forked from openucx/ucc
-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge branch 'master' into nsarka/sliding-window-allreduce-stubs
- Loading branch information
Showing
131 changed files
with
4,706 additions
and
1,201 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,20 @@ | ||
| ARG CUDA_VER='12.1.1' | ||
| FROM harbor.mellanox.com/torch-ucc/ucc/1.0.0/x86_64/centos8/cuda${CUDA_VER}:base | ||
| RUN rm -rf ${SRC_DIR}/ucc | ||
| COPY . ${SRC_DIR}/ucc | ||
|
|
||
| RUN apt update && apt install -y sudo && \ | ||
| echo "swx-jenkins ALL=(ALL) NOPASSWD: ALL" >> /etc/sudoers | ||
| RUN pip install 'protobuf<=3.19.0' | ||
| #============================================================================== | ||
| # Build UCC | ||
| RUN ${SRC_DIR}/ucc/.ci/scripts/build_ucc.sh | ||
| #============================================================================== | ||
| # Install torch_ucc (UCC version) python module and build a wheel package | ||
| RUN chown -R 6213:11429 /opt/nvidia | ||
| #============================================================================== | ||
| RUN groupadd -g 11429 swx-jenkins | ||
| RUN adduser --no-create-home --uid 6213 --gid 11429 --home /labhome/swx-jenkins swx-jenkins | ||
| #============================================================================== | ||
| USER swx-jenkins | ||
|
|
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,69 @@ | ||
| ARG CUDA_VER='12.1.1' | ||
| FROM nvcr.io/nvidia/pytorch:23.11-py3 | ||
| #============================================================================== | ||
| ARG NVIDIA_ROOT_DIR=/opt/nvidia | ||
| ENV DEBIAN_FRONTEND=noninteractive | ||
| ENV TZ=Etc/UTC | ||
| ENV SRC_DIR=${NVIDIA_ROOT_DIR}/src | ||
| ENV PKG_DIR=${NVIDIA_ROOT_DIR}/pkg | ||
| ENV BIN_DIR=${NVIDIA_ROOT_DIR}/bin | ||
| ENV WORKLOADS_DIR=${NVIDIA_ROOT_DIR}/workloads | ||
| ENV CUDA_HOME=/usr/local/cuda | ||
| ENV UCX_GITHUB_URL=https://github.com/openucx/ucx.git | ||
| ENV UCX_BRANCH=master | ||
| ENV UCX_BUILD_TYPE=release-mt | ||
| ENV UCX_INSTALL_DIR=${BIN_DIR}/ucx/build-${UCX_BUILD_TYPE} | ||
| ENV UCC_INSTALL_DIR=${BIN_DIR}/ucc/build | ||
| ENV OFED_PKG='lsof kmod udev swig libelf1 libfuse2 pciutils tk gfortran libpci3 libusb-1.0-0 libltdl-dev libmnl0 bison tcl flex chrpath debhelper ethtool graphviz' | ||
| ENV PACKAGES='numactl openssh-server protobuf-compiler rdma-core vim libevent-dev build-essential git make autoconf libtool' | ||
| ENV OS_VERSION=ubuntu22.04 | ||
| ENV PLATFORM=x86_64 | ||
| ENV MOFED_VERSION=23.10-0.5.5.0 | ||
| ENV MOFED_URL="https://content.mellanox.com/ofed/MLNX_OFED-${MOFED_VERSION}/MLNX_OFED_LINUX-${MOFED_VERSION}-${OS_VERSION}-${PLATFORM}.tgz" | ||
| ENV OMPI_PATH="/opt/hpcx/ompi" | ||
| #============================================================================== | ||
| RUN apt update && apt install -y ${OFED_PKG} && \ | ||
| mkdir -p /tmp/ofed && wget --quiet -O /tmp/ofed/ofed.tgz ${MOFED_URL} && \ | ||
| tar -xvf /tmp/ofed/ofed.tgz --strip-components=2 -C /tmp/ofed && \ | ||
| /tmp/ofed/mlnxofedinstall --user-space-only --without-fw-update -q --distro ${OS_VERSION} --basic && \ | ||
| rm -rf /tmp/ofed | ||
|
|
||
| RUN apt install -y ${PACKAGES} | ||
|
|
||
| # Remove old UCX | ||
| RUN rm -rf /opt/hpcx/uc? | ||
| ENV PATH=${OMPI_PATH}/bin:$PATH | ||
| RUN echo "export PATH=\"\$OMPI_PATH:\$PATH\"" >> /etc/bashrc && \ | ||
| export LD_LIBRARY_PATH=\"\$OMPI_PATH/lib64:\${LD_LIBRARY_PATH}\" >> /etc/bashrc | ||
| #============================================================================== | ||
| # Configure SSH | ||
| RUN mkdir -p /var/run/sshd && \ | ||
| cat /etc/ssh/ssh_config | grep -v StrictHostKeyChecking > /etc/ssh/ssh_config.new && \ | ||
| echo " StrictHostKeyChecking no" >> /etc/ssh/ssh_config.new && \ | ||
| mv /etc/ssh/ssh_config.new /etc/ssh/ssh_config && \ | ||
| ssh-keygen -A && \ | ||
| rm -f /run/nologin | ||
| #============================================================================== | ||
|
|
||
| #============================================================================== | ||
| RUN mkdir -p ${SRC_DIR} ${PKG_DIR} ${BIN_DIR} ${WORKLOADS_DIR} && \ | ||
| cd ${SRC_DIR} && \ | ||
| mkdir -p ${SRC_DIR}/ucx && \ | ||
| git clone --recursive ${UCX_GITHUB_URL} ${SRC_DIR}/ucx && \ | ||
| cd ${SRC_DIR}/ucx && \ | ||
| git checkout ${UCX_BRANCH} | ||
|
|
||
| COPY . ${SRC_DIR}/ucc | ||
| #============================================================================== | ||
| # Build UCX | ||
| RUN ${SRC_DIR}/ucc/.ci/scripts/build_ucx.sh | ||
| ENV PATH=${UCX_INSTALL_DIR}/bin:${PATH} | ||
| #============================================================================== | ||
| # Install workloads | ||
| WORKDIR ${WORKLOADS_DIR} | ||
| RUN git clone https://github.com/facebookresearch/dlrm.git && \ | ||
| cd ${WORKLOADS_DIR}/dlrm && \ | ||
| pip3 install -r ${WORKLOADS_DIR}/dlrm/requirements.txt && \ | ||
| pip3 install tensorboard | ||
| RUN git clone https://github.com/facebookresearch/param.git && \ | ||
| pip3 install -r ${WORKLOADS_DIR}/param/requirements.txt |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,25 @@ | ||
| name: HPC_SDK | ||
|
|
||
| on: [push, pull_request] | ||
|
|
||
| env: | ||
| HPCXDIR: /opt/nvidia/hpc_sdk/Linux_x86_64/2023/comm_libs/12.2/hpcx/latest/ | ||
| NCCLDIR: /opt/nvidia/hpc_sdk/Linux_x86_64/2023/comm_libs/12.2/nccl/ | ||
| CUDADIR: /opt/nvidia/hpc_sdk/Linux_x86_64/2023/cuda/12.2/ | ||
|
|
||
| jobs: | ||
| build: | ||
| runs-on: ubuntu-20.04 | ||
| container: | ||
| image: nvcr.io/nvidia/nvhpc:23.9-devel-cuda12.2-ubuntu22.04 | ||
| steps: | ||
| - name: Install dependencies | ||
| run: | | ||
| apt-get update | ||
| apt-get install -y --no-install-recommends libiberty-dev | ||
| - uses: actions/checkout@v1 | ||
| - name: Build UCC | ||
| run: | | ||
| ./autogen.sh | ||
| CC=nvc CXX=nvc++ ./configure --with-tls=ucp,mlx5,cuda,self,nccl,sharp --with-mpi=${HPCXDIR}/ompi --with-sharp=${HPCXDIR}/sharp --with-ucx=${HPCXDIR}/ucx --with-cuda=${CUDADIR} --with-nccl=${NCCLDIR} --with-nvcc-gencode="-gencode=arch=compute_80,code=sm_80" | ||
| make -j`nproc` install |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.