; k < 56; k += 1) {
+65:       for (int32_t a = 0; a < 56; a += 1) {
+65:         Scale_out[((3136 * j) + ((56 * k) + a))] = (0.5 + (2 * var_3[((3136 * j) + ((56 * k) + a))]));
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Scale_out);
+65: }
+65: 
+65: void fn_softmax_4(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _softmax_temp_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _softmax_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _softmax_temp_out);
+65:   cinn_buffer_malloc((void*)(0), _softmax_out);
+65:   float* softmax_out = ((float*)(_softmax_out->memory));
+65:   float* softmax_temp_out = ((float*)(_softmax_temp_out->memory));
+65:   float* softmax_temp_out_init = ((float*)(_softmax_temp_out->memory));
+65:   const float* var_4 = ((const float*)(_var_4->memory));
+65:   for (int32_t j = 0; j < 144; j += 1) {
+65:     for (int32_t k = 0; k < 56; k += 1) {
+65:       for (int32_t a = 0; a < 56; a += 1) {
+65:         softmax_temp_out_init[((3136 * j) + ((56 * k) + a))] = 0;
+65:       };
+65:     };
+65:   };
+65:   for (int32_t j = 0; j < 144; j += 1) {
+65:     for (int32_t k = 0; k < 56; k += 1) {
+65:       for (int32_t a = 0; a < 56; a += 1) {
+65:         for (int32_t axis_j = 0; axis_j < 144; axis_j += 1) {
+65:           softmax_temp_out[((3136 * j) + ((56 * k) + a))] = (softmax_temp_out[((3136 * j) + ((56 * k) + a))] + cinn_cpu_exp_fp32(var_4[((3136 * axis_j) + ((56 * k) + a))]));
+65:         };
+65:       };
+65:     };
+65:   };
+65:   for (int32_t j = 0; j < 144; j += 1) {
+65:     for (int32_t k = 0; k < 56; k += 1) {
+65:       for (int32_t a = 0; a < 56; a += 1) {
+65:         softmax_out[((3136 * j) + ((56 * k) + a))] = (cinn_cpu_exp_fp32(var_4[((3136 * j) + ((56 * k) + a))]) * (1 / softmax_temp_out[((3136 * j) + ((56 * k) + a))]));
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _softmax_temp_out);
+65:   cinn_buffer_free((void*)(0), _softmax_out);
+65: }
+65: 
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register relu
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register relu6
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register conv2d
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register depthwise_conv2d
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register batchnorm
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register pool1d
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register pool2d
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register pool3d
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register sigmoid
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register softmax
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register slice
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register depthwise_conv2d
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register elementwise_add
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register elementwise_mul
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register scale
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register matmul
+68: I0100 00:00:00.000000 28814 registry.h:89] RAW: Register mul
+68: WARNING: Logging before InitGoogleLogging() is written to STDERR
+68: I0924 13:32:22.196000 28814 syntax.cc:130] Loading Paddle model from /home/wangyue50/CINN-my/CINN/build/thirds/MobileNetV2
+68: I0924 13:32:22.196074 28814 model_parser.cc:188] model_dir is: /home/wangyue50/CINN-my/CINN/build/thirds/MobileNetV2
+68: I0924 13:32:22.196079 28814 model_parser.cc:189] model_file is: __model__
+68: I0924 13:32:22.196080 28814 model_parser.cc:190] param_file is: 
+68: I0924 13:32:22.255563 28814 paddle_model_to_program.cc:26] detect model output: [save_infer_model/scale_0]
+68: I0924 13:32:22.257247 28814 executor.cc:49] Program:
+68: Program {
+68: var_3, var_4, var_5 = conv2d(image, conv1_1_weights, stride=[2,2], padding=[1,1], groups=1, dilation=[1,1])
+68: var_15 = batchnorm(var_5, conv1_1_bn_scale, conv1_1_bn_offset, conv1_1_bn_mean, conv1_1_bn_variance, epsilon=1e-05)
+68: var_17 = relu6(var_15)
+68: var_21, var_22, var_23 = conv2d(var_17, conv2_1_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_33 = batchnorm(var_23, conv2_1_expand_bn_scale, conv2_1_expand_bn_offset, conv2_1_expand_bn_mean, conv2_1_expand_bn_variance, epsilon=1e-05)
+68: var_35 = relu6(var_33)
+68: var_39, var_40 = depthwise_conv2d(var_35, conv2_1_dwise_weights, stride=[1,1], padding=[1,1], groups=32, dilation=[1,1])
+68: var_50 = batchnorm(var_40, conv2_1_dwise_bn_scale, conv2_1_dwise_bn_offset, conv2_1_dwise_bn_mean, conv2_1_dwise_bn_variance, epsilon=1e-05)
+68: var_52 = relu6(var_50)
+68: var_56, var_57, var_58 = conv2d(var_52, conv2_1_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_68 = batchnorm(var_58, conv2_1_linear_bn_scale, conv2_1_linear_bn_offset, conv2_1_linear_bn_mean, conv2_1_linear_bn_variance, epsilon=1e-05)
+68: var_72, var_73, var_74 = conv2d(var_68, conv3_1_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_84 = batchnorm(var_74, conv3_1_expand_bn_scale, conv3_1_expand_bn_offset, conv3_1_expand_bn_mean, conv3_1_expand_bn_variance, epsilon=1e-05)
+68: var_86 = relu6(var_84)
+68: var_90, var_91 = depthwise_conv2d(var_86, conv3_1_dwise_weights, stride=[2,2], padding=[1,1], groups=96, dilation=[1,1])
+68: var_101 = batchnorm(var_91, conv3_1_dwise_bn_scale, conv3_1_dwise_bn_offset, conv3_1_dwise_bn_mean, conv3_1_dwise_bn_variance, epsilon=1e-05)
+68: var_103 = relu6(var_101)
+68: var_107, var_108, var_109 = conv2d(var_103, conv3_1_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_119 = batchnorm(var_109, conv3_1_linear_bn_scale, conv3_1_linear_bn_offset, conv3_1_linear_bn_mean, conv3_1_linear_bn_variance, epsilon=1e-05)
+68: var_123, var_124, var_125 = conv2d(var_119, conv3_2_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_135 = batchnorm(var_125, conv3_2_expand_bn_scale, conv3_2_expand_bn_offset, conv3_2_expand_bn_mean, conv3_2_expand_bn_variance, epsilon=1e-05)
+68: var_137 = relu6(var_135)
+68: var_141, var_142 = depthwise_conv2d(var_137, conv3_2_dwise_weights, stride=[1,1], padding=[1,1], groups=144, dilation=[1,1])
+68: var_152 = batchnorm(var_142, conv3_2_dwise_bn_scale, conv3_2_dwise_bn_offset, conv3_2_dwise_bn_mean, conv3_2_dwise_bn_variance, epsilon=1e-05)
+68: var_154 = relu6(var_152)
+68: var_158, var_159, var_160 = conv2d(var_154, conv3_2_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_170 = batchnorm(var_160, conv3_2_linear_bn_scale, conv3_2_linear_bn_offset, conv3_2_linear_bn_mean, conv3_2_linear_bn_variance, epsilon=1e-05)
+68: var_172 = elementwise_add(var_119, var_170, axis=-1)
+68: var_176, var_177, var_178 = conv2d(var_172, conv4_1_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_188 = batchnorm(var_178, conv4_1_expand_bn_scale, conv4_1_expand_bn_offset, conv4_1_expand_bn_mean, conv4_1_expand_bn_variance, epsilon=1e-05)
+68: var_190 = relu6(var_188)
+68: var_194, var_195 = depthwise_conv2d(var_190, conv4_1_dwise_weights, stride=[2,2], padding=[1,1], groups=144, dilation=[1,1])
+68: var_205 = batchnorm(var_195, conv4_1_dwise_bn_scale, conv4_1_dwise_bn_offset, conv4_1_dwise_bn_mean, conv4_1_dwise_bn_variance, epsilon=1e-05)
+68: var_207 = relu6(var_205)
+68: var_211, var_212, var_213 = conv2d(var_207, conv4_1_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_223 = batchnorm(var_213, conv4_1_linear_bn_scale, conv4_1_linear_bn_offset, conv4_1_linear_bn_mean, conv4_1_linear_bn_variance, epsilon=1e-05)
+68: var_227, var_228, var_229 = conv2d(var_223, conv4_2_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_239 = batchnorm(var_229, conv4_2_expand_bn_scale, conv4_2_expand_bn_offset, conv4_2_expand_bn_mean, conv4_2_expand_bn_variance, epsilon=1e-05)
+68: var_241 = relu6(var_239)
+68: var_245, var_246 = depthwise_conv2d(var_241, conv4_2_dwise_weights, stride=[1,1], padding=[1,1], groups=192, dilation=[1,1])
+68: var_256 = batchnorm(var_246, conv4_2_dwise_bn_scale, conv4_2_dwise_bn_offset, conv4_2_dwise_bn_mean, conv4_2_dwise_bn_variance, epsilon=1e-05)
+68: var_258 = relu6(var_256)
+68: var_262, var_263, var_264 = conv2d(var_258, conv4_2_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_274 = batchnorm(var_264, conv4_2_linear_bn_scale, conv4_2_linear_bn_offset, conv4_2_linear_bn_mean, conv4_2_linear_bn_variance, epsilon=1e-05)
+68: var_276 = elementwise_add(var_223, var_274, axis=-1)
+68: var_280, var_281, var_282 = conv2d(var_276, conv4_3_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_292 = batchnorm(var_282, conv4_3_expand_bn_scale, conv4_3_expand_bn_offset, conv4_3_expand_bn_mean, conv4_3_expand_bn_variance, epsilon=1e-05)
+68: var_294 = relu6(var_292)
+68: var_298, var_299 = depthwise_conv2d(var_294, conv4_3_dwise_weights, stride=[1,1], padding=[1,1], groups=192, dilation=[1,1])
+68: var_309 = batchnorm(var_299, conv4_3_dwise_bn_scale, conv4_3_dwise_bn_offset, conv4_3_dwise_bn_mean, conv4_3_dwise_bn_variance, epsilon=1e-05)
+68: var_311 = relu6(var_309)
+68: var_315, var_316, var_317 = conv2d(var_311, conv4_3_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_327 = batchnorm(var_317, conv4_3_linear_bn_scale, conv4_3_linear_bn_offset, conv4_3_linear_bn_mean, conv4_3_linear_bn_variance, epsilon=1e-05)
+68: var_329 = elementwise_add(var_276, var_327, axis=-1)
+68: var_333, var_334, var_335 = conv2d(var_329, conv5_1_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_345 = batchnorm(var_335, conv5_1_expand_bn_scale, conv5_1_expand_bn_offset, conv5_1_expand_bn_mean, conv5_1_expand_bn_variance, epsilon=1e-05)
+68: var_347 = relu6(var_345)
+68: var_351, var_352 = depthwise_conv2d(var_347, conv5_1_dwise_weights, stride=[2,2], padding=[1,1], groups=192, dilation=[1,1])
+68: var_362 = batchnorm(var_352, conv5_1_dwise_bn_scale, conv5_1_dwise_bn_offset, conv5_1_dwise_bn_mean, conv5_1_dwise_bn_variance, epsilon=1e-05)
+68: var_364 = relu6(var_362)
+68: var_368, var_369, var_370 = conv2d(var_364, conv5_1_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_380 = batchnorm(var_370, conv5_1_linear_bn_scale, conv5_1_linear_bn_offset, conv5_1_linear_bn_mean, conv5_1_linear_bn_variance, epsilon=1e-05)
+68: var_384, var_385, var_386 = conv2d(var_380, conv5_2_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_396 = batchnorm(var_386, conv5_2_expand_bn_scale, conv5_2_expand_bn_offset, conv5_2_expand_bn_mean, conv5_2_expand_bn_variance, epsilon=1e-05)
+68: var_398 = relu6(var_396)
+68: var_402, var_403 = depthwise_conv2d(var_398, conv5_2_dwise_weights, stride=[1,1], padding=[1,1], groups=384, dilation=[1,1])
+68: var_413 = batchnorm(var_403, conv5_2_dwise_bn_scale, conv5_2_dwise_bn_offset, conv5_2_dwise_bn_mean, conv5_2_dwise_bn_variance, epsilon=1e-05)
+68: var_415 = relu6(var_413)
+68: var_419, var_420, var_421 = conv2d(var_415, conv5_2_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_431 = batchnorm(var_421, conv5_2_linear_bn_scale, conv5_2_linear_bn_offset, conv5_2_linear_bn_mean, conv5_2_linear_bn_variance, epsilon=1e-05)
+68: var_433 = elementwise_add(var_380, var_431, axis=-1)
+68: var_437, var_438, var_439 = conv2d(var_433, conv5_3_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_449 = batchnorm(var_439, conv5_3_expand_bn_scale, conv5_3_expand_bn_offset, conv5_3_expand_bn_mean, conv5_3_expand_bn_variance, epsilon=1e-05)
+68: var_451 = relu6(var_449)
+68: var_455, var_456 = depthwise_conv2d(var_451, conv5_3_dwise_weights, stride=[1,1], padding=[1,1], groups=384, dilation=[1,1])
+68: var_466 = batchnorm(var_456, conv5_3_dwise_bn_scale, conv5_3_dwise_bn_offset, conv5_3_dwise_bn_mean, conv5_3_dwise_bn_variance, epsilon=1e-05)
+68: var_468 = relu6(var_466)
+68: var_472, var_473, var_474 = conv2d(var_468, conv5_3_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_484 = batchnorm(var_474, conv5_3_linear_bn_scale, conv5_3_linear_bn_offset, conv5_3_linear_bn_mean, conv5_3_linear_bn_variance, epsilon=1e-05)
+68: var_486 = elementwise_add(var_433, var_484, axis=-1)
+68: var_490, var_491, var_492 = conv2d(var_486, conv5_4_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_502 = batchnorm(var_492, conv5_4_expand_bn_scale, conv5_4_expand_bn_offset, conv5_4_expand_bn_mean, conv5_4_expand_bn_variance, epsilon=1e-05)
+68: var_504 = relu6(var_502)
+68: var_508, var_509 = depthwise_conv2d(var_504, conv5_4_dwise_weights, stride=[1,1], padding=[1,1], groups=384, dilation=[1,1])
+68: var_519 = batchnorm(var_509, conv5_4_dwise_bn_scale, conv5_4_dwise_bn_offset, conv5_4_dwise_bn_mean, conv5_4_dwise_bn_variance, epsilon=1e-05)
+68: var_521 = relu6(var_519)
+68: var_525, var_526, var_527 = conv2d(var_521, conv5_4_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_537 = batchnorm(var_527, conv5_4_linear_bn_scale, conv5_4_linear_bn_offset, conv5_4_linear_bn_mean, conv5_4_linear_bn_variance, epsilon=1e-05)
+68: var_539 = elementwise_add(var_486, var_537, axis=-1)
+68: var_543, var_544, var_545 = conv2d(var_539, conv6_1_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_555 = batchnorm(var_545, conv6_1_expand_bn_scale, conv6_1_expand_bn_offset, conv6_1_expand_bn_mean, conv6_1_expand_bn_variance, epsilon=1e-05)
+68: var_557 = relu6(var_555)
+68: var_561, var_562 = depthwise_conv2d(var_557, conv6_1_dwise_weights, stride=[1,1], padding=[1,1], groups=384, dilation=[1,1])
+68: var_572 = batchnorm(var_562, conv6_1_dwise_bn_scale, conv6_1_dwise_bn_offset, conv6_1_dwise_bn_mean, conv6_1_dwise_bn_variance, epsilon=1e-05)
+68: var_574 = relu6(var_572)
+68: var_578, var_579, var_580 = conv2d(var_574, conv6_1_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_590 = batchnorm(var_580, conv6_1_linear_bn_scale, conv6_1_linear_bn_offset, conv6_1_linear_bn_mean, conv6_1_linear_bn_variance, epsilon=1e-05)
+68: var_594, var_595, var_596 = conv2d(var_590, conv6_2_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_606 = batchnorm(var_596, conv6_2_expand_bn_scale, conv6_2_expand_bn_offset, conv6_2_expand_bn_mean, conv6_2_expand_bn_variance, epsilon=1e-05)
+68: var_608 = relu6(var_606)
+68: var_612, var_613 = depthwise_conv2d(var_608, conv6_2_dwise_weights, stride=[1,1], padding=[1,1], groups=576, dilation=[1,1])
+68: var_623 = batchnorm(var_613, conv6_2_dwise_bn_scale, conv6_2_dwise_bn_offset, conv6_2_dwise_bn_mean, conv6_2_dwise_bn_variance, epsilon=1e-05)
+68: var_625 = relu6(var_623)
+68: var_629, var_630, var_631 = conv2d(var_625, conv6_2_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_641 = batchnorm(var_631, conv6_2_linear_bn_scale, conv6_2_linear_bn_offset, conv6_2_linear_bn_mean, conv6_2_linear_bn_variance, epsilon=1e-05)
+68: var_643 = elementwise_add(var_590, var_641, axis=-1)
+68: var_647, var_648, var_649 = conv2d(var_643, conv6_3_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_659 = batchnorm(var_649, conv6_3_expand_bn_scale, conv6_3_expand_bn_offset, conv6_3_expand_bn_mean, conv6_3_expand_bn_variance, epsilon=1e-05)
+68: var_661 = relu6(var_659)
+68: var_665, var_666 = depthwise_conv2d(var_661, conv6_3_dwise_weights, stride=[1,1], padding=[1,1], groups=576, dilation=[1,1])
+68: var_676 = batchnorm(var_666, conv6_3_dwise_bn_scale, conv6_3_dwise_bn_offset, conv6_3_dwise_bn_mean, conv6_3_dwise_bn_variance, epsilon=1e-05)
+68: var_678 = relu6(var_676)
+68: var_682, var_683, var_684 = conv2d(var_678, conv6_3_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_694 = batchnorm(var_684, conv6_3_linear_bn_scale, conv6_3_linear_bn_offset, conv6_3_linear_bn_mean, conv6_3_linear_bn_variance, epsilon=1e-05)
+68: var_696 = elementwise_add(var_643, var_694, axis=-1)
+68: var_700, var_701, var_702 = conv2d(var_696, conv7_1_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_712 = batchnorm(var_702, conv7_1_expand_bn_scale, conv7_1_expand_bn_offset, conv7_1_expand_bn_mean, conv7_1_expand_bn_variance, epsilon=1e-05)
+68: var_714 = relu6(var_712)
+68: var_718, var_719 = depthwise_conv2d(var_714, conv7_1_dwise_weights, stride=[2,2], padding=[1,1], groups=576, dilation=[1,1])
+68: var_729 = batchnorm(var_719, conv7_1_dwise_bn_scale, conv7_1_dwise_bn_offset, conv7_1_dwise_bn_mean, conv7_1_dwise_bn_variance, epsilon=1e-05)
+68: var_731 = relu6(var_729)
+68: var_735, var_736, var_737 = conv2d(var_731, conv7_1_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_747 = batchnorm(var_737, conv7_1_linear_bn_scale, conv7_1_linear_bn_offset, conv7_1_linear_bn_mean, conv7_1_linear_bn_variance, epsilon=1e-05)
+68: var_751, var_752, var_753 = conv2d(var_747, conv7_2_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_763 = batchnorm(var_753, conv7_2_expand_bn_scale, conv7_2_expand_bn_offset, conv7_2_expand_bn_mean, conv7_2_expand_bn_variance, epsilon=1e-05)
+68: var_765 = relu6(var_763)
+68: var_769, var_770 = depthwise_conv2d(var_765, conv7_2_dwise_weights, stride=[1,1], padding=[1,1], groups=960, dilation=[1,1])
+68: var_780 = batchnorm(var_770, conv7_2_dwise_bn_scale, conv7_2_dwise_bn_offset, conv7_2_dwise_bn_mean, conv7_2_dwise_bn_variance, epsilon=1e-05)
+68: var_782 = relu6(var_780)
+68: var_786, var_787, var_788 = conv2d(var_782, conv7_2_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_798 = batchnorm(var_788, conv7_2_linear_bn_scale, conv7_2_linear_bn_offset, conv7_2_linear_bn_mean, conv7_2_linear_bn_variance, epsilon=1e-05)
+68: var_800 = elementwise_add(var_747, var_798, axis=-1)
+68: var_804, var_805, var_806 = conv2d(var_800, conv7_3_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_816 = batchnorm(var_806, conv7_3_expand_bn_scale, conv7_3_expand_bn_offset, conv7_3_expand_bn_mean, conv7_3_expand_bn_variance, epsilon=1e-05)
+68: var_818 = relu6(var_816)
+68: var_822, var_823 = depthwise_conv2d(var_818, conv7_3_dwise_weights, stride=[1,1], padding=[1,1], groups=960, dilation=[1,1])
+68: var_833 = batchnorm(var_823, conv7_3_dwise_bn_scale, conv7_3_dwise_bn_offset, conv7_3_dwise_bn_mean, conv7_3_dwise_bn_variance, epsilon=1e-05)
+68: var_835 = relu6(var_833)
+68: var_839, var_840, var_841 = conv2d(var_835, conv7_3_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_851 = batchnorm(var_841, conv7_3_linear_bn_scale, conv7_3_linear_bn_offset, conv7_3_linear_bn_mean, conv7_3_linear_bn_variance, epsilon=1e-05)
+68: var_853 = elementwise_add(var_800, var_851, axis=-1)
+68: var_857, var_858, var_859 = conv2d(var_853, conv8_1_expand_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_869 = batchnorm(var_859, conv8_1_expand_bn_scale, conv8_1_expand_bn_offset, conv8_1_expand_bn_mean, conv8_1_expand_bn_variance, epsilon=1e-05)
+68: var_871 = relu6(var_869)
+68: var_875, var_876 = depthwise_conv2d(var_871, conv8_1_dwise_weights, stride=[1,1], padding=[1,1], groups=960, dilation=[1,1])
+68: var_886 = batchnorm(var_876, conv8_1_dwise_bn_scale, conv8_1_dwise_bn_offset, conv8_1_dwise_bn_mean, conv8_1_dwise_bn_variance, epsilon=1e-05)
+68: var_888 = relu6(var_886)
+68: var_892, var_893, var_894 = conv2d(var_888, conv8_1_linear_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_904 = batchnorm(var_894, conv8_1_linear_bn_scale, conv8_1_linear_bn_offset, conv8_1_linear_bn_mean, conv8_1_linear_bn_variance, epsilon=1e-05)
+68: var_908, var_909, var_910 = conv2d(var_904, conv9_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+68: var_920 = batchnorm(var_910, conv9_bn_scale, conv9_bn_offset, conv9_bn_mean, conv9_bn_variance, epsilon=1e-05)
+68: var_922 = relu6(var_920)
+68: var_924, var_925 = pool2d(var_922, pool_type=avg, stride_size=[1,1], exclusive=true, kernel_size=[7,7], data_format=AnyLayout, padding_size=[0,0,0,0], ceil_mode=false)
+68: var_929 = mul(var_925, fc10_weights, y_num_col_dims=1, x_num_col_dims=1)
+68: var_933 = elementwise_add(var_929, fc10_offset, axis=1)
+68: var_935, var_936 = softmax(var_933, axis=-1)
+68: var_938 = scale(var_936, scale=1)
+68: }
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register relu
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register relu6
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register conv2d
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register depthwise_conv2d
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register batchnorm
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register pool1d
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register pool2d
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register pool3d
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register sigmoid
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register softmax
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register slice
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register depthwise_conv2d
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register elementwise_add
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register elementwise_mul
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register scale
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register matmul
+67: I0100 00:00:00.000000 28813 registry.h:89] RAW: Register mul
+67: WARNING: Logging before InitGoogleLogging() is written to STDERR
+67: I0924 13:32:22.166492 28813 syntax.cc:130] Loading Paddle model from /home/wangyue50/CINN-my/CINN/build/thirds/ResNet18
+67: I0924 13:32:22.166563 28813 model_parser.cc:188] model_dir is: /home/wangyue50/CINN-my/CINN/build/thirds/ResNet18
+67: I0924 13:32:22.166568 28813 model_parser.cc:189] model_file is: __model__
+67: I0924 13:32:22.166569 28813 model_parser.cc:190] param_file is: 
+67: I0924 13:32:22.216248 28813 paddle_model_to_program.cc:26] detect model output: [save_infer_model/scale_0]
+67: I0924 13:32:22.217094 28813 executor.cc:49] Program:
+67: Program {
+67: var_3, var_4, var_5 = conv2d(image, conv1_weights, stride=[2,2], padding=[3,3], groups=1, dilation=[1,1])
+67: var_15 = batchnorm(var_5, bn_conv1_scale, bn_conv1_offset, bn_conv1_mean, bn_conv1_variance, epsilon=1e-05)
+67: var_17 = relu(var_15)
+67: var_19, var_20 = pool2d(var_17, pool_type=max, stride_size=[2,2], exclusive=true, kernel_size=[3,3], data_format=AnyLayout, padding_size=[1,1,1,1], ceil_mode=false)
+67: var_24, var_25, var_26 = conv2d(var_20, res2a_branch2a_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_36 = batchnorm(var_26, bn2a_branch2a_scale, bn2a_branch2a_offset, bn2a_branch2a_mean, bn2a_branch2a_variance, epsilon=1e-05)
+67: var_38 = relu(var_36)
+67: var_42, var_43, var_44 = conv2d(var_38, res2a_branch2b_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_54 = batchnorm(var_44, bn2a_branch2b_scale, bn2a_branch2b_offset, bn2a_branch2b_mean, bn2a_branch2b_variance, epsilon=1e-05)
+67: var_58, var_59, var_60 = conv2d(var_20, res2a_branch1_weights, stride=[1,1], padding=[0,0], groups=1, dilation=[1,1])
+67: var_70 = batchnorm(var_60, bn2a_branch1_scale, bn2a_branch1_offset, bn2a_branch1_mean, bn2a_branch1_variance, epsilon=1e-05)
+67: var_72 = elementwise_add(var_70, var_54, axis=-1)
+67: var_74 = relu(var_72)
+67: var_78, var_79, var_80 = conv2d(var_74, res2b_branch2a_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_90 = batchnorm(var_80, bn2b_branch2a_scale, bn2b_branch2a_offset, bn2b_branch2a_mean, bn2b_branch2a_variance, epsilon=1e-05)
+67: var_92 = relu(var_90)
+67: var_96, var_97, var_98 = conv2d(var_92, res2b_branch2b_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_108 = batchnorm(var_98, bn2b_branch2b_scale, bn2b_branch2b_offset, bn2b_branch2b_mean, bn2b_branch2b_variance, epsilon=1e-05)
+67: var_110 = elementwise_add(var_74, var_108, axis=-1)
+67: var_112 = relu(var_110)
+67: var_116, var_117, var_118 = conv2d(var_112, res3a_branch2a_weights, stride=[2,2], padding=[1,1], groups=1, dilation=[1,1])
+67: var_128 = batchnorm(var_118, bn3a_branch2a_scale, bn3a_branch2a_offset, bn3a_branch2a_mean, bn3a_branch2a_variance, epsilon=1e-05)
+67: var_130 = relu(var_128)
+67: var_134, var_135, var_136 = conv2d(var_130, res3a_branch2b_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_146 = batchnorm(var_136, bn3a_branch2b_scale, bn3a_branch2b_offset, bn3a_branch2b_mean, bn3a_branch2b_variance, epsilon=1e-05)
+67: var_150, var_151, var_152 = conv2d(var_112, res3a_branch1_weights, stride=[2,2], padding=[0,0], groups=1, dilation=[1,1])
+67: var_162 = batchnorm(var_152, bn3a_branch1_scale, bn3a_branch1_offset, bn3a_branch1_mean, bn3a_branch1_variance, epsilon=1e-05)
+67: var_164 = elementwise_add(var_162, var_146, axis=-1)
+67: var_166 = relu(var_164)
+67: var_170, var_171, var_172 = conv2d(var_166, res3b_branch2a_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_182 = batchnorm(var_172, bn3b_branch2a_scale, bn3b_branch2a_offset, bn3b_branch2a_mean, bn3b_branch2a_variance, epsilon=1e-05)
+67: var_184 = relu(var_182)
+67: var_188, var_189, var_190 = conv2d(var_184, res3b_branch2b_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_200 = batchnorm(var_190, bn3b_branch2b_scale, bn3b_branch2b_offset, bn3b_branch2b_mean, bn3b_branch2b_variance, epsilon=1e-05)
+67: var_202 = elementwise_add(var_166, var_200, axis=-1)
+67: var_204 = relu(var_202)
+67: var_208, var_209, var_210 = conv2d(var_204, res4a_branch2a_weights, stride=[2,2], padding=[1,1], groups=1, dilation=[1,1])
+67: var_220 = batchnorm(var_210, bn4a_branch2a_scale, bn4a_branch2a_offset, bn4a_branch2a_mean, bn4a_branch2a_variance, epsilon=1e-05)
+67: var_222 = relu(var_220)
+67: var_226, var_227, var_228 = conv2d(var_222, res4a_branch2b_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_238 = batchnorm(var_228, bn4a_branch2b_scale, bn4a_branch2b_offset, bn4a_branch2b_mean, bn4a_branch2b_variance, epsilon=1e-05)
+67: var_242, var_243, var_244 = conv2d(var_204, res4a_branch1_weights, stride=[2,2], padding=[0,0], groups=1, dilation=[1,1])
+67: var_254 = batchnorm(var_244, bn4a_branch1_scale, bn4a_branch1_offset, bn4a_branch1_mean, bn4a_branch1_variance, epsilon=1e-05)
+67: var_256 = elementwise_add(var_254, var_238, axis=-1)
+67: var_258 = relu(var_256)
+67: var_262, var_263, var_264 = conv2d(var_258, res4b_branch2a_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_274 = batchnorm(var_264, bn4b_branch2a_scale, bn4b_branch2a_offset, bn4b_branch2a_mean, bn4b_branch2a_variance, epsilon=1e-05)
+67: var_276 = relu(var_274)
+67: var_280, var_281, var_282 = conv2d(var_276, res4b_branch2b_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_292 = batchnorm(var_282, bn4b_branch2b_scale, bn4b_branch2b_offset, bn4b_branch2b_mean, bn4b_branch2b_variance, epsilon=1e-05)
+67: var_294 = elementwise_add(var_258, var_292, axis=-1)
+67: var_296 = relu(var_294)
+67: var_300, var_301, var_302 = conv2d(var_296, res5a_branch2a_weights, stride=[2,2], padding=[1,1], groups=1, dilation=[1,1])
+67: var_312 = batchnorm(var_302, bn5a_branch2a_scale, bn5a_branch2a_offset, bn5a_branch2a_mean, bn5a_branch2a_variance, epsilon=1e-05)
+67: var_314 = relu(var_312)
+67: var_318, var_319, var_320 = conv2d(var_314, res5a_branch2b_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_330 = batchnorm(var_320, bn5a_branch2b_scale, bn5a_branch2b_offset, bn5a_branch2b_mean, bn5a_branch2b_variance, epsilon=1e-05)
+67: var_334, var_335, var_336 = conv2d(var_296, res5a_branch1_weights, stride=[2,2], padding=[0,0], groups=1, dilation=[1,1])
+67: var_346 = batchnorm(var_336, bn5a_branch1_scale, bn5a_branch1_offset, bn5a_branch1_mean, bn5a_branch1_variance, epsilon=1e-05)
+67: var_348 = elementwise_add(var_346, var_330, axis=-1)
+67: var_350 = relu(var_348)
+67: var_354, var_355, var_356 = conv2d(var_350, res5b_branch2a_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_366 = batchnorm(var_356, bn5b_branch2a_scale, bn5b_branch2a_offset, bn5b_branch2a_mean, bn5b_branch2a_variance, epsilon=1e-05)
+67: var_368 = relu(var_366)
+67: var_372, var_373, var_374 = conv2d(var_368, res5b_branch2b_weights, stride=[1,1], padding=[1,1], groups=1, dilation=[1,1])
+67: var_384 = batchnorm(var_374, bn5b_branch2b_scale, bn5b_branch2b_offset, bn5b_branch2b_mean, bn5b_branch2b_variance, epsilon=1e-05)
+67: var_386 = elementwise_add(var_350, var_384, axis=-1)
+67: var_388 = relu(var_386)
+67: var_390, var_391 = pool2d(var_388, pool_type=avg, stride_size=[1,1], exclusive=true, kernel_size=[7,7], data_format=AnyLayout, padding_size=[0,0,0,0], ceil_mode=false)
+67: var_395 = mul(var_391, fc_0__w_0, y_num_col_dims=1, x_num_col_dims=1)
+67: var_399 = elementwise_add(var_395, fc_0__b_0, axis=1)
+67: var_401, var_402 = softmax(var_399, axis=-1)
+67: var_404 = scale(var_402, scale=1)
+67: }
+67: I0924 13:32:22.343083 28813 nn.cc:477] kernel_size length is: 2
+67: I0924 13:32:22.343093 28813 nn.cc:478] kernel_size is: 3
+67: I0924 13:32:22.343096 28813 nn.cc:479] padding_size length is: 4
+67: I0924 13:32:22.343097 28813 nn.cc:480] padding_size is: 1
+54: WARNING: Logging before InitGoogleLogging() is written to STDERR
+54: I0924 13:32:23.535948 27840 test02_matmul_case.cc:132] Testing matmul
+65: .I0924 13:32:24.203958 28049 syntax.cc:130] Loading Paddle model from /home/wangyue50/CINN-my/CINN/build/thirds/naive_mul_model
+65: I0924 13:32:24.204044 28049 model_parser.cc:188] model_dir is: /home/wangyue50/CINN-my/CINN/build/thirds/naive_mul_model
+65: I0924 13:32:24.204051 28049 model_parser.cc:189] model_file is: __model__
+65: I0924 13:32:24.204054 28049 model_parser.cc:190] param_file is: 
+65: I0924 13:32:24.205775 28049 paddle_model_to_program.cc:26] detect model output: [save_infer_model/scale_0.tmp_0]
+65: I0924 13:32:24.205875 28049 executor.cc:49] Program:
+65: Program {
+65: var_11 = mul(A, fc_0__w_0, y_num_col_dims=1, x_num_col_dims=1)
+65: var_15 = elementwise_add(var_11, fc_0__b_0, axis=1)
+65: var_17 = relu(var_15)
+65: var_19 = scale(var_17, scale=1)
+65: }
+65: I0924 13:32:24.252491 28049 graph_compiler.cc:39] [Debug] C Code is:
+65: #include <cinn_runtime.h>
+65: #include <stdio.h>
+65: 
+65: void fn_mul_0(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_0__w_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _Mul_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _Mul_out);
+65:   const float* A_reshape = ((const float*)(_A->memory));
+65:   float* Mul_out = ((float*)(_Mul_out->memory));
+65:   float* Mul_out_init = ((float*)(_Mul_out->memory));
+65:   const float* fc_0__w_0_reshape = ((const float*)(_fc_0__w_0->memory));
+65:   for (int32_t i = 0; i < 4; i += 1) {
+65:     for (int32_t j = 0; j < 30; j += 1) {
+65:       Mul_out_init[((30 * i) + j)] = 0;
+65:     };
+65:   };
+65:   for (int32_t i = 0; i < 4; i += 1) {
+65:     for (int32_t j = 0; j < 30; j += 1) {
+65:       for (int32_t axis_k = 0; axis_k < 30; axis_k += 1) {
+65:         Mul_out[((30 * i) + j)] = (Mul_out[((30 * i) + j)] + (A_reshape[((30 * i) + axis_k)] * fc_0__w_0_reshape[((30 * axis_k) + j)]));
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Mul_out);
+65: }
+65: 
+65: void fn_elementwise_add_1(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_11 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_0__b_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _C_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _C_0);
+65:   float* C_0 = ((float*)(_C_0->memory));
+65:   const float* fc_0__b_0 = ((const float*)(_fc_0__b_0->memory));
+65:   const float* var_11 = ((const float*)(_var_11->memory));
+65:   for (int32_t i = 0; i < 4; i += 1) {
+65:     for (int32_t j = 0; j < 30; j += 1) {
+65:       C_0[((30 * i) + j)] = (var_11[((30 * i) + j)] + fc_0__b_0[j]);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _C_0);
+65: }
+65: 
+65: void fn_relu_2(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_15 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Relu_output_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Relu_output_0);
+65:   float* Relu_output_0 = ((float*)(_Relu_output_0->memory));
+65:   const float* var_15 = ((const float*)(_var_15->memory));
+65:   for (int32_t i = 0; i < 4; i += 1) {
+65:     for (int32_t j = 0; j < 30; j += 1) {
+65:       Relu_output_0[((30 * i) + j)] = cinn_max(var_15[((30 * i) + j)], 0);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Relu_output_0);
+65: }
+65: 
+65: void fn_scale_3(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_17 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Scale_out_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Scale_out_0);
+65:   float* Scale_out_0 = ((float*)(_Scale_out_0->memory));
+65:   const float* var_17 = ((const float*)(_var_17->memory));
+65:   for (int32_t i = 0; i < 4; i += 1) {
+65:     for (int32_t j = 0; j < 30; j += 1) {
+65:       Scale_out_0[((30 * i) + j)] = var_17[((30 * i) + j)];
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Scale_out_0);
+65: }
+65: 
+65: I0924 13:32:24.283535 28049 analysis_predictor.cc:138] Profiler is deactivated, and no profiling report will be generated.
+65: I0924 13:32:24.283852 28049 analysis_predictor.cc:875] MODEL VERSION: 1.8.4
+65: I0924 13:32:24.283857 28049 analysis_predictor.cc:877] PREDICTOR VERSION: 1.8.4
+65: I0924 13:32:24.283890 28049 analysis_predictor.cc:474] ir_optim is turned off, no IR pass will be executed
+65: --- Running analysis [ir_graph_build_pass]
+65: --- Running analysis [ir_graph_clean_pass]
+65: --- Running analysis [ir_analysis_pass]
+65: --- Running analysis [ir_params_sync_among_devices_pass]
+65: --- Running analysis [adjust_cudnn_workspace_size_pass]
+65: --- Running analysis [inference_op_replace_pass]
+65: --- Running analysis [ir_graph_to_program_pass]
+65: I0924 13:32:24.284680 28049 analysis_predictor.cc:496] ======= optimize end =======
+65: .I0924 13:32:24.286247 28049 syntax.cc:130] Loading Paddle model from /home/wangyue50/CINN-my/CINN/build/thirds/multi_fc_model
+65: I0924 13:32:24.286298 28049 model_parser.cc:188] model_dir is: /home/wangyue50/CINN-my/CINN/build/thirds/multi_fc_model
+65: I0924 13:32:24.286303 28049 model_parser.cc:189] model_file is: __model__
+65: I0924 13:32:24.286304 28049 model_parser.cc:190] param_file is: 
+65: I0924 13:32:24.290424 28049 paddle_model_to_program.cc:26] detect model output: [save_infer_model/scale_0.tmp_0]
+65: I0924 13:32:24.290608 28049 executor.cc:49] Program:
+65: Program {
+65: var_25 = mul(A, fc_0__w_0, y_num_col_dims=1, x_num_col_dims=1)
+65: var_29 = elementwise_add(var_25, fc_bias, axis=1)
+65: var_31 = relu(var_29)
+65: var_35 = mul(var_31, fc_1__w_0, y_num_col_dims=1, x_num_col_dims=1)
+65: var_37 = elementwise_add(var_35, fc_bias, axis=1)
+65: var_39 = relu(var_37)
+65: var_43 = mul(var_39, fc_2__w_0, y_num_col_dims=1, x_num_col_dims=1)
+65: var_45 = elementwise_add(var_43, fc_bias, axis=1)
+65: var_47 = relu(var_45)
+65: var_51 = mul(var_47, fc_3__w_0, y_num_col_dims=1, x_num_col_dims=1)
+65: var_53 = elementwise_add(var_51, fc_bias, axis=1)
+65: var_55 = relu(var_53)
+65: var_59 = mul(var_55, fc_4__w_0, y_num_col_dims=1, x_num_col_dims=1)
+65: var_61 = elementwise_add(var_59, fc_bias, axis=1)
+65: var_63 = relu(var_61)
+65: var_67 = mul(var_63, fc_5__w_0, y_num_col_dims=1, x_num_col_dims=1)
+65: var_69 = elementwise_add(var_67, fc_bias, axis=1)
+65: var_71 = relu(var_69)
+65: var_73 = scale(var_71, scale=1)
+65: }
+65: I0924 13:32:24.521476 28049 graph_compiler.cc:39] [Debug] C Code is:
+65: #include <cinn_runtime.h>
+65: #include <stdio.h>
+65: 
+65: void fn_mul_0(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_0__w_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _Mul_out_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _Mul_out_0);
+65:   const float* A_reshape_0 = ((const float*)(_A->memory));
+65:   float* Mul_out_0 = ((float*)(_Mul_out_0->memory));
+65:   float* Mul_out_0_init = ((float*)(_Mul_out_0->memory));
+65:   const float* fc_0__w_0_reshape_0 = ((const float*)(_fc_0__w_0->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Mul_out_0_init[((64 * i) + j)] = 0;
+65:     };
+65:   };
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       for (int32_t axis_k_0 = 0; axis_k_0 < 64; axis_k_0 += 1) {
+65:         Mul_out_0[((64 * i) + j)] = (Mul_out_0[((64 * i) + j)] + (A_reshape_0[((64 * i) + axis_k_0)] * fc_0__w_0_reshape_0[((64 * axis_k_0) + j)]));
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Mul_out_0);
+65: }
+65: 
+65: void fn_elementwise_add_1(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_25 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_bias = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _C_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _C_1);
+65:   float* C_1 = ((float*)(_C_1->memory));
+65:   const float* fc_bias = ((const float*)(_fc_bias->memory));
+65:   const float* var_25 = ((const float*)(_var_25->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       C_1[((64 * i) + j)] = (var_25[((64 * i) + j)] + fc_bias[j]);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _C_1);
+65: }
+65: 
+65: void fn_relu_2(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_29 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Relu_output_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Relu_output_1);
+65:   float* Relu_output_1 = ((float*)(_Relu_output_1->memory));
+65:   const float* var_29 = ((const float*)(_var_29->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Relu_output_1[((64 * i) + j)] = cinn_max(var_29[((64 * i) + j)], 0);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Relu_output_1);
+65: }
+65: 
+65: void fn_mul_3(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_31 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_1__w_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _Mul_out_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _Mul_out_1);
+65:   float* Mul_out_1 = ((float*)(_Mul_out_1->memory));
+65:   float* Mul_out_1_init = ((float*)(_Mul_out_1->memory));
+65:   const float* fc_1__w_0_reshape = ((const float*)(_fc_1__w_0->memory));
+65:   const float* var_31_reshape = ((const float*)(_var_31->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Mul_out_1_init[((64 * i) + j)] = 0;
+65:     };
+65:   };
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       for (int32_t axis_k_1 = 0; axis_k_1 < 64; axis_k_1 += 1) {
+65:         Mul_out_1[((64 * i) + j)] = (Mul_out_1[((64 * i) + j)] + (var_31_reshape[((64 * i) + axis_k_1)] * fc_1__w_0_reshape[((64 * axis_k_1) + j)]));
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Mul_out_1);
+65: }
+65: 
+65: void fn_elementwise_add_4(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_35 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_bias = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _C_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _C_2);
+65:   float* C_2 = ((float*)(_C_2->memory));
+65:   const float* fc_bias = ((const float*)(_fc_bias->memory));
+65:   const float* var_35 = ((const float*)(_var_35->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       C_2[((64 * i) + j)] = (var_35[((64 * i) + j)] + fc_bias[j]);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _C_2);
+65: }
+65: 
+65: void fn_relu_5(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_37 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Relu_output_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Relu_output_2);
+65:   float* Relu_output_2 = ((float*)(_Relu_output_2->memory));
+65:   const float* var_37 = ((const float*)(_var_37->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Relu_output_2[((64 * i) + j)] = cinn_max(var_37[((64 * i) + j)], 0);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Relu_output_2);
+65: }
+65: 
+65: void fn_mul_6(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_39 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_2__w_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _Mul_out_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _Mul_out_2);
+65:   float* Mul_out_2 = ((float*)(_Mul_out_2->memory));
+65:   float* Mul_out_2_init = ((float*)(_Mul_out_2->memory));
+65:   const float* fc_2__w_0_reshape = ((const float*)(_fc_2__w_0->memory));
+65:   const float* var_39_reshape = ((const float*)(_var_39->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Mul_out_2_init[((64 * i) + j)] = 0;
+65:     };
+65:   };
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       for (int32_t axis_k_2 = 0; axis_k_2 < 64; axis_k_2 += 1) {
+65:         Mul_out_2[((64 * i) + j)] = (Mul_out_2[((64 * i) + j)] + (var_39_reshape[((64 * i) + axis_k_2)] * fc_2__w_0_reshape[((64 * axis_k_2) + j)]));
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Mul_out_2);
+65: }
+65: 
+65: void fn_elementwise_add_7(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_43 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_bias = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _C_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _C_3);
+65:   float* C_3 = ((float*)(_C_3->memory));
+65:   const float* fc_bias = ((const float*)(_fc_bias->memory));
+65:   const float* var_43 = ((const float*)(_var_43->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       C_3[((64 * i) + j)] = (var_43[((64 * i) + j)] + fc_bias[j]);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _C_3);
+65: }
+65: 
+65: void fn_relu_8(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_45 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Relu_output_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Relu_output_3);
+65:   float* Relu_output_3 = ((float*)(_Relu_output_3->memory));
+65:   const float* var_45 = ((const float*)(_var_45->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Relu_output_3[((64 * i) + j)] = cinn_max(var_45[((64 * i) + j)], 0);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Relu_output_3);
+65: }
+65: 
+65: void fn_mul_9(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_47 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_3__w_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _Mul_out_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _Mul_out_3);
+65:   float* Mul_out_3 = ((float*)(_Mul_out_3->memory));
+65:   float* Mul_out_3_init = ((float*)(_Mul_out_3->memory));
+65:   const float* fc_3__w_0_reshape = ((const float*)(_fc_3__w_0->memory));
+65:   const float* var_47_reshape = ((const float*)(_var_47->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Mul_out_3_init[((64 * i) + j)] = 0;
+65:     };
+65:   };
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       for (int32_t axis_k_3 = 0; axis_k_3 < 64; axis_k_3 += 1) {
+65:         Mul_out_3[((64 * i) + j)] = (Mul_out_3[((64 * i) + j)] + (var_47_reshape[((64 * i) + axis_k_3)] * fc_3__w_0_reshape[((64 * axis_k_3) + j)]));
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Mul_out_3);
+65: }
+65: 
+65: void fn_elementwise_add_10(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_51 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_bias = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _C_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _C_4);
+65:   float* C_4 = ((float*)(_C_4->memory));
+65:   const float* fc_bias = ((const float*)(_fc_bias->memory));
+65:   const float* var_51 = ((const float*)(_var_51->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       C_4[((64 * i) + j)] = (var_51[((64 * i) + j)] + fc_bias[j]);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _C_4);
+65: }
+65: 
+65: void fn_relu_11(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_53 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Relu_output_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Relu_output_4);
+65:   float* Relu_output_4 = ((float*)(_Relu_output_4->memory));
+65:   const float* var_53 = ((const float*)(_var_53->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Relu_output_4[((64 * i) + j)] = cinn_max(var_53[((64 * i) + j)], 0);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Relu_output_4);
+65: }
+65: 
+65: void fn_mul_12(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_55 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_4__w_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _Mul_out_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _Mul_out_4);
+65:   float* Mul_out_4 = ((float*)(_Mul_out_4->memory));
+65:   float* Mul_out_4_init = ((float*)(_Mul_out_4->memory));
+65:   const float* fc_4__w_0_reshape = ((const float*)(_fc_4__w_0->memory));
+65:   const float* var_55_reshape = ((const float*)(_var_55->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Mul_out_4_init[((64 * i) + j)] = 0;
+65:     };
+65:   };
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       for (int32_t axis_k_4 = 0; axis_k_4 < 64; axis_k_4 += 1) {
+65:         Mul_out_4[((64 * i) + j)] = (Mul_out_4[((64 * i) + j)] + (var_55_reshape[((64 * i) + axis_k_4)] * fc_4__w_0_reshape[((64 * axis_k_4) + j)]));
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Mul_out_4);
+65: }
+65: 
+65: void fn_elementwise_add_13(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_59 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_bias = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _C_5 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _C_5);
+65:   float* C_5 = ((float*)(_C_5->memory));
+65:   const float* fc_bias = ((const float*)(_fc_bias->memory));
+65:   const float* var_59 = ((const float*)(_var_59->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       C_5[((64 * i) + j)] = (var_59[((64 * i) + j)] + fc_bias[j]);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _C_5);
+65: }
+65: 
+65: void fn_relu_14(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_61 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Relu_output_5 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Relu_output_5);
+65:   float* Relu_output_5 = ((float*)(_Relu_output_5->memory));
+65:   const float* var_61 = ((const float*)(_var_61->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Relu_output_5[((64 * i) + j)] = cinn_max(var_61[((64 * i) + j)], 0);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Relu_output_5);
+65: }
+65: 
+65: void fn_mul_15(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_63 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_5__w_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _Mul_out_5 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _Mul_out_5);
+65:   float* Mul_out_5 = ((float*)(_Mul_out_5->memory));
+65:   float* Mul_out_5_init = ((float*)(_Mul_out_5->memory));
+65:   const float* fc_5__w_0_reshape = ((const float*)(_fc_5__w_0->memory));
+65:   const float* var_63_reshape = ((const float*)(_var_63->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Mul_out_5_init[((64 * i) + j)] = 0;
+65:     };
+65:   };
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       for (int32_t axis_k_5 = 0; axis_k_5 < 64; axis_k_5 += 1) {
+65:         Mul_out_5[((64 * i) + j)] = (Mul_out_5[((64 * i) + j)] + (var_63_reshape[((64 * i) + axis_k_5)] * fc_5__w_0_reshape[((64 * axis_k_5) + j)]));
+65:       };
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Mul_out_5);
+65: }
+65: 
+65: void fn_elementwise_add_16(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_67 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   const cinn_buffer_t* _fc_bias = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_t* _C_6 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+65:   cinn_buffer_malloc((void*)(0), _C_6);
+65:   float* C_6 = ((float*)(_C_6->memory));
+65:   const float* fc_bias = ((const float*)(_fc_bias->memory));
+65:   const float* var_67 = ((const float*)(_var_67->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       C_6[((64 * i) + j)] = (var_67[((64 * i) + j)] + fc_bias[j]);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _C_6);
+65: }
+65: 
+65: void fn_relu_17(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_69 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Relu_output_6 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Relu_output_6);
+65:   float* Relu_output_6 = ((float*)(_Relu_output_6->memory));
+65:   const float* var_69 = ((const float*)(_var_69->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Relu_output_6[((64 * i) + j)] = cinn_max(var_69[((64 * i) + j)], 0);
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Relu_output_6);
+65: }
+65: 
+65: void fn_scale_18(void* _args, int32_t num_args)
+65: {
+65:   const cinn_buffer_t* _var_71 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+65:   cinn_buffer_t* _Scale_out_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+65:   cinn_buffer_malloc((void*)(0), _Scale_out_1);
+65:   float* Scale_out_1 = ((float*)(_Scale_out_1->memory));
+65:   const float* var_71 = ((const float*)(_var_71->memory));
+65:   for (int32_t i = 0; i < 8; i += 1) {
+65:     for (int32_t j = 0; j < 64; j += 1) {
+65:       Scale_out_1[((64 * i) + j)] = var_71[((64 * i) + j)];
+65:     };
+65:   };
+65:   cinn_buffer_free((void*)(0), _Scale_out_1);
+65: }
+65: 
+65: I0924 13:32:24.582810 28049 analysis_predictor.cc:138] Profiler is deactivated, and no profiling report will be generated.
+65: I0924 13:32:24.583323 28049 analysis_predictor.cc:875] MODEL VERSION: 1.8.4
+65: I0924 13:32:24.583328 28049 analysis_predictor.cc:877] PREDICTOR VERSION: 1.8.4
+65: I0924 13:32:24.583356 28049 analysis_predictor.cc:474] ir_optim is turned off, no IR pass will be executed
+65: --- Running analysis [ir_graph_build_pass]
+65: --- Running analysis [ir_graph_clean_pass]
+65: --- Running analysis [ir_analysis_pass]
+65: --- Running analysis [ir_params_sync_among_devices_pass]
+65: --- Running analysis [adjust_cudnn_workspace_size_pass]
+65: --- Running analysis [inference_op_replace_pass]
+65: --- Running analysis [ir_graph_to_program_pass]
+65: I0924 13:32:24.585050 28049 analysis_predictor.cc:496] ======= optimize end =======
+65: .
+65: ----------------------------------------------------------------------
+65: Ran 3 tests in 3.034s
+65: 
+65: OK
+65: f var_3
+65: var = elementwise_add(A, B)
+65: var_0 = relu(var)
+65: var_1, var_2, var_3 = conv2d(var_0, E, dilation=[1,1], padding=[0,0], stride=[1,1])
+65: var_4 = scale(var_3, scale=2, bias=0.5)
+65: var_5, var_6 = softmax(var_4, axis=1)
+65: result in paddle_verify: 
+65: 
+65: x_data [[0.5488281  0.71533203 0.60253906 0.5449219  0.42358398 0.6459961
+65:   0.4375     0.89160156 0.9638672  0.38354492 0.7915039  0.5288086
+65:   0.5678711  0.92578125 0.07104492 0.0871582  0.0202179  0.83251953
+65:   0.7783203  0.8701172  0.9785156  0.7993164  0.46142578 0.7807617
+65:   0.11828613 0.6401367  0.14331055 0.9448242  0.52197266 0.41455078]
+65:  [0.26464844 0.77441406 0.4560547  0.5683594  0.01878357 0.6176758
+65:   0.6123047  0.6166992  0.94384766 0.6816406  0.35961914 0.43701172
+65:   0.6977539  0.06021118 0.6669922  0.67041016 0.21032715 0.12890625
+65:   0.3154297  0.36376953 0.5703125  0.4387207  0.98828125 0.10205078
+65:   0.2088623  0.16125488 0.6533203  0.25317383 0.4663086  0.24438477]
+65:  [0.15893555 0.11035156 0.65625    0.1381836  0.1965332  0.36865234
+65:   0.8208008  0.09710693 0.8378906  0.09606934 0.9765625  0.46875
+65:   0.9765625  0.60498047 0.7392578  0.03918457 0.28271484 0.12017822
+65:   0.29614258 0.11871338 0.3178711  0.41430664 0.06414795 0.6923828
+65:   0.56640625 0.26538086 0.5234375  0.09393311 0.5761719  0.9291992 ]
+65:  [0.31860352 0.66748047 0.13183594 0.7163086  0.28930664 0.18322754
+65:   0.5864258  0.02011108 0.82910156 0.00469589 0.6777344  0.27001953
+65:   0.73535156 0.96240234 0.2487793  0.5761719  0.59228516 0.5722656
+65:   0.22302246 0.9526367  0.44702148 0.8461914  0.69970703 0.29736328
+65:   0.81396484 0.39648438 0.8808594  0.5810547  0.88183594 0.6923828 ]]
+67: I0924 13:32:24.899466 28813 nn.cc:477] kernel_size length is: 2
+67: I0924 13:32:24.899487 28813 nn.cc:478] kernel_size is: 7
+67: I0924 13:32:24.899490 28813 nn.cc:479] padding_size length is: 4
+67: I0924 13:32:24.899492 28813 nn.cc:480] padding_size is: 0
+64/68 Test #65: test_cinn_frontend .....................   Passed    4.61 sec
+67: I0924 13:32:25.657984 28813 graph_compiler.cc:39] [Debug] C Code is:
+67: #include <cinn_runtime.h>
+67: #include <stdio.h>
+67: 
+67: void fn_conv2d_0(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _image = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _conv1_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_t* _T_pad_out_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   cinn_buffer_t* _weights_dilation = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   cinn_buffer_t* _Conv2d_nchw_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_malloc((void*)(0), _T_pad_out_0);
+67:   cinn_buffer_malloc((void*)(0), _weights_dilation);
+67:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out);
+67:   float* Conv2d_nchw_out = ((float*)(_Conv2d_nchw_out->memory));
+67:   float* Conv2d_nchw_out_init = ((float*)(_Conv2d_nchw_out->memory));
+67:   float* T_pad_out_0 = ((float*)(_T_pad_out_0->memory));
+67:   const float* conv1_weights = ((const float*)(_conv1_weights->memory));
+67:   const float* image = ((const float*)(_image->memory));
+67:   float* weights_dilation = ((float*)(_weights_dilation->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 112; k += 1) {
+67:       for (int32_t a = 0; a < 112; a += 1) {
+67:         Conv2d_nchw_out_init[((12544 * j) + ((112 * k) + a))] = 0;
+67:       };
+67:     };
+67:   };
+67:   for (int32_t i = 0; i < 64; i += 1) {
+67:     for (int32_t j = 0; j < 3; j += 1) {
+67:       for (int32_t k = 0; k < 7; k += 1) {
+67:         for (int32_t a = 0; a < 7; a += 1) {
+67:           weights_dilation[((147 * i) + ((49 * j) + ((7 * k) + a)))] = (((((a % 1) == 0) && ((k % 1) == 0))) ? conv1_weights[((a / 1) + (((k / 1) * 7) + ((147 * i) + (49 * j))))] : 0);
+67:         };
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 3; j += 1) {
+67:     for (int32_t k = 0; k < 230; k += 1) {
+67:       for (int32_t a = 0; a < 230; a += 1) {
+67:         T_pad_out_0[((52900 * j) + ((230 * k) + a))] = ((((a < 227) && ((a >= 3) && ((k < 227) && (k >= 3))))) ? image[(-675 + ((50176 * j) + ((224 * k) + a)))] : 0);
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 112; k += 1) {
+67:       for (int32_t a = 0; a < 112; a += 1) {
+67:         for (int32_t fc = 0; fc < 3; fc += 1) {
+67:           for (int32_t fy = 0; fy < 7; fy += 1) {
+67:             for (int32_t fx = 0; fx < 7; fx += 1) {
+67:               Conv2d_nchw_out[((12544 * j) + ((112 * k) + a))] = (Conv2d_nchw_out[((12544 * j) + ((112 * k) + a))] + (T_pad_out_0[((2 * a) + ((52900 * fc) + ((230 * fy) + ((460 * k) + fx))))] * weights_dilation[((49 * fc) + ((7 * fy) + ((147 * j) + fx)))]));
+67:             };
+67:           };
+67:         };
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _T_pad_out_0);
+67:   cinn_buffer_free((void*)(0), _weights_dilation);
+67:   cinn_buffer_free((void*)(0), _Conv2d_nchw_out);
+67: }
+67: 
+67: void fn_batchnorm_1(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_5 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _bn_conv1_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   const cinn_buffer_t* _bn_conv1_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   const cinn_buffer_t* _bn_conv1_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   const cinn_buffer_t* _bn_conv1_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_t* _BatchNorm_output_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+67:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_0);
+67:   float* BatchNorm_output_0 = ((float*)(_BatchNorm_output_0->memory));
+67:   const float* bn_conv1_mean = ((const float*)(_bn_conv1_mean->memory));
+67:   const float* bn_conv1_offset = ((const float*)(_bn_conv1_offset->memory));
+67:   const float* bn_conv1_scale = ((const float*)(_bn_conv1_scale->memory));
+67:   const float* bn_conv1_variance = ((const float*)(_bn_conv1_variance->memory));
+67:   const float* var_5 = ((const float*)(_var_5->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 112; k += 1) {
+67:       for (int32_t a = 0; a < 112; a += 1) {
+67:         BatchNorm_output_0[((12544 * j) + ((112 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((bn_conv1_variance[j] + 1e-05))) * (bn_conv1_scale[j] * var_5[((12544 * j) + ((112 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((bn_conv1_variance[j] + 1e-05))) * (bn_conv1_scale[j] * bn_conv1_mean[j]))) + bn_conv1_offset[j]));
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _BatchNorm_output_0);
+67: }
+67: 
+67: void fn_relu_2(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_15 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   cinn_buffer_t* _Relu_output = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_malloc((void*)(0), _Relu_output);
+67:   float* Relu_output = ((float*)(_Relu_output->memory));
+67:   const float* var_15 = ((const float*)(_var_15->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 112; k += 1) {
+67:       for (int32_t a = 0; a < 112; a += 1) {
+67:         Relu_output[((12544 * j) + ((112 * k) + a))] = cinn_max(var_15[((12544 * j) + ((112 * k) + a))], 0);
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _Relu_output);
+67: }
+67: 
+67: void fn_pool2d_3(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_17 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   cinn_buffer_t* _pad_temp_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_t* _T_Pool2d_out_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   cinn_buffer_malloc((void*)(0), _pad_temp_0);
+67:   cinn_buffer_malloc((void*)(0), _T_Pool2d_out_0);
+67:   float* T_Pool2d_out_0 = ((float*)(_T_Pool2d_out_0->memory));
+67:   float* T_Pool2d_out_0_init = ((float*)(_T_Pool2d_out_0->memory));
+67:   float* pad_temp_0 = ((float*)(_pad_temp_0->memory));
+67:   const float* var_17 = ((const float*)(_var_17->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         T_Pool2d_out_0_init[((3136 * j) + ((56 * k) + a))] = 0;
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 114; k += 1) {
+67:       for (int32_t a = 0; a < 114; a += 1) {
+67:         pad_temp_0[((12996 * j) + ((114 * k) + a))] = ((((a < 113) && ((a >= 1) && ((k < 113) && (k >= 1))))) ? var_17[(-113 + ((12544 * j) + ((112 * k) + a)))] : -3.40282e+38);
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         for (int32_t kernel_idx = 0; kernel_idx < 3; kernel_idx += 1) {
+67:           for (int32_t kernel_idx_0 = 0; kernel_idx_0 < 3; kernel_idx_0 += 1) {
+67:             T_Pool2d_out_0[((3136 * j) + ((56 * k) + a))] = cinn_max(T_Pool2d_out_0[((3136 * j) + ((56 * k) + a))], pad_temp_0[((2 * a) + ((12996 * j) + ((228 * k) + ((114 * kernel_idx) + kernel_idx_0))))]);
+67:           };
+67:         };
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _pad_temp_0);
+67:   cinn_buffer_free((void*)(0), _T_Pool2d_out_0);
+67: }
+67: 
+67: void fn_conv2d_9(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_20 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _res2a_branch1_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_t* _T_Identity_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   cinn_buffer_t* _weights_dilation_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   cinn_buffer_t* _Conv2d_nchw_out_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_malloc((void*)(0), _T_Identity_out);
+67:   cinn_buffer_malloc((void*)(0), _weights_dilation_0);
+67:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out_0);
+67:   float* Conv2d_nchw_out_0 = ((float*)(_Conv2d_nchw_out_0->memory));
+67:   float* Conv2d_nchw_out_0_init = ((float*)(_Conv2d_nchw_out_0->memory));
+67:   float* T_Identity_out = ((float*)(_T_Identity_out->memory));
+67:   const float* res2a_branch1_weights = ((const float*)(_res2a_branch1_weights->memory));
+67:   const float* var_20 = ((const float*)(_var_20->memory));
+67:   float* weights_dilation_0 = ((float*)(_weights_dilation_0->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         Conv2d_nchw_out_0_init[((3136 * j) + ((56 * k) + a))] = 0;
+67:       };
+67:     };
+67:   };
+67:   for (int32_t i = 0; i < 64; i += 1) {
+67:     for (int32_t j = 0; j < 64; j += 1) {
+67:       weights_dilation_0[((64 * i) + j)] = (((((0 % 1) == 0) && ((0 % 1) == 0))) ? res2a_branch1_weights[((64 * i) + j)] : 0);
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         T_Identity_out[((3136 * j) + ((56 * k) + a))] = var_20[((3136 * j) + ((56 * k) + a))];
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         for (int32_t fc_0 = 0; fc_0 < 64; fc_0 += 1) {
+67:           Conv2d_nchw_out_0[((3136 * j) + ((56 * k) + a))] = (Conv2d_nchw_out_0[((3136 * j) + ((56 * k) + a))] + (T_Identity_out[((3136 * fc_0) + ((56 * k) + a))] * weights_dilation_0[((64 * j) + fc_0)]));
+67:         };
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _T_Identity_out);
+67:   cinn_buffer_free((void*)(0), _weights_dilation_0);
+67:   cinn_buffer_free((void*)(0), _Conv2d_nchw_out_0);
+67: }
+67: 
+67: void fn_conv2d_4(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_20 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _res2a_branch2a_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_t* _T_pad_out_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   cinn_buffer_t* _weights_dilation_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   cinn_buffer_t* _Conv2d_nchw_out_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_malloc((void*)(0), _T_pad_out_1);
+67:   cinn_buffer_malloc((void*)(0), _weights_dilation_1);
+67:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out_1);
+67:   float* Conv2d_nchw_out_1 = ((float*)(_Conv2d_nchw_out_1->memory));
+67:   float* Conv2d_nchw_out_1_init = ((float*)(_Conv2d_nchw_out_1->memory));
+67:   float* T_pad_out_1 = ((float*)(_T_pad_out_1->memory));
+67:   const float* res2a_branch2a_weights = ((const float*)(_res2a_branch2a_weights->memory));
+67:   const float* var_20 = ((const float*)(_var_20->memory));
+67:   float* weights_dilation_1 = ((float*)(_weights_dilation_1->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         Conv2d_nchw_out_1_init[((3136 * j) + ((56 * k) + a))] = 0;
+67:       };
+67:     };
+67:   };
+67:   for (int32_t i = 0; i < 64; i += 1) {
+67:     for (int32_t j = 0; j < 64; j += 1) {
+67:       for (int32_t k = 0; k < 3; k += 1) {
+67:         for (int32_t a = 0; a < 3; a += 1) {
+67:           weights_dilation_1[((576 * i) + ((9 * j) + ((3 * k) + a)))] = (((((a % 1) == 0) && ((k % 1) == 0))) ? res2a_branch2a_weights[((a / 1) + (((k / 1) * 3) + ((576 * i) + (9 * j))))] : 0);
+67:         };
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 58; k += 1) {
+67:       for (int32_t a = 0; a < 58; a += 1) {
+67:         T_pad_out_1[((3364 * j) + ((58 * k) + a))] = ((((a < 57) && ((a >= 1) && ((k < 57) && (k >= 1))))) ? var_20[(-57 + ((3136 * j) + ((56 * k) + a)))] : 0);
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         for (int32_t fc_1 = 0; fc_1 < 64; fc_1 += 1) {
+67:           for (int32_t fy_1 = 0; fy_1 < 3; fy_1 += 1) {
+67:             for (int32_t fx_1 = 0; fx_1 < 3; fx_1 += 1) {
+67:               Conv2d_nchw_out_1[((3136 * j) + ((56 * k) + a))] = (Conv2d_nchw_out_1[((3136 * j) + ((56 * k) + a))] + (T_pad_out_1[((3364 * fc_1) + ((58 * fy_1) + ((58 * k) + (a + fx_1))))] * weights_dilation_1[((9 * fc_1) + ((3 * fy_1) + ((576 * j) + fx_1)))]));
+67:             };
+67:           };
+67:         };
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _T_pad_out_1);
+67:   cinn_buffer_free((void*)(0), _weights_dilation_1);
+67:   cinn_buffer_free((void*)(0), _Conv2d_nchw_out_1);
+67: }
+67: 
+67: void fn_batchnorm_10(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_60 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _bn2a_branch1_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   const cinn_buffer_t* _bn2a_branch1_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   const cinn_buffer_t* _bn2a_branch1_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   const cinn_buffer_t* _bn2a_branch1_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_t* _BatchNorm_output_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+67:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_1);
+67:   float* BatchNorm_output_1 = ((float*)(_BatchNorm_output_1->memory));
+67:   const float* bn2a_branch1_mean = ((const float*)(_bn2a_branch1_mean->memory));
+67:   const float* bn2a_branch1_offset = ((const float*)(_bn2a_branch1_offset->memory));
+67:   const float* bn2a_branch1_scale = ((const float*)(_bn2a_branch1_scale->memory));
+67:   const float* bn2a_branch1_variance = ((const float*)(_bn2a_branch1_variance->memory));
+67:   const float* var_60 = ((const float*)(_var_60->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         BatchNorm_output_1[((3136 * j) + ((56 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((bn2a_branch1_variance[j] + 1e-05))) * (bn2a_branch1_scale[j] * var_60[((3136 * j) + ((56 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((bn2a_branch1_variance[j] + 1e-05))) * (bn2a_branch1_scale[j] * bn2a_branch1_mean[j]))) + bn2a_branch1_offset[j]));
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _BatchNorm_output_1);
+67: }
+67: 
+67: void fn_batchnorm_5(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_26 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _bn2a_branch2a_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   const cinn_buffer_t* _bn2a_branch2a_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   const cinn_buffer_t* _bn2a_branch2a_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   const cinn_buffer_t* _bn2a_branch2a_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_t* _BatchNorm_output_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+67:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_2);
+67:   float* BatchNorm_output_2 = ((float*)(_BatchNorm_output_2->memory));
+67:   const float* bn2a_branch2a_mean = ((const float*)(_bn2a_branch2a_mean->memory));
+67:   const float* bn2a_branch2a_offset = ((const float*)(_bn2a_branch2a_offset->memory));
+67:   const float* bn2a_branch2a_scale = ((const float*)(_bn2a_branch2a_scale->memory));
+67:   const float* bn2a_branch2a_variance = ((const float*)(_bn2a_branch2a_variance->memory));
+67:   const float* var_26 = ((const float*)(_var_26->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         BatchNorm_output_2[((3136 * j) + ((56 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((bn2a_branch2a_variance[j] + 1e-05))) * (bn2a_branch2a_scale[j] * var_26[((3136 * j) + ((56 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((bn2a_branch2a_variance[j] + 1e-05))) * (bn2a_branch2a_scale[j] * bn2a_branch2a_mean[j]))) + bn2a_branch2a_offset[j]));
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _BatchNorm_output_2);
+67: }
+67: 
+67: void fn_relu_6(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_36 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   cinn_buffer_t* _Relu_output_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_malloc((void*)(0), _Relu_output_0);
+67:   float* Relu_output_0 = ((float*)(_Relu_output_0->memory));
+67:   const float* var_36 = ((const float*)(_var_36->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         Relu_output_0[((3136 * j) + ((56 * k) + a))] = cinn_max(var_36[((3136 * j) + ((56 * k) + a))], 0);
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _Relu_output_0);
+67: }
+67: 
+67: void fn_conv2d_7(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_38 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _res2a_branch2b_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_t* _T_pad_out_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   cinn_buffer_t* _weights_dilation_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   cinn_buffer_t* _Conv2d_nchw_out_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_malloc((void*)(0), _T_pad_out_2);
+67:   cinn_buffer_malloc((void*)(0), _weights_dilation_2);
+67:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out_2);
+67:   float* Conv2d_nchw_out_2 = ((float*)(_Conv2d_nchw_out_2->memory));
+67:   float* Conv2d_nchw_out_2_init = ((float*)(_Conv2d_nchw_out_2->memory));
+67:   float* T_pad_out_2 = ((float*)(_T_pad_out_2->memory));
+67:   const float* res2a_branch2b_weights = ((const float*)(_res2a_branch2b_weights->memory));
+67:   const float* var_38 = ((const float*)(_var_38->memory));
+67:   float* weights_dilation_2 = ((float*)(_weights_dilation_2->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         Conv2d_nchw_out_2_init[((3136 * j) + ((56 * k) + a))] = 0;
+67:       };
+67:     };
+67:   };
+67:   for (int32_t i = 0; i < 64; i += 1) {
+67:     for (int32_t j = 0; j < 64; j += 1) {
+67:       for (int32_t k = 0; k < 3; k += 1) {
+67:         for (int32_t a = 0; a < 3; a += 1) {
+67:           weights_dilation_2[((576 * i) + ((9 * j) + ((3 * k) + a)))] = (((((a % 1) == 0) && ((k % 1) == 0))) ? res2a_branch2b_weights[((a / 1) + (((k / 1) * 3) + ((576 * i) + (9 * j))))] : 0);
+67:         };
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 58; k += 1) {
+67:       for (int32_t a = 0; a < 58; a += 1) {
+67:         T_pad_out_2[((3364 * j) + ((58 * k) + a))] = ((((a < 57) && ((a >= 1) && ((k < 57) && (k >= 1))))) ? var_38[(-57 + ((3136 * j) + ((56 * k) + a)))] : 0);
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         for (int32_t fc_2 = 0; fc_2 < 64; fc_2 += 1) {
+67:           for (int32_t fy_2 = 0; fy_2 < 3; fy_2 += 1) {
+67:             for (int32_t fx_2 = 0; fx_2 < 3; fx_2 += 1) {
+67:               Conv2d_nchw_out_2[((3136 * j) + ((56 * k) + a))] = (Conv2d_nchw_out_2[((3136 * j) + ((56 * k) + a))] + (T_pad_out_2[((3364 * fc_2) + ((58 * fy_2) + ((58 * k) + (a + fx_2))))] * weights_dilation_2[((9 * fc_2) + ((3 * fy_2) + ((576 * j) + fx_2)))]));
+67:             };
+67:           };
+67:         };
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _T_pad_out_2);
+67:   cinn_buffer_free((void*)(0), _weights_dilation_2);
+67:   cinn_buffer_free((void*)(0), _Conv2d_nchw_out_2);
+67: }
+67: 
+67: void fn_batchnorm_8(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_44 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _bn2a_branch2b_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   const cinn_buffer_t* _bn2a_branch2b_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   const cinn_buffer_t* _bn2a_branch2b_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   const cinn_buffer_t* _bn2a_branch2b_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_t* _BatchNorm_output_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+67:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_3);
+67:   float* BatchNorm_output_3 = ((float*)(_BatchNorm_output_3->memory));
+67:   const float* bn2a_branch2b_mean = ((const float*)(_bn2a_branch2b_mean->memory));
+67:   const float* bn2a_branch2b_offset = ((const float*)(_bn2a_branch2b_offset->memory));
+67:   const float* bn2a_branch2b_scale = ((const float*)(_bn2a_branch2b_scale->memory));
+67:   const float* bn2a_branch2b_variance = ((const float*)(_bn2a_branch2b_variance->memory));
+67:   const float* var_44 = ((const float*)(_var_44->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         BatchNorm_output_3[((3136 * j) + ((56 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((bn2a_branch2b_variance[j] + 1e-05))) * (bn2a_branch2b_scale[j] * var_44[((3136 * j) + ((56 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((bn2a_branch2b_variance[j] + 1e-05))) * (bn2a_branch2b_scale[j] * bn2a_branch2b_mean[j]))) + bn2a_branch2b_offset[j]));
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _BatchNorm_output_3);
+67: }
+67: 
+67: void fn_elementwise_add_11(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_70 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _var_54 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_t* _C = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   cinn_buffer_malloc((void*)(0), _C);
+67:   float* C = ((float*)(_C->memory));
+67:   const float* var_54 = ((const float*)(_var_54->memory));
+67:   const float* var_70 = ((const float*)(_var_70->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         C[((3136 * j) + ((56 * k) + a))] = (var_70[((3136 * j) + ((56 * k) + a))] + var_54[((3136 * j) + ((56 * k) + a))]);
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _C);
+67: }
+67: 
+67: void fn_relu_12(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_72 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   cinn_buffer_t* _Relu_output_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_malloc((void*)(0), _Relu_output_1);
+67:   float* Relu_output_1 = ((float*)(_Relu_output_1->memory));
+67:   const float* var_72 = ((const float*)(_var_72->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         Relu_output_1[((3136 * j) + ((56 * k) + a))] = cinn_max(var_72[((3136 * j) + ((56 * k) + a))], 0);
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _Relu_output_1);
+67: }
+67: 
+67: void fn_conv2d_13(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_74 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _res2b_branch2a_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_t* _T_pad_out_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   cinn_buffer_t* _weights_dilation_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   cinn_buffer_t* _Conv2d_nchw_out_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_malloc((void*)(0), _T_pad_out_3);
+67:   cinn_buffer_malloc((void*)(0), _weights_dilation_3);
+67:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out_3);
+67:   float* Conv2d_nchw_out_3 = ((float*)(_Conv2d_nchw_out_3->memory));
+67:   float* Conv2d_nchw_out_3_init = ((float*)(_Conv2d_nchw_out_3->memory));
+67:   float* T_pad_out_3 = ((float*)(_T_pad_out_3->memory));
+67:   const float* res2b_branch2a_weights = ((const float*)(_res2b_branch2a_weights->memory));
+67:   const float* var_74 = ((const float*)(_var_74->memory));
+67:   float* weights_dilation_3 = ((float*)(_weights_dilation_3->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         Conv2d_nchw_out_3_init[((3136 * j) + ((56 * k) + a))] = 0;
+67:       };
+67:     };
+67:   };
+67:   for (int32_t i = 0; i < 64; i += 1) {
+67:     for (int32_t j = 0; j < 64; j += 1) {
+67:       for (int32_t k = 0; k < 3; k += 1) {
+67:         for (int32_t a = 0; a < 3; a += 1) {
+67:           weights_dilation_3[((576 * i) + ((9 * j) + ((3 * k) + a)))] = (((((a % 1) == 0) && ((k % 1) == 0))) ? res2b_branch2a_weights[((a / 1) + (((k / 1) * 3) + ((576 * i) + (9 * j))))] : 0);
+67:         };
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 58; k += 1) {
+67:       for (int32_t a = 0; a < 58; a += 1) {
+67:         T_pad_out_3[((3364 * j) + ((58 * k) + a))] = ((((a < 57) && ((a >= 1) && ((k < 57) && (k >= 1))))) ? var_74[(-57 + ((3136 * j) + ((56 * k) + a)))] : 0);
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         for (int32_t fc_3 = 0; fc_3 < 64; fc_3 += 1) {
+67:           for (int32_t fy_3 = 0; fy_3 < 3; fy_3 += 1) {
+67:             for (int32_t fx_3 = 0; fx_3 < 3; fx_3 += 1) {
+67:               Conv2d_nchw_out_3[((3136 * j) + ((56 * k) + a))] = (Conv2d_nchw_out_3[((3136 * j) + ((56 * k) + a))] + (T_pad_out_3[((3364 * fc_3) + ((58 * fy_3) + ((58 * k) + (a + fx_3))))] * weights_dilation_3[((9 * fc_3) + ((3 * fy_3) + ((576 * j) + fx_3)))]));
+67:             };
+67:           };
+67:         };
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _T_pad_out_3);
+67:   cinn_buffer_free((void*)(0), _weights_dilation_3);
+67:   cinn_buffer_free((void*)(0), _Conv2d_nchw_out_3);
+67: }
+67: 
+67: void fn_batchnorm_14(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_80 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _bn2b_branch2a_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   const cinn_buffer_t* _bn2b_branch2a_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   const cinn_buffer_t* _bn2b_branch2a_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   const cinn_buffer_t* _bn2b_branch2a_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_t* _BatchNorm_output_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+67:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_4);
+67:   float* BatchNorm_output_4 = ((float*)(_BatchNorm_output_4->memory));
+67:   const float* bn2b_branch2a_mean = ((const float*)(_bn2b_branch2a_mean->memory));
+67:   const float* bn2b_branch2a_offset = ((const float*)(_bn2b_branch2a_offset->memory));
+67:   const float* bn2b_branch2a_scale = ((const float*)(_bn2b_branch2a_scale->memory));
+67:   const float* bn2b_branch2a_variance = ((const float*)(_bn2b_branch2a_variance->memory));
+67:   const float* var_80 = ((const float*)(_var_80->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         BatchNorm_output_4[((3136 * j) + ((56 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((bn2b_branch2a_variance[j] + 1e-05))) * (bn2b_branch2a_scale[j] * var_80[((3136 * j) + ((56 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((bn2b_branch2a_variance[j] + 1e-05))) * (bn2b_branch2a_scale[j] * bn2b_branch2a_mean[j]))) + bn2b_branch2a_offset[j]));
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _BatchNorm_output_4);
+67: }
+67: 
+67: void fn_relu_15(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_90 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   cinn_buffer_t* _Relu_output_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_malloc((void*)(0), _Relu_output_2);
+67:   float* Relu_output_2 = ((float*)(_Relu_output_2->memory));
+67:   const float* var_90 = ((const float*)(_var_90->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         Relu_output_2[((3136 * j) + ((56 * k) + a))] = cinn_max(var_90[((3136 * j) + ((56 * k) + a))], 0);
+67:       };
+67:     };
+67:   };
+67:   cinn_buffer_free((void*)(0), _Relu_output_2);
+67: }
+67: 
+67: void fn_conv2d_16(void* _args, int32_t num_args)
+67: {
+67:   const cinn_buffer_t* _var_92 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+67:   const cinn_buffer_t* _res2b_branch2b_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+67:   cinn_buffer_t* _T_pad_out_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+67:   cinn_buffer_t* _weights_dilation_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+67:   cinn_buffer_t* _Conv2d_nchw_out_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+67:   cinn_buffer_malloc((void*)(0), _T_pad_out_4);
+67:   cinn_buffer_malloc((void*)(0), _weights_dilation_4);
+67:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out_4);
+67:   float* Conv2d_nchw_out_4 = ((float*)(_Conv2d_nchw_out_4->memory));
+67:   float* Conv2d_nchw_out_4_init = ((float*)(_Conv2d_nchw_out_4->memory));
+67:   float* T_pad_out_4 = ((float*)(_T_pad_out_4->memory));
+67:   const float* res2b_branch2b_weights = ((const float*)(_res2b_branch2b_weights->memory));
+67:   const float* var_92 = ((const float*)(_var_92->memory));
+67:   float* weights_dilation_4 = ((float*)(_weights_dilation_4->memory));
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 56; k += 1) {
+67:       for (int32_t a = 0; a < 56; a += 1) {
+67:         Conv2d_nchw_out_4_init[((3136 * j) + ((56 * k) + a))] = 0;
+67:       };
+67:     };
+67:   };
+67:   for (int32_t i = 0; i < 64; i += 1) {
+67:     for (int32_t j = 0; j < 64; j += 1) {
+67:       for (int32_t k = 0; k < 3; k += 1) {
+67:         for (int32_t a = 0; a < 3; a += 1) {
+67:           weights_dilation_4[((576 * i) + ((9 * j) + ((3 * k) + a)))] = (((((a % 1) == 0) && ((k % 1) == 0))) ? res2b_branch2b_weights[((a / 1) + (((k / 1) * 3) + ((576 * i) + (9 * j))))] : 0);
+67:         };
+67:       };
+67:     };
+67:   };
+67:   for (int32_t j = 0; j < 64; j += 1) {
+67:     for (int32_t k = 0; k < 58; k += 1) {
+67:       for (int32_t a = 0; a < 58; a += 1) {
+67:         T_pad_out_4[((3364 *
+67: I0924 13:32:26.094007 28813 codegen_llvm.cc:344] instr: i32 49
+67: W0924 13:32:26.746232 28813 init.cc:226] Warning: PaddlePaddle catches a failure signal, it may not work properly
+67: W0924 13:32:26.746279 28813 init.cc:228] You could check whether you killed PaddlePaddle thread/process accidentally or report the case to PaddlePaddle
+67: W0924 13:32:26.746281 28813 init.cc:231] The detail failure signal is:
+67: 
+67: W0924 13:32:26.746284 28813 init.cc:234] *** Aborted at 1600954346 (unix time) try "date -d @1600954346" if you are using GNU date ***
+67: W0924 13:32:26.747192 28813 init.cc:234] PC: @                0x0 (unknown)
+67: W0924 13:32:26.747408 28813 init.cc:234] *** SIGSEGV (@0x7fb48a9f8e4c) received by PID 28813 (TID 0x7fb4b0ce7740) from PID 18446744071740296780; stack trace: ***
+67: W0924 13:32:26.748113 28813 init.cc:234]     @     0x7fb4b071af20 (unknown)
+67: W0924 13:32:26.748739 28813 init.cc:234]     @     0x7fb4b0b99661 (unknown)
+67: W0924 13:32:26.750941 28813 init.cc:234]     @     0x7fb347c177d5 cinn::hlir::framework::Instruction::Run()
+67: W0924 13:32:26.752528 28813 init.cc:234]     @     0x7fb347c178f2 cinn::hlir::framework::Program::Execute()
+67: W0924 13:32:26.754045 28813 init.cc:234]     @     0x7fb347ea62d6 cinn::frontend::Executor::Run()
+67: W0924 13:32:26.754400 28813 init.cc:234]     @     0x7fb347c216d9 _ZZN8pybind1112cpp_functionC4IvN4cinn8frontend8ExecutorEJEJNS_4nameENS_9is_methodENS_7siblingEEEEMT0_FT_DpT1_EDpRKT2_ENKUlPS4_E_clESI_
+67: W0924 13:32:26.755192 28813 init.cc:234]     @     0x7fb347c34443 _ZN8pybind116detail15argument_loaderIJPN4cinn8frontend8ExecutorEEE9call_implIvRZNS_12cpp_functionC4IvS4_JEJNS_4nameENS_9is_methodENS_7siblingEEEEMT0_FT_DpT1_EDpRKT2_EUlS5_E_JLm0EENS0_9void_typeEEESE_OSD_St16integer_sequenceImJXspT1_EEEOT2_
+67: W0924 13:32:26.755627 28813 init.cc:234]     @     0x7fb347c2deec _ZNO8pybind116detail15argument_loaderIJPN4cinn8frontend8ExecutorEEE4callIvNS0_9void_typeERZNS_12cpp_functionC4IvS4_JEJNS_4nameENS_9is_methodENS_7siblingEEEEMT0_FT_DpT1_EDpRKT2_EUlS5_E_EENSt9enable_ifIXsrSt7is_voidISF_E5valueES8_E4typeEOT1_
+67: W0924 13:32:26.756399 28813 init.cc:234]     @     0x7fb347c29ac6 _ZZN8pybind1112cpp_function10initializeIZNS0_C4IvN4cinn8frontend8ExecutorEJEJNS_4nameENS_9is_methodENS_7siblingEEEEMT0_FT_DpT1_EDpRKT2_EUlPS5_E_vJSJ_EJS6_S7_S8_EEEvOSA_PFS9_SC_ESI_ENKUlRNS_6detail13function_callEE1_clESQ_
+67: W0924 13:32:26.756747 28813 init.cc:234]     @     0x7fb347c29bf3 _ZZN8pybind1112cpp_function10initializeIZNS0_C4IvN4cinn8frontend8ExecutorEJEJNS_4nameENS_9is_methodENS_7siblingEEEEMT0_FT_DpT1_EDpRKT2_EUlPS5_E_vJSJ_EJS6_S7_S8_EEEvOSA_PFS9_SC_ESI_ENUlRNS_6detail13function_callEE1_4_FUNESQ_
+67: W0924 13:32:26.757316 28813 init.cc:234]     @     0x7fb34784e4d9 pybind11::cpp_function::dispatcher()
+67: W0924 13:32:26.757386 28813 init.cc:234]     @           0x5674fc _PyCFunction_FastCallDict
+67: W0924 13:32:26.757441 28813 init.cc:234]     @           0x50abb3 (unknown)
+67: W0924 13:32:26.757491 28813 init.cc:234]     @           0x50c5b9 _PyEval_EvalFrameDefault
+67: W0924 13:32:26.757541 28813 init.cc:234]     @           0x509d48 (unknown)
+67: W0924 13:32:26.757591 28813 init.cc:234]     @           0x50aa7d (unknown)
+67: W0924 13:32:26.757639 28813 init.cc:234]     @           0x50c5b9 _PyEval_EvalFrameDefault
+67: W0924 13:32:26.757689 28813 init.cc:234]     @           0x508245 (unknown)
+67: W0924 13:32:26.757722 28813 init.cc:234]     @           0x509642 _PyFunction_FastCallDict
+67: W0924 13:32:26.757778 28813 init.cc:234]     @           0x595311 (unknown)
+67: W0924 13:32:26.757822 28813 init.cc:234]     @           0x5a067e PyObject_Call
+67: W0924 13:32:26.757869 28813 init.cc:234]     @           0x50d966 _PyEval_EvalFrameDefault
+67: W0924 13:32:26.757920 28813 init.cc:234]     @           0x508245 (unknown)
+67: W0924 13:32:26.757954 28813 init.cc:234]     @           0x509642 _PyFunction_FastCallDict
+67: W0924 13:32:26.758008 28813 init.cc:234]     @           0x595311 (unknown)
+67: W0924 13:32:26.758064 28813 init.cc:234]     @           0x54b1e1 (unknown)
+67: W0924 13:32:26.758112 28813 init.cc:234]     @           0x5aa6ec _PyObject_FastCallKeywords
+67: W0924 13:32:26.758174 28813 init.cc:234]     @           0x50abb3 (unknown)
+67: W0924 13:32:26.758225 28813 init.cc:234]     @           0x50c5b9 _PyEval_EvalFrameDefault
+67: W0924 13:32:26.758278 28813 init.cc:234]     @           0x508245 (unknown)
+67: W0924 13:32:26.758311 28813 init.cc:234]     @           0x509642 _PyFunction_FastCallDict
+67: W0924 13:32:26.758368 28813 init.cc:234]     @           0x595311 (unknown)
+67: Segmentation fault
+65/68 Test #67: test_cinn_real_resnet18 ................***Failed    5.96 sec
+57: ./home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+57: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+57: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+57: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+57: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+68: I0924 13:32:27.612771 28814 nn.cc:477] kernel_size length is: 2
+68: I0924 13:32:27.612797 28814 nn.cc:478] kernel_size is: 7
+68: I0924 13:32:27.612800 28814 nn.cc:479] padding_size length is: 4
+68: I0924 13:32:27.612802 28814 nn.cc:480] padding_size is: 0
+68: I0924 13:32:28.877796 28814 graph_compiler.cc:39] [Debug] C Code is:
+68: #include <cinn_runtime.h>
+68: #include <stdio.h>
+68: 
+68: void fn_conv2d_0(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _image = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv1_1_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_t* _T_pad_out_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   cinn_buffer_t* _weights_dilation = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   cinn_buffer_t* _Conv2d_nchw_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_malloc((void*)(0), _T_pad_out_0);
+68:   cinn_buffer_malloc((void*)(0), _weights_dilation);
+68:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out);
+68:   float* Conv2d_nchw_out = ((float*)(_Conv2d_nchw_out->memory));
+68:   float* Conv2d_nchw_out_init = ((float*)(_Conv2d_nchw_out->memory));
+68:   float* T_pad_out_0 = ((float*)(_T_pad_out_0->memory));
+68:   const float* conv1_1_weights = ((const float*)(_conv1_1_weights->memory));
+68:   const float* image = ((const float*)(_image->memory));
+68:   float* weights_dilation = ((float*)(_weights_dilation->memory));
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         Conv2d_nchw_out_init[((12544 * j) + ((112 * k) + a))] = 0;
+68:       };
+68:     };
+68:   };
+68:   for (int32_t i = 0; i < 32; i += 1) {
+68:     for (int32_t j = 0; j < 3; j += 1) {
+68:       for (int32_t k = 0; k < 3; k += 1) {
+68:         for (int32_t a = 0; a < 3; a += 1) {
+68:           weights_dilation[((27 * i) + ((9 * j) + ((3 * k) + a)))] = (((((a % 1) == 0) && ((k % 1) == 0))) ? conv1_1_weights[((a / 1) + (((k / 1) * 3) + ((27 * i) + (9 * j))))] : 0);
+68:         };
+68:       };
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 3; j += 1) {
+68:     for (int32_t k = 0; k < 226; k += 1) {
+68:       for (int32_t a = 0; a < 226; a += 1) {
+68:         T_pad_out_0[((51076 * j) + ((226 * k) + a))] = ((((a < 225) && ((a >= 1) && ((k < 225) && (k >= 1))))) ? image[(-225 + ((50176 * j) + ((224 * k) + a)))] : 0);
+68:       };
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         for (int32_t fc = 0; fc < 3; fc += 1) {
+68:           for (int32_t fy = 0; fy < 3; fy += 1) {
+68:             for (int32_t fx = 0; fx < 3; fx += 1) {
+68:               Conv2d_nchw_out[((12544 * j) + ((112 * k) + a))] = (Conv2d_nchw_out[((12544 * j) + ((112 * k) + a))] + (T_pad_out_0[((2 * a) + ((51076 * fc) + ((226 * fy) + ((452 * k) + fx))))] * weights_dilation[((9 * fc) + ((3 * fy) + ((27 * j) + fx)))]));
+68:             };
+68:           };
+68:         };
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _T_pad_out_0);
+68:   cinn_buffer_free((void*)(0), _weights_dilation);
+68:   cinn_buffer_free((void*)(0), _Conv2d_nchw_out);
+68: }
+68: 
+68: void fn_batchnorm_1(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_5 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv1_1_bn_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   const cinn_buffer_t* _conv1_1_bn_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   const cinn_buffer_t* _conv1_1_bn_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   const cinn_buffer_t* _conv1_1_bn_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_t* _BatchNorm_output_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+68:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_0);
+68:   float* BatchNorm_output_0 = ((float*)(_BatchNorm_output_0->memory));
+68:   const float* conv1_1_bn_mean = ((const float*)(_conv1_1_bn_mean->memory));
+68:   const float* conv1_1_bn_offset = ((const float*)(_conv1_1_bn_offset->memory));
+68:   const float* conv1_1_bn_scale = ((const float*)(_conv1_1_bn_scale->memory));
+68:   const float* conv1_1_bn_variance = ((const float*)(_conv1_1_bn_variance->memory));
+68:   const float* var_5 = ((const float*)(_var_5->memory));
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         BatchNorm_output_0[((12544 * j) + ((112 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((conv1_1_bn_variance[j] + 1e-05))) * (conv1_1_bn_scale[j] * var_5[((12544 * j) + ((112 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((conv1_1_bn_variance[j] + 1e-05))) * (conv1_1_bn_scale[j] * conv1_1_bn_mean[j]))) + conv1_1_bn_offset[j]));
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _BatchNorm_output_0);
+68: }
+68: 
+68: void fn_relu6_2(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_15 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   cinn_buffer_t* _Relu6_output = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_malloc((void*)(0), _Relu6_output);
+68:   float* Relu6_output = ((float*)(_Relu6_output->memory));
+68:   const float* var_15 = ((const float*)(_var_15->memory));
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         Relu6_output[((12544 * j) + ((112 * k) + a))] = cinn_min(cinn_max(var_15[((12544 * j) + ((112 * k) + a))], 0), 6);
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _Relu6_output);
+68: }
+68: 
+68: void fn_conv2d_3(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_17 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv2_1_expand_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_t* _T_Identity_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   cinn_buffer_t* _weights_dilation_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   cinn_buffer_t* _Conv2d_nchw_out_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_malloc((void*)(0), _T_Identity_out);
+68:   cinn_buffer_malloc((void*)(0), _weights_dilation_0);
+68:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out_0);
+68:   float* Conv2d_nchw_out_0 = ((float*)(_Conv2d_nchw_out_0->memory));
+68:   float* Conv2d_nchw_out_0_init = ((float*)(_Conv2d_nchw_out_0->memory));
+68:   float* T_Identity_out = ((float*)(_T_Identity_out->memory));
+68:   const float* conv2_1_expand_weights = ((const float*)(_conv2_1_expand_weights->memory));
+68:   const float* var_17 = ((const float*)(_var_17->memory));
+68:   float* weights_dilation_0 = ((float*)(_weights_dilation_0->memory));
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         Conv2d_nchw_out_0_init[((12544 * j) + ((112 * k) + a))] = 0;
+68:       };
+68:     };
+68:   };
+68:   for (int32_t i = 0; i < 32; i += 1) {
+68:     for (int32_t j = 0; j < 32; j += 1) {
+68:       weights_dilation_0[((32 * i) + j)] = (((((0 % 1) == 0) && ((0 % 1) == 0))) ? conv2_1_expand_weights[((32 * i) + j)] : 0);
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         T_Identity_out[((12544 * j) + ((112 * k) + a))] = var_17[((12544 * j) + ((112 * k) + a))];
+68:       };
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         for (int32_t fc_0 = 0; fc_0 < 32; fc_0 += 1) {
+68:           Conv2d_nchw_out_0[((12544 * j) + ((112 * k) + a))] = (Conv2d_nchw_out_0[((12544 * j) + ((112 * k) + a))] + (T_Identity_out[((12544 * fc_0) + ((112 * k) + a))] * weights_dilation_0[((32 * j) + fc_0)]));
+68:         };
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _T_Identity_out);
+68:   cinn_buffer_free((void*)(0), _weights_dilation_0);
+68:   cinn_buffer_free((void*)(0), _Conv2d_nchw_out_0);
+68: }
+68: 
+68: void fn_batchnorm_4(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_23 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv2_1_expand_bn_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   const cinn_buffer_t* _conv2_1_expand_bn_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   const cinn_buffer_t* _conv2_1_expand_bn_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   const cinn_buffer_t* _conv2_1_expand_bn_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_t* _BatchNorm_output_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+68:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_1);
+68:   float* BatchNorm_output_1 = ((float*)(_BatchNorm_output_1->memory));
+68:   const float* conv2_1_expand_bn_mean = ((const float*)(_conv2_1_expand_bn_mean->memory));
+68:   const float* conv2_1_expand_bn_offset = ((const float*)(_conv2_1_expand_bn_offset->memory));
+68:   const float* conv2_1_expand_bn_scale = ((const float*)(_conv2_1_expand_bn_scale->memory));
+68:   const float* conv2_1_expand_bn_variance = ((const float*)(_conv2_1_expand_bn_variance->memory));
+68:   const float* var_23 = ((const float*)(_var_23->memory));
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         BatchNorm_output_1[((12544 * j) + ((112 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((conv2_1_expand_bn_variance[j] + 1e-05))) * (conv2_1_expand_bn_scale[j] * var_23[((12544 * j) + ((112 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((conv2_1_expand_bn_variance[j] + 1e-05))) * (conv2_1_expand_bn_scale[j] * conv2_1_expand_bn_mean[j]))) + conv2_1_expand_bn_offset[j]));
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _BatchNorm_output_1);
+68: }
+68: 
+68: void fn_relu6_5(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_33 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   cinn_buffer_t* _Relu6_output_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_malloc((void*)(0), _Relu6_output_0);
+68:   float* Relu6_output_0 = ((float*)(_Relu6_output_0->memory));
+68:   const float* var_33 = ((const float*)(_var_33->memory));
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         Relu6_output_0[((12544 * j) + ((112 * k) + a))] = cinn_min(cinn_max(var_33[((12544 * j) + ((112 * k) + a))], 0), 6);
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _Relu6_output_0);
+68: }
+68: 
+68: void fn_depthwise_conv2d_6(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_35 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv2_1_dwise_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_t* _T_pad_out_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   cinn_buffer_t* _T_depthwise_conv2d_nchw_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   cinn_buffer_malloc((void*)(0), _T_pad_out_1);
+68:   cinn_buffer_malloc((void*)(0), _T_depthwise_conv2d_nchw_out);
+68:   float* T_depthwise_conv2d_nchw_out = ((float*)(_T_depthwise_conv2d_nchw_out->memory));
+68:   float* T_depthwise_conv2d_nchw_out_init = ((float*)(_T_depthwise_conv2d_nchw_out->memory));
+68:   float* T_pad_out_1 = ((float*)(_T_pad_out_1->memory));
+68:   const float* conv2_1_dwise_weights = ((const float*)(_conv2_1_dwise_weights->memory));
+68:   const float* var_35 = ((const float*)(_var_35->memory));
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         T_depthwise_conv2d_nchw_out_init[((12544 * j) + ((112 * k) + a))] = 0;
+68:       };
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 114; k += 1) {
+68:       for (int32_t a = 0; a < 114; a += 1) {
+68:         T_pad_out_1[((12996 * j) + ((114 * k) + a))] = ((((a < 113) && ((a >= 1) && ((k < 113) && (k >= 1))))) ? var_35[(-113 + ((12544 * j) + ((112 * k) + a)))] : 0);
+68:       };
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         for (int32_t kh = 0; kh < 3; kh += 1) {
+68:           for (int32_t kw = 0; kw < 3; kw += 1) {
+68:             T_depthwise_conv2d_nchw_out[((12544 * j) + ((112 * k) + a))] = (T_depthwise_conv2d_nchw_out[((12544 * j) + ((112 * k) + a))] + (T_pad_out_1[(((j / 1) * 12996) + ((114 * k) + ((114 * kh) + (a + kw))))] * conv2_1_dwise_weights[(((j / 1) * 9) + ((3 * kh) + kw))]));
+68:           };
+68:         };
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _T_pad_out_1);
+68:   cinn_buffer_free((void*)(0), _T_depthwise_conv2d_nchw_out);
+68: }
+68: 
+68: void fn_batchnorm_7(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_40 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv2_1_dwise_bn_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   const cinn_buffer_t* _conv2_1_dwise_bn_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   const cinn_buffer_t* _conv2_1_dwise_bn_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   const cinn_buffer_t* _conv2_1_dwise_bn_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_t* _BatchNorm_output_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+68:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_2);
+68:   float* BatchNorm_output_2 = ((float*)(_BatchNorm_output_2->memory));
+68:   const float* conv2_1_dwise_bn_mean = ((const float*)(_conv2_1_dwise_bn_mean->memory));
+68:   const float* conv2_1_dwise_bn_offset = ((const float*)(_conv2_1_dwise_bn_offset->memory));
+68:   const float* conv2_1_dwise_bn_scale = ((const float*)(_conv2_1_dwise_bn_scale->memory));
+68:   const float* conv2_1_dwise_bn_variance = ((const float*)(_conv2_1_dwise_bn_variance->memory));
+68:   const float* var_40 = ((const float*)(_var_40->memory));
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         BatchNorm_output_2[((12544 * j) + ((112 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((conv2_1_dwise_bn_variance[j] + 1e-05))) * (conv2_1_dwise_bn_scale[j] * var_40[((12544 * j) + ((112 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((conv2_1_dwise_bn_variance[j] + 1e-05))) * (conv2_1_dwise_bn_scale[j] * conv2_1_dwise_bn_mean[j]))) + conv2_1_dwise_bn_offset[j]));
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _BatchNorm_output_2);
+68: }
+68: 
+68: void fn_relu6_8(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_50 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   cinn_buffer_t* _Relu6_output_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_malloc((void*)(0), _Relu6_output_1);
+68:   float* Relu6_output_1 = ((float*)(_Relu6_output_1->memory));
+68:   const float* var_50 = ((const float*)(_var_50->memory));
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         Relu6_output_1[((12544 * j) + ((112 * k) + a))] = cinn_min(cinn_max(var_50[((12544 * j) + ((112 * k) + a))], 0), 6);
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _Relu6_output_1);
+68: }
+68: 
+68: void fn_conv2d_9(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_52 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv2_1_linear_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_t* _T_Identity_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   cinn_buffer_t* _weights_dilation_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   cinn_buffer_t* _Conv2d_nchw_out_1 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_malloc((void*)(0), _T_Identity_out);
+68:   cinn_buffer_malloc((void*)(0), _weights_dilation_1);
+68:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out_1);
+68:   float* Conv2d_nchw_out_1 = ((float*)(_Conv2d_nchw_out_1->memory));
+68:   float* Conv2d_nchw_out_1_init = ((float*)(_Conv2d_nchw_out_1->memory));
+68:   float* T_Identity_out = ((float*)(_T_Identity_out->memory));
+68:   const float* conv2_1_linear_weights = ((const float*)(_conv2_1_linear_weights->memory));
+68:   const float* var_52 = ((const float*)(_var_52->memory));
+68:   float* weights_dilation_1 = ((float*)(_weights_dilation_1->memory));
+68:   for (int32_t j = 0; j < 16; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         Conv2d_nchw_out_1_init[((12544 * j) + ((112 * k) + a))] = 0;
+68:       };
+68:     };
+68:   };
+68:   for (int32_t i = 0; i < 16; i += 1) {
+68:     for (int32_t j = 0; j < 32; j += 1) {
+68:       weights_dilation_1[((32 * i) + j)] = (((((0 % 1) == 0) && ((0 % 1) == 0))) ? conv2_1_linear_weights[((32 * i) + j)] : 0);
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 32; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         T_Identity_out[((12544 * j) + ((112 * k) + a))] = var_52[((12544 * j) + ((112 * k) + a))];
+68:       };
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 16; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         for (int32_t fc_1 = 0; fc_1 < 32; fc_1 += 1) {
+68:           Conv2d_nchw_out_1[((12544 * j) + ((112 * k) + a))] = (Conv2d_nchw_out_1[((12544 * j) + ((112 * k) + a))] + (T_Identity_out[((12544 * fc_1) + ((112 * k) + a))] * weights_dilation_1[((32 * j) + fc_1)]));
+68:         };
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _T_Identity_out);
+68:   cinn_buffer_free((void*)(0), _weights_dilation_1);
+68:   cinn_buffer_free((void*)(0), _Conv2d_nchw_out_1);
+68: }
+68: 
+68: void fn_batchnorm_10(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_58 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv2_1_linear_bn_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   const cinn_buffer_t* _conv2_1_linear_bn_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   const cinn_buffer_t* _conv2_1_linear_bn_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   const cinn_buffer_t* _conv2_1_linear_bn_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_t* _BatchNorm_output_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+68:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_3);
+68:   float* BatchNorm_output_3 = ((float*)(_BatchNorm_output_3->memory));
+68:   const float* conv2_1_linear_bn_mean = ((const float*)(_conv2_1_linear_bn_mean->memory));
+68:   const float* conv2_1_linear_bn_offset = ((const float*)(_conv2_1_linear_bn_offset->memory));
+68:   const float* conv2_1_linear_bn_scale = ((const float*)(_conv2_1_linear_bn_scale->memory));
+68:   const float* conv2_1_linear_bn_variance = ((const float*)(_conv2_1_linear_bn_variance->memory));
+68:   const float* var_58 = ((const float*)(_var_58->memory));
+68:   for (int32_t j = 0; j < 16; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         BatchNorm_output_3[((12544 * j) + ((112 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((conv2_1_linear_bn_variance[j] + 1e-05))) * (conv2_1_linear_bn_scale[j] * var_58[((12544 * j) + ((112 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((conv2_1_linear_bn_variance[j] + 1e-05))) * (conv2_1_linear_bn_scale[j] * conv2_1_linear_bn_mean[j]))) + conv2_1_linear_bn_offset[j]));
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _BatchNorm_output_3);
+68: }
+68: 
+68: void fn_conv2d_11(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_68 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv3_1_expand_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_t* _T_Identity_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   cinn_buffer_t* _weights_dilation_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   cinn_buffer_t* _Conv2d_nchw_out_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_malloc((void*)(0), _T_Identity_out);
+68:   cinn_buffer_malloc((void*)(0), _weights_dilation_2);
+68:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out_2);
+68:   float* Conv2d_nchw_out_2 = ((float*)(_Conv2d_nchw_out_2->memory));
+68:   float* Conv2d_nchw_out_2_init = ((float*)(_Conv2d_nchw_out_2->memory));
+68:   float* T_Identity_out = ((float*)(_T_Identity_out->memory));
+68:   const float* conv3_1_expand_weights = ((const float*)(_conv3_1_expand_weights->memory));
+68:   const float* var_68 = ((const float*)(_var_68->memory));
+68:   float* weights_dilation_2 = ((float*)(_weights_dilation_2->memory));
+68:   for (int32_t j = 0; j < 96; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         Conv2d_nchw_out_2_init[((12544 * j) + ((112 * k) + a))] = 0;
+68:       };
+68:     };
+68:   };
+68:   for (int32_t i = 0; i < 96; i += 1) {
+68:     for (int32_t j = 0; j < 16; j += 1) {
+68:       weights_dilation_2[((16 * i) + j)] = (((((0 % 1) == 0) && ((0 % 1) == 0))) ? conv3_1_expand_weights[((16 * i) + j)] : 0);
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 16; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         T_Identity_out[((12544 * j) + ((112 * k) + a))] = var_68[((12544 * j) + ((112 * k) + a))];
+68:       };
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 96; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         for (int32_t fc_2 = 0; fc_2 < 16; fc_2 += 1) {
+68:           Conv2d_nchw_out_2[((12544 * j) + ((112 * k) + a))] = (Conv2d_nchw_out_2[((12544 * j) + ((112 * k) + a))] + (T_Identity_out[((12544 * fc_2) + ((112 * k) + a))] * weights_dilation_2[((16 * j) + fc_2)]));
+68:         };
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _T_Identity_out);
+68:   cinn_buffer_free((void*)(0), _weights_dilation_2);
+68:   cinn_buffer_free((void*)(0), _Conv2d_nchw_out_2);
+68: }
+68: 
+68: void fn_batchnorm_12(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_74 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv3_1_expand_bn_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   const cinn_buffer_t* _conv3_1_expand_bn_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   const cinn_buffer_t* _conv3_1_expand_bn_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   const cinn_buffer_t* _conv3_1_expand_bn_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_t* _BatchNorm_output_4 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+68:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_4);
+68:   float* BatchNorm_output_4 = ((float*)(_BatchNorm_output_4->memory));
+68:   const float* conv3_1_expand_bn_mean = ((const float*)(_conv3_1_expand_bn_mean->memory));
+68:   const float* conv3_1_expand_bn_offset = ((const float*)(_conv3_1_expand_bn_offset->memory));
+68:   const float* conv3_1_expand_bn_scale = ((const float*)(_conv3_1_expand_bn_scale->memory));
+68:   const float* conv3_1_expand_bn_variance = ((const float*)(_conv3_1_expand_bn_variance->memory));
+68:   const float* var_74 = ((const float*)(_var_74->memory));
+68:   for (int32_t j = 0; j < 96; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         BatchNorm_output_4[((12544 * j) + ((112 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((conv3_1_expand_bn_variance[j] + 1e-05))) * (conv3_1_expand_bn_scale[j] * var_74[((12544 * j) + ((112 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((conv3_1_expand_bn_variance[j] + 1e-05))) * (conv3_1_expand_bn_scale[j] * conv3_1_expand_bn_mean[j]))) + conv3_1_expand_bn_offset[j]));
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _BatchNorm_output_4);
+68: }
+68: 
+68: void fn_relu6_13(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_84 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   cinn_buffer_t* _Relu6_output_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_malloc((void*)(0), _Relu6_output_2);
+68:   float* Relu6_output_2 = ((float*)(_Relu6_output_2->memory));
+68:   const float* var_84 = ((const float*)(_var_84->memory));
+68:   for (int32_t j = 0; j < 96; j += 1) {
+68:     for (int32_t k = 0; k < 112; k += 1) {
+68:       for (int32_t a = 0; a < 112; a += 1) {
+68:         Relu6_output_2[((12544 * j) + ((112 * k) + a))] = cinn_min(cinn_max(var_84[((12544 * j) + ((112 * k) + a))], 0), 6);
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _Relu6_output_2);
+68: }
+68: 
+68: void fn_depthwise_conv2d_14(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_86 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv3_1_dwise_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_t* _T_pad_out_2 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   cinn_buffer_t* _T_depthwise_conv2d_nchw_out_0 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   cinn_buffer_malloc((void*)(0), _T_pad_out_2);
+68:   cinn_buffer_malloc((void*)(0), _T_depthwise_conv2d_nchw_out_0);
+68:   float* T_depthwise_conv2d_nchw_out_0 = ((float*)(_T_depthwise_conv2d_nchw_out_0->memory));
+68:   float* T_depthwise_conv2d_nchw_out_0_init = ((float*)(_T_depthwise_conv2d_nchw_out_0->memory));
+68:   float* T_pad_out_2 = ((float*)(_T_pad_out_2->memory));
+68:   const float* conv3_1_dwise_weights = ((const float*)(_conv3_1_dwise_weights->memory));
+68:   const float* var_86 = ((const float*)(_var_86->memory));
+68:   for (int32_t j = 0; j < 96; j += 1) {
+68:     for (int32_t k = 0; k < 56; k += 1) {
+68:       for (int32_t a = 0; a < 56; a += 1) {
+68:         T_depthwise_conv2d_nchw_out_0_init[((3136 * j) + ((56 * k) + a))] = 0;
+68:       };
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 96; j += 1) {
+68:     for (int32_t k = 0; k < 114; k += 1) {
+68:       for (int32_t a = 0; a < 114; a += 1) {
+68:         T_pad_out_2[((12996 * j) + ((114 * k) + a))] = ((((a < 113) && ((a >= 1) && ((k < 113) && (k >= 1))))) ? var_86[(-113 + ((12544 * j) + ((112 * k) + a)))] : 0);
+68:       };
+68:     };
+68:   };
+68:   for (int32_t j = 0; j < 96; j += 1) {
+68:     for (int32_t k = 0; k < 56; k += 1) {
+68:       for (int32_t a = 0; a < 56; a += 1) {
+68:         for (int32_t kh = 0; kh < 3; kh += 1) {
+68:           for (int32_t kw = 0; kw < 3; kw += 1) {
+68:             T_depthwise_conv2d_nchw_out_0[((3136 * j) + ((56 * k) + a))] = (T_depthwise_conv2d_nchw_out_0[((3136 * j) + ((56 * k) + a))] + (T_pad_out_2[(((j / 1) * 12996) + ((2 * a) + ((228 * k) + ((114 * kh) + kw))))] * conv3_1_dwise_weights[(((j / 1) * 9) + ((3 * kh) + kw))]));
+68:           };
+68:         };
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _T_pad_out_2);
+68:   cinn_buffer_free((void*)(0), _T_depthwise_conv2d_nchw_out_0);
+68: }
+68: 
+68: void fn_batchnorm_15(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_91 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv3_1_dwise_bn_scale = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   const cinn_buffer_t* _conv3_1_dwise_bn_offset = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   const cinn_buffer_t* _conv3_1_dwise_bn_mean = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   const cinn_buffer_t* _conv3_1_dwise_bn_variance = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_t* _BatchNorm_output_5 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[5]));
+68:   cinn_buffer_malloc((void*)(0), _BatchNorm_output_5);
+68:   float* BatchNorm_output_5 = ((float*)(_BatchNorm_output_5->memory));
+68:   const float* conv3_1_dwise_bn_mean = ((const float*)(_conv3_1_dwise_bn_mean->memory));
+68:   const float* conv3_1_dwise_bn_offset = ((const float*)(_conv3_1_dwise_bn_offset->memory));
+68:   const float* conv3_1_dwise_bn_scale = ((const float*)(_conv3_1_dwise_bn_scale->memory));
+68:   const float* conv3_1_dwise_bn_variance = ((const float*)(_conv3_1_dwise_bn_variance->memory));
+68:   const float* var_91 = ((const float*)(_var_91->memory));
+68:   for (int32_t j = 0; j < 96; j += 1) {
+68:     for (int32_t k = 0; k < 56; k += 1) {
+68:       for (int32_t a = 0; a < 56; a += 1) {
+68:         BatchNorm_output_5[((3136 * j) + ((56 * k) + a))] = (((1 / cinn_cpu_sqrt_fp32((conv3_1_dwise_bn_variance[j] + 1e-05))) * (conv3_1_dwise_bn_scale[j] * var_91[((3136 * j) + ((56 * k) + a))])) + ((-1 * ((1 / cinn_cpu_sqrt_fp32((conv3_1_dwise_bn_variance[j] + 1e-05))) * (conv3_1_dwise_bn_scale[j] * conv3_1_dwise_bn_mean[j]))) + conv3_1_dwise_bn_offset[j]));
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _BatchNorm_output_5);
+68: }
+68: 
+68: void fn_relu6_16(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_101 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   cinn_buffer_t* _Relu6_output_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_malloc((void*)(0), _Relu6_output_3);
+68:   float* Relu6_output_3 = ((float*)(_Relu6_output_3->memory));
+68:   const float* var_101 = ((const float*)(_var_101->memory));
+68:   for (int32_t j = 0; j < 96; j += 1) {
+68:     for (int32_t k = 0; k < 56; k += 1) {
+68:       for (int32_t a = 0; a < 56; a += 1) {
+68:         Relu6_output_3[((3136 * j) + ((56 * k) + a))] = cinn_min(cinn_max(var_101[((3136 * j) + ((56 * k) + a))], 0), 6);
+68:       };
+68:     };
+68:   };
+68:   cinn_buffer_free((void*)(0), _Relu6_output_3);
+68: }
+68: 
+68: void fn_conv2d_17(void* _args, int32_t num_args)
+68: {
+68:   const cinn_buffer_t* _var_103 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+68:   const cinn_buffer_t* _conv3_1_linear_weights = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+68:   cinn_buffer_t* _T_Identity_out = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+68:   cinn_buffer_t* _weights_dilation_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[3]));
+68:   cinn_buffer_t* _Conv2d_nchw_out_3 = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[4]));
+68:   cinn_buffer_malloc((void*)(0), _T_Identity_out);
+68:   cinn_buffer_malloc((void*)(0), _weights_dilation_3);
+68:   cinn_buffer_malloc((void*)(0), _Conv2d_nchw_out_3);
+68:   float* Conv2d_nchw_out_3 = ((float*)(_Conv2d_nchw_out_3->memory));
+68:   float* Conv2d_nchw_out_
+68: I0924 13:32:29.892637 28814 codegen_llvm.cc:344] instr: i32 49
+54: I0924 13:32:30.845150 27840 test02_matmul_case.cc:132] 3654.5
+54: I0924 13:32:30.863699 27840 test02_matmul_case.cc:134] Testing matmul_tile
+68: malloc_consolidate(): invalid chunk size
+57: .
+57: ----------------------------------------------------------------------
+57: Ran 2 tests in 14.638s
+57: 
+57: OK
+57: func function matmul (_A, _B, _c)
+57: {
+57:   for (i, 1024)
+57:   {
+57:     for (j, 1024)
+57:     {
+57:       c_init[i, j] = 0
+57:     }
+57:   }
+57:   for (i, 1024)
+57:   {
+57:     for (j, 1024)
+57:     {
+57:       for (k1, 1024)
+57:       {
+57:         c[i, j] = (c[i, j] + (A[i, k1] * B[k1, j]))
+57:       }
+57:     }
+57:   }
+57: }
+57: func function matmul_tile (_A, _B, _c)
+57: {
+57:   for (i, 1024)
+57:   {
+57:     for (j, 1024)
+57:     {
+57:       c_init[i, j] = 0
+57:     }
+57:   }
+57:   for (i_outer, 256)
+57:   {
+57:     for (i_inner, 4)
+57:     {
+57:       for (j_outer, 256)
+57:       {
+57:         for (j_inner, 4)
+57:         {
+57:           for (k1, 1024)
+57:           {
+57:             c[((4 * i_outer) + i_inner), ((4 * j_outer) + j_inner)] = (c[((4 * i_outer) + i_inner), ((4 * j_outer) + j_inner)] + (A[((4 * i_outer) + i_inner), k1] * B[k1, ((4 * j_outer) + j_inner)]))
+57:           }
+57:         }
+57:       }
+57:     }
+57:   }
+57: }
+57: module:
+57:  #include <cinn_runtime.h>
+57: #include <stdio.h>
+57: 
+57: void matmul_tile(void* _args, int32_t num_args)
+57: {
+57:   const cinn_buffer_t* _A = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[0]));
+57:   const cinn_buffer_t* _B = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[1]));
+57:   cinn_buffer_t* _c = cinn_pod_value_to_buffer_p(&(((cinn_pod_value_t*)(_args))[2]));
+57:   cinn_buffer_malloc((void*)(0), _c);
+57:   const float* A = ((const float*)(_A->memory));
+57:   const float* B = ((const float*)(_B->memory));
+57:   float* c = ((float*)(_c->memory));
+57:   float* c_init = ((float*)(_c->memory));
+57:   for (int32_t i = 0; i < 1024; i += 1) {
+57:     for (int32_t j = 0; j < 1024; j += 1) {
+57:       c_init[((1024 * i) + j)] = 0;
+57:     };
+57:   };
+57:   for (int32_t i_outer = 0; i_outer < 256; i_outer += 1) {
+57:     for (int32_t i_inner = 0; i_inner < 4; i_inner += 1) {
+57:       for (int32_t j_outer = 0; j_outer < 256; j_outer += 1) {
+57:         for (int32_t j_inner = 0; j_inner < 4; j_inner += 1) {
+57:           for (int32_t k1 = 0; k1 < 1024; k1 += 1) {
+57:             c[((1024 * i_inner) + ((4096 * i_outer) + ((4 * j_outer) + j_inner)))] = (c[((1024 * i_inner) + ((4096 * i_outer) + ((4 * j_outer) + j_inner)))] + (A[((1024 * i_inner) + ((4096 * i_outer) + k1))] * B[((4 * j_outer) + ((1024 * k1) + j_inner))]));
+57:           };
+57:         };
+57:       };
+57:     };
+57:   };
+57:   cinn_buffer_free((void*)(0), _c);
+57: }
+57: 
+57: 
+66/68 Test #57: test_cinn_python_api ...................   Passed   15.10 sec
+54: I0924 13:32:38.111878 27840 test02_matmul_case.cc:134] 3624
+54: I0924 13:32:38.129998 27840 test02_matmul_case.cc:136] Testing matmul_split
+54: I0924 13:32:45.469558 27840 test02_matmul_case.cc:136] 3669.5
+54: I0924 13:32:45.488063 27840 test02_matmul_case.cc:138] Testing matmul_block
+54: I0924 13:32:46.234335 27840 test02_matmul_case.cc:138] 373
+54: I0924 13:32:46.252069 27840 test02_matmul_case.cc:140] Testing matmul_vectorize
+54: I0924 13:32:46.931332 27840 test02_matmul_case.cc:140] 339.5
+54: I0924 13:32:46.949447 27840 test02_matmul_case.cc:142] Testing matmul_loop_permutation
+54: I0924 13:32:47.270104 27840 test02_matmul_case.cc:142] 160
+54: I0924 13:32:47.287595 27840 test02_matmul_case.cc:144] Testing matmul_array_packing
+54: I0924 13:32:47.568820 27840 test02_matmul_case.cc:144] 140.5
+54: I0924 13:32:47.586277 27840 test02_matmul_case.cc:146] Testing matmul_dynamic_shape
+54: I0924 13:32:54.837365 27840 test02_matmul_case.cc:146] 3625.5
+54: I0924 13:32:54.854187 27840 test02_matmul_case.cc:148] Testing matmul_dynamic_shape_tile
+54: I0924 13:33:02.101543 27840 test02_matmul_case.cc:148] 3623.5
+54: I0924 13:33:02.119555 27840 test02_matmul_case.cc:150] Testing matmul_array_packing_dynamic_shape
+54: /home/wangyue50/CINN-my/CINN/cinn/runtime/cinn_x86_device_impl.cc:21:cinn_x86_malloc(): buf.memory size is 4194304
+54: I0924 13:33:02.403213 27840 test02_matmul_case.cc:150] 141.5
+54: I0924 13:33:02.463613 27840 test02_matmul_case.cc:178] Testing matmul_basic
+54: I0924 13:33:15.219161 27840 test02_matmul_case.cc:178] 6377.5
+54: I0924 13:33:15.293460 27840 test02_matmul_case.cc:179] Testing matmul_tile
+54: I0924 13:33:30.818352 27840 test02_matmul_case.cc:179] 7762
+54: I0924 13:33:30.905640 27840 test02_matmul_case.cc:180] Testing matmul_block
+54: I0924 13:33:40.564250 27840 test02_matmul_case.cc:180] 4829
+54: I0924 13:33:40.663898 27840 test02_matmul_case.cc:181] Testing matmul_vectorize
+54: I0924 13:33:41.966856 27840 test02_matmul_case.cc:181] 651
+54: I0924 13:33:42.108276 27840 test02_matmul_case.cc:182] Testing matmul_loop_permutation
+54: I0924 13:33:42.751489 27840 test02_matmul_case.cc:182] 321.5
+54: I0924 13:33:42.865382 27840 test02_matmul_case.cc:183] Testing matmul_array_packing
+54: I0924 13:33:44.057888 27840 test02_matmul_case.cc:183] 596
+54: warning: <unknown>:0:0: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering
+54: I0924 13:33:44.756733 27840 test02_matmul_case.cc:189] Testing matmul_fn
+54: I0924 13:33:53.072249 27840 test02_matmul_case.cc:189] 4157.5
+54: [       OK ] test02.basic (93555 ms)
+54: [----------] 1 test from test02 (93555 ms total)
+54: 
+54: [----------] Global test environment tear-down
+54: [==========] 1 test from 1 test case ran. (93555 ms total)
+54: [  PASSED  ] 1 test.
+67/68 Test #54: test02_matmul_case .....................   Passed   93.58 sec
diff --git a/Testing/Temporary/CTestCostData.txt b/Testing/Temporary/CTestCostData.txt
new file mode 100644
index 0000000000..ed97d539c0
--- /dev/null
+++ b/Testing/Temporary/CTestCostData.txt
@@ -0,0 +1 @@
+---
diff --git a/Testing/Temporary/LastTest.log b/Testing/Temporary/LastTest.log
new file mode 100644
index 0000000000..cd3e6b3b6b
--- /dev/null
+++ b/Testing/Temporary/LastTest.log
@@ -0,0 +1,3 @@
+Start testing: Sep 25 08:24 UTC
+----------------------------------------------------------
+End testing: Sep 25 08:24 UTC
diff --git a/cinn/hlir/pe/nn.cc b/cinn/hlir/pe/nn.cc
index 31ae8f5da5..531f94f766 100644
--- a/cinn/hlir/pe/nn.cc
+++ b/cinn/hlir/pe/nn.cc
@@ -37,6 +37,55 @@ Tensor PRelu(const Tensor &A, const Tensor &slope, const int axis, const std::st
       output_name);
 }
 
+// std::vector<ir::Tensor> Conv2d_NCHW(const ir::Tensor &input,
+//                                     const ir::Tensor &weights,
+//                                     int pad_h,
+//                                     int pad_w,
+//                                     int stride_h,
+//                                     int stride_w,
+//                                     int dilation_h,
+//                                     int dilation_w,
+//                                     const std::vector<std::vector<int>> &output_shapes,
+//                                     const std::string &output_name) {
+//   CHECK_EQ(4, input->shape.size()) << "Input's dimension of Conv2d op is not 4! Please check.";
+//   CHECK_EQ(4, weights->shape.size()) << "Weight's dimension of Conv2d op is not 4! Please check.";
+//   CHECK_EQ(3, output_shapes.size()) << "The size of output_shapes of Conv2d op is not 3! Please check.";
+//   CHECK_EQ(4, output_shapes[0].size()) << "The size of output_shapes[0] of Conv2d op is not 4! Please check.";
+//   CHECK_EQ(4, output_shapes[1].size()) << "The size of output_shapes[1] of Conv2d op is not 4! Please check.";
+//   CHECK_EQ(4, output_shapes[2].size()) << "The size of output_shapes[2] of Conv2d op is not 4! Please check.";
+//   std::vector<Expr> output_shape{
+//       Expr(output_shapes[2][0]), Expr(output_shapes[2][1]), Expr(output_shapes[2][2]), Expr(output_shapes[2][3])};
+//   auto input_pad = Compute(
+//       {Expr(output_shapes[0][0]), Expr(output_shapes[0][1]), Expr(output_shapes[0][2]), Expr(output_shapes[0][3])},
+//       [=](Expr nn, Expr cc, Expr yy, Expr xx) {
+//         auto cond =
+//             ir::logic_and({yy >= pad_h, yy - pad_h < input->shape[2], xx >= pad_w, xx - pad_w < input->shape[3]});
+//         return ir::Select::Make(cond, input(nn, cc, yy - pad_h, xx - pad_w), Expr(0.f));
+//       },
+//       UniqName("input_pad"));
+//   auto weights_dilation = Compute(
+//       {Expr(output_shapes[1][0]), Expr(output_shapes[1][1]), Expr(output_shapes[1][2]), Expr(output_shapes[1][3])},
+//       [=](Expr nn, Expr cc, Expr yy, Expr xx) {
+//         auto cond = ir::logic_and({(xx) % dilation_h == 0, yy % dilation_w == 0});
+//         return ir::Select::Make(cond, weights(nn, cc, yy / dilation_h, xx / dilation_w), Expr(0.f));
+//       },
+//       UniqName("weights_dilation"));
+
+//   Var rc(input_pad->shape[1], UniqName("rc"));
+//   Var ry(weights_dilation->shape[2], UniqName("ry"));
+//   Var rx(weights_dilation->shape[3], UniqName("rx"));
+
+//   auto res = Compute(output_shape,
+//                      [=](Expr nn, Expr ff, Expr yy, Expr xx) {
+//                        return ir::ReduceSum(
+//                            input_pad(nn, rc, yy * stride_h + ry, xx * stride_w + rx) * weights_dilation(ff, rc, ry,
+//                            rx), Expr(0.f));
+//                      },
+//                      output_name,
+//                      {ry, rx, rc});
+//   return {input_pad, weights_dilation, res};
+// }
+
 std::vector<ir::Tensor> Conv2d_NCHW(const ir::Tensor &input,
                                     const ir::Tensor &weights,
                                     int pad_h,
@@ -51,6 +100,7 @@ std::vector<ir::Tensor> Conv2d_NCHW(const ir::Tensor &input,
   CHECK_EQ(4, weights->shape.size()) << "Weight's dimension of Conv2d_NCHW op is not 4! Please check.";
   std::vector<Expr> output_shape;
   std::vector<Expr> new_weights_shape;
+  std::vector<Expr> input_pad_shape;
   if (output_shapes.size() == 3) {
     // already computed by infer_shape
     CHECK_EQ(4, output_shapes[0].size()) << "The size of output_shapes[0] of Conv2d op is not 4! Please check.";
@@ -60,6 +110,8 @@ std::vector<ir::Tensor> Conv2d_NCHW(const ir::Tensor &input,
         Expr(output_shapes[2][0]), Expr(output_shapes[2][1]), Expr(output_shapes[2][2]), Expr(output_shapes[2][3])};
     new_weights_shape = {
         Expr(output_shapes[1][0]), Expr(output_shapes[1][1]), Expr(output_shapes[1][2]), Expr(output_shapes[1][3])};
+    input_pad_shape = {
+        Expr(output_shapes[0][0]), Expr(output_shapes[0][1]), Expr(output_shapes[0][2]), Expr(output_shapes[0][3])};
   } else {
     output_shape = {
         input->shape[0],                                                                                  // B
@@ -71,10 +123,16 @@ std::vector<ir::Tensor> Conv2d_NCHW(const ir::Tensor &input,
                          weights->shape[1],
                          dilation_h * (weights->shape[2] - 1) + 1,
                          dilation_w * (weights->shape[3] - 1) + 1};
+    input_pad_shape   = {input->shape[0], input->shape[1], input->shape[2] + 2 * pad_h, input->shape[3] + 2 * pad_w};
   }
-  auto input_pad =
-      (pad_h == 0 && pad_w == 0) ? Identity(input) : Pad(input, {Expr(0), Expr(0), Expr(pad_h), Expr(pad_w)});
-
+  auto input_pad = Compute(
+      input_pad_shape,
+      [=](Expr nn, Expr cc, Expr yy, Expr xx) {
+        auto cond =
+            ir::logic_and({yy >= pad_h, yy - pad_h < input->shape[2], xx >= pad_w, xx - pad_w < input->shape[3]});
+        return ir::Select::Make(cond, input(nn, cc, yy - pad_h, xx - pad_w), ir::Zero(input->type()));
+      },
+      UniqName("input_pad"));
   auto weights_dilation = Compute(
       new_weights_shape,
       [=](Expr nn, Expr cc, Expr yy, Expr xx) {
@@ -120,6 +178,7 @@ std::vector<ir::Tensor> Conv2d_NHWC(const ir::Tensor &input,
   CHECK_EQ(4, weights->shape.size()) << "Weight's dimension of Conv2d_NHWC op is not 4! Please check.";
   std::vector<Expr> output_shape;
   std::vector<Expr> new_weights_shape;
+  std::vector<Expr> input_pad_shape;
   if (output_shapes.size() == 3) {
     // already computed by infer_shape
     CHECK_EQ(4, output_shapes[0].size()) << "The size of output_shapes[0] of Conv2d op is not 4! Please check.";
@@ -129,6 +188,8 @@ std::vector<ir::Tensor> Conv2d_NHWC(const ir::Tensor &input,
         Expr(output_shapes[2][0]), Expr(output_shapes[2][1]), Expr(output_shapes[2][2]), Expr(output_shapes[2][3])};
     new_weights_shape = {
         Expr(output_shapes[1][0]), Expr(output_shapes[1][1]), Expr(output_shapes[1][2]), Expr(output_shapes[1][3])};
+    input_pad_shape = {
+        Expr(output_shapes[0][0]), Expr(output_shapes[0][1]), Expr(output_shapes[0][2]), Expr(output_shapes[0][3])};
   } else {
     output_shape = {
         input->shape[0],                                                                                  // B
@@ -140,9 +201,16 @@ std::vector<ir::Tensor> Conv2d_NHWC(const ir::Tensor &input,
                          weights->shape[1],
                          dilation_h * (weights->shape[2] - 1) + 1,
                          dilation_w * (weights->shape[3] - 1) + 1};
+    input_pad_shape   = {input->shape[0], input->shape[1] + 2 * pad_h, input->shape[2] + 2 * pad_w, input->shape[3]};
   }
-  auto input_pad =
-      (pad_h == 0 && pad_w == 0) ? Identity(input) : Pad(input, {Expr(0), Expr(pad_h), Expr(pad_w), Expr(0)});
+  auto input_pad = Compute(
+      input_pad_shape,
+      [=](Expr nn, Expr yy, Expr xx, Expr cc) {
+        auto cond =
+            ir::logic_and({yy >= pad_h, yy - pad_h < input->shape[1], xx >= pad_w, xx - pad_w < input->shape[2]});
+        return ir::Select::Make(cond, input(nn, yy - pad_h, xx - pad_w, cc), ir::Zero(input->type()));
+      },
+      UniqName("input_pad"));
 
   auto weights_dilation = Compute(
       new_weights_shape,
diff --git a/python/tests/conv2d_utils.py b/python/tests/conv2d_utils.py
index 636968aef6..02c9f880ae 100644
--- a/python/tests/conv2d_utils.py
+++ b/python/tests/conv2d_utils.py
@@ -73,10 +73,10 @@ def conv2d_native(inputs_data, input_shape, filter_size, attrs, is_depthwise):
 
     res_shape = output.shape[1:]
     pad_shape = list(input_shape)
-    dialtion_shape = list(filter_size_new)
+    dilation_shape = list(filter_size_new)
     assert len(padding) == 2
     assert len(pad_shape) == 4
-    assert len(dialtion_shape) == 4
+    assert len(dilation_shape) == 4
     if data_format == "NCHW":
         h_index = 2
         w_index = 3
@@ -86,11 +86,12 @@ def conv2d_native(inputs_data, input_shape, filter_size, attrs, is_depthwise):
 
     pad_shape[h_index] += 2 * padding[0]
     pad_shape[w_index] += 2 * padding[1]
-    dialtion_shape[h_index] = (filter_size_new[h_index] - 1) * dilation[0] + 1
-    dialtion_shape[w_index] = (filter_size_new[w_index] - 1) * dilation[1] + 1
+    dilation_shape[2] = (filter_size_new[2] - 1) * dilation[0] + 1
+    dilation_shape[3] = (filter_size_new[3] - 1) * dilation[1] + 1
 
     print("pad's shape is:", pad_shape)
+    print("dilation's shape is:", dilation_shape)
     if is_depthwise:
         return output, [pad_shape, res_shape]
     else:
-        return output, [pad_shape, dialtion_shape, res_shape]
+        return output, [pad_shape, dilation_shape, res_shape]
diff --git a/python/tests/test_op_nn.py b/python/tests/test_op_nn.py
index f1632ef9bd..57c8fa2c9d 100644
--- a/python/tests/test_op_nn.py
+++ b/python/tests/test_op_nn.py
@@ -46,7 +46,7 @@ def init_testcase(self):
         self.groups = 1
         assert np.mod(self.input_size[1], self.groups) == 0
         f_c = self.input_size[1] // self.groups
-        self.filter_size = [64, f_c, 7, 7]
+        self.filter_size = [2, f_c, 2, 2]
         assert np.mod(self.filter_size[0], self.groups) == 0
         self.data_format = "NCHW"
         self.attrs = framework.NodeAttr()
@@ -170,9 +170,9 @@ def init_testcase(self):
         self.filter_size = [16, f_c, 7, 7]
         self.data_format = "NHWC"
         self.attrs = framework.NodeAttr()
-        self.padding = [1, 1]
+        self.padding = [2, 2]
         self.stride = [2, 2]
-        self.dilation = [1, 1]
+        self.dilation = [2, 2]
         self.attrs.attr_store = {
             "stride": self.stride,
             "padding": self.padding,
diff --git a/python/tests/test_pe_reduction.py b/python/tests/test_pe_reduction.py
index d803eb6071..51c0cadd93 100644
--- a/python/tests/test_pe_reduction.py
+++ b/python/tests/test_pe_reduction.py
@@ -99,7 +99,7 @@ def reduction_tester(self, fn_name, cinn_fn, np_fn, axes, keep_dims,
         stages = create_stages([x.to_tensor()])
         if initial:
             y = cinn_fn(x.to_tensor(), stages, axes_expr, keep_dims,
-                                ir.Expr(initial))
+                        ir.Expr(initial))
             func = lang.lower(func_name, stages, [x.to_tensor(), y])
         else:
             y = cinn_fn(x.to_tensor(), stages, axes_expr, keep_dims)