rename define __CUDA to __NCLGLIB_CUDA

simonpintarelli · Jan 19, 2021 · fcfd605 · fcfd605
1 parent 313a43d
commit fcfd605
Show file tree

Hide file tree

Showing 9 changed files with 19 additions and 19 deletions.
diff --git a/cmake/nlcglib_macros.cmake b/cmake/nlcglib_macros.cmake
@@ -23,5 +23,5 @@ MACRO(NLCGLIB_SETUP_TARGET _target)
     target_link_libraries(${_target} PRIVATE my_lapack)
   endif()
   target_compile_definitions(${_target} PUBLIC $<$<BOOL:${USE_OPENMP}>:__USE_OPENMP>)
-  target_compile_definitions(${_target} PUBLIC $<$<BOOL:${USE_CUDA}>:__CUDA>)
+  target_compile_definitions(${_target} PUBLIC $<$<BOOL:${USE_CUDA}>:__NLCGLIB__CUDA>)
 ENDMACRO()
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
@@ -6,7 +6,7 @@ target_link_libraries(nlcglib_internal INTERFACE my_lapack)
 target_include_directories(nlcglib_internal PUBLIC ${CMAKE_SOURCE_DIR}/src)
 target_include_directories(nlcglib_internal PUBLIC ${CMAKE_SOURCE_DIR}/include)
 target_include_directories(nlcglib_internal PUBLIC ${CUDA_INCLUDE_DIRS})
-target_compile_definitions(nlcglib_internal PUBLIC $<$<BOOL:${USE_CUDA}>:__CUDA>)
+target_compile_definitions(nlcglib_internal PUBLIC $<$<BOOL:${USE_CUDA}>:__NLCGLIB__CUDA>)
 set(nlcglib_internal_location $<TARGET_FILE:nlcglib_internal>)
 
 
@@ -18,7 +18,7 @@ target_include_directories(nlcglib
 nlcglib_setup_target(nlcglib)
 set_target_properties(nlcglib PROPERTIES
   PUBLIC_HEADER ${CMAKE_SOURCE_DIR}/include/nlcglib.hpp)
-target_compile_definitions(nlcglib PUBLIC $<$<BOOL:${USE_CUDA}>:__CUDA>)
+target_compile_definitions(nlcglib PUBLIC $<$<BOOL:${USE_CUDA}>:__NLCGLIB__CUDA>)
 
 include(GNUInstallDirs)
 

diff --git a/src/exec_space.hpp b/src/exec_space.hpp
@@ -9,7 +9,7 @@ struct exec
 {
 };
 
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
 template <>
 struct exec<Kokkos::CudaSpace>
 {

diff --git a/src/gpu/acc.cpp b/src/gpu/acc.cpp
@@ -1,4 +1,4 @@
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
 #include <Kokkos_Complex.hpp>
 #include <cassert>
 #include <complex>

diff --git a/src/la/cusolver.cpp b/src/la/cusolver.cpp
@@ -1,4 +1,4 @@
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
 #include "cusolver.hpp"
 #include <iostream>
 

diff --git a/src/la/lapack.hpp b/src/la/lapack.hpp
@@ -4,7 +4,7 @@
 #include <utility>
 #include <la/map.hpp>
 #include "lapack_cpu.hpp"
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
 #include "lapack_cuda.hpp"
 #endif
 #include "mvector.hpp"
@@ -42,7 +42,7 @@ namespace nlcglib {
 //   return d;
 // }
 
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
 /// diag (on CUDA-GPU)
 template <class T, class LAYOUT, class... KOKKOS>
 std::enable_if_t<
@@ -367,7 +367,7 @@ struct innerh_tr
   //   return sum;
   // }
 
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
   template <class M1, class M2>
   std::enable_if_t<
     Kokkos::SpaceAccessibility<Kokkos::Cuda, typename M1::storage_t::memory_space>::accessible,
@@ -463,7 +463,7 @@ double l2norm(const mvector<X>& x) {
   return Kokkos::real(z);
 }
 
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
 template<class memspace>
 std::enable_if_t<Kokkos::SpaceAccessibility<Kokkos::Cuda, memspace>::accessible>
 loewdin_aux(Kokkos::View<double*, memspace>& w)

diff --git a/src/la/mvector.hpp b/src/la/mvector.hpp
@@ -397,7 +397,7 @@ make_mmatrix(std::shared_ptr<MatrixBaseZ> matrix_base, std::enable_if_t<std::is_
     auto buffer = matrix_base->get(i);
     auto kindex = matrix_base->kpoint_index(i);
     Communicator comm(buffer.mpi_comm);
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
     if (Kokkos::SpaceAccessibility<Kokkos::Cuda, memspace>::accessible) {
       // make sure is memory type device
       if (buffer.memtype != memory_type::device)
@@ -433,7 +433,7 @@ make_mmatrix(std::shared_ptr<MatrixBaseZ> matrix_base, std::enable_if_t<!std::is
     auto buffer = matrix_base->get(i);
     auto kindex = matrix_base->kpoint_index(i);
     Communicator comm(buffer.mpi_comm);
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
     if (Kokkos::SpaceAccessibility<Kokkos::Cuda, memspace>::accessible) {
       // make sure is memory type device
       if (buffer.memtype != memory_type::device)
@@ -467,7 +467,7 @@ auto make_mmvector(std::shared_ptr<VectorBaseZ> vector_base)
     // vector_t vector();
     auto buffer = vector_base->get(i);
     if (buffer.memtype == memory_type::device) {
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
       Kokkos::View<double*, Kokkos::CudaSpace, Kokkos::MemoryUnmanaged> src(buffer.data, buffer.size[0]);
       vector_t dst("vector", buffer.size[0]);
       Kokkos::deep_copy(dst, src);

diff --git a/src/nlcglib.cpp b/src/nlcglib.cpp
@@ -321,7 +321,7 @@ nlcg_check_gradient_host(EnergyBase& energy)
 void
 nlcg_check_gradient_cuda(EnergyBase& energy)
 {
-#if defined (__CLANG) && defined (__CUDA)
+#if defined (__CLANG) && defined (__NLCGLIB__CUDA)
   Kokkos::initialize();
   nlcg_check_gradient<Kokkos::CudaSpace>(energy);
   Kokkos::finalize();
@@ -342,7 +342,7 @@ nlcg_mvp2_cpu(EnergyBase& energy_base, smearing_type smearing, double temp, doub
 nlcg_info
 nlcg_mvp2_device(EnergyBase& energy_base, smearing_type smearing, double temp, double tol, double kappa, double tau, int maxiter, int restart)
 {
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
   Kokkos::initialize();
   auto info = nlcg<Kokkos::CudaSpace>(energy_base, smearing, temp, maxiter, tol, kappa, tau, restart);
   Kokkos::finalize();
@@ -365,7 +365,7 @@ nlcg_mvp2_device_cpu(EnergyBase& energy_base,
                       int maxiter,
                       int restart)
 {
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
   Kokkos::initialize();
   auto info = nlcg<Kokkos::CudaSpace, Kokkos::HostSpace>(energy_base, smearing, temp, maxiter, tol, kappa, tau, restart);
   Kokkos::finalize();
@@ -388,7 +388,7 @@ nlcg_mvp2_cpu_device(EnergyBase& energy_base,
                       int maxiter,
                       int restart)
 {
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
   Kokkos::initialize();
   auto info = nlcg<Kokkos::HostSpace, Kokkos::CudaSpace>(
       energy_base, smearing, temp, maxiter, tol, kappa, tau, restart);

diff --git a/test/test.cpp b/test/test.cpp
@@ -52,7 +52,7 @@ run_unmanaged()
   eigh(H, eigvals, S);
 }
 
-#ifdef __CUDA
+#ifdef __NLCGLIB__CUDA
 void
 run_unmanaged_cuda()
 {
@@ -117,7 +117,7 @@ int main(int argc, char *argv[])
 
   run();
 
-  #ifdef __CUDA
+  #ifdef __NLCGLIB__CUDA
   run_unmanaged_cuda();
   #endif