Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

test tensorflow efa #4449

Merged
merged 24 commits into from
Dec 5, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions tensorflow/training/buildspec-2-16-ec2.yml
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,7 @@ images:
tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *OS_VERSION, "-ec2" ]
latest_release_tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *OS_VERSION,
"-ec2" ]
# build_tag_override: "beta:2.16.2-cpu-py310-ubuntu20.04-ec2"
docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /Dockerfile., *DEVICE_TYPE ]
target: ec2
enable_test_promotion: true
Expand All @@ -55,6 +56,7 @@ images:
tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *CUDA_VERSION, "-", *OS_VERSION, "-ec2" ]
latest_release_tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *CUDA_VERSION, "-",
*OS_VERSION, "-ec2" ]
# build_tag_override: "beta:2.16.2-gpu-py310-cu123-ubuntu20.04-ec2"
docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /, *CUDA_VERSION, /Dockerfile.,
*DEVICE_TYPE ]
target: ec2
Expand Down
1 change: 0 additions & 1 deletion test/dlc_tests/container_tests/bin/efa/testEFA
Original file line number Diff line number Diff line change
Expand Up @@ -64,7 +64,6 @@ check_efa_nccl_all_reduce(){
# the container. Not using full-paths of mpirun and other executables because these paths can change across PyTorch
# versions in DLC images.
mpirun -x FI_PROVIDER="efa" -n $NODES -N $GPU_COUNT --hostfile $NUM_HOSTS_FILE \
-x NCCL_TUNER_PLUGIN=/usr/local/lib/libnccl-ofi-tuner.so \
-x NCCL_DEBUG=INFO ${USE_DEVICE_RDMA_ARG} -x NCCL_PROTO=simple -x NCCL_ALGO=ring -x RDMAV_FORK_SAFE=1 \
-x PATH -x LD_LIBRARY_PATH=${CUDA_HOME}/lib:${CUDA_HOME}/lib64:$LD_LIBRARY_PATH \
-x NCCL_SOCKET_IFNAME=^lo --mca pml ^cm --mca btl tcp,self --mca btl_tcp_if_exclude lo,docker0 --bind-to none \
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -31,4 +31,12 @@ git clone https://github.com/pytorch/benchmark.git
cd benchmark
git checkout 350bb04 # pin commit for Python 3.11 support
pip install -r requirements.txt

# pin pynvml to avoid issues with 12.0.0
# https://github.com/pytorch/benchmark/issues/2552
# PR: https://github.com/pytorch/benchmark/pull/2553
# when this issue is fixed can revert back
pip uninstall pynvml -y
pip install "pynvml<12"

pip install numba
Loading