ci(torch): Build against CUDA 12.6; remove 12.3, 12.1, and lower builds · coreweave/ml-containers@ebbd1db

Triggered via push October 11, 2024 21:05

Eta0

pushed ebbd1db

es/torch-updates

Status Cancelled

Total duration 2h 23m 17s

Artifacts –

torch-nccl.yml

on: push

Get torch:nccl Config / Read Configuration File

10s

Matrix: Build torch:nccl

Annotations

10 errors and 3 warnings

Build torch:nccl (12.6.1, cudnn, ubuntu20.04, 2.23.4-1, 2ff05b2) / Build torch / Build Images

buildx failed with: ERROR: failed to solve: process "/bin/sh -c export MAX_JOBS=\"$(./scale.sh \"$(./effective_cpu_count.sh)\" 3 32)\" && cd audio && mkdir build && ln -s /usr/bin/cc build/cc && ln -s /usr/bin/c++ build/c++ && { if [ -d /opt/nccl-tests ]; then export USE_DISTRIBUTED=1 USE_NCCL=1 USE_SYSTEM_NCCL=1 UCC_HOME=${HPCX_UCC_DIR} UCX_HOME=${HPCX_UCX_DIR} USE_NCCL_WITH_UCC=1 USE_UCC=1 USE_SYSTEM_UCC=1; fi; } && USE_CUDNN=1 USE_OPENCV=1 BUILD_TORCH=ON BUILD_TEST=0 CUDA_HOST_COMPILER=cc USE_CUDA=1 FORCE_CUDA=1 USE_NNPACK=1 CC=cc CXX=c++ BUILD_VERSION=\"$(../version-string.sh \"$TORCH_AUDIO_VERSION\")\" TORCH_NVCC_FLAGS=\"-Xfatbin -compress-all\" python3 setup.py bdist_wheel --dist-dir ../dist" did not complete successfully: exit code: 1

Build torch:nccl (12.6.1, cudnn, ubuntu22.04, 2.23.4-1, 2ff05b2) / Build torch / Build Images

buildx failed with: ERROR: failed to solve: process "/bin/sh -c export MAX_JOBS=\"$(./scale.sh \"$(./effective_cpu_count.sh)\" 3 32)\" && cd audio && mkdir build && ln -s /usr/bin/cc build/cc && ln -s /usr/bin/c++ build/c++ && { if [ -d /opt/nccl-tests ]; then export USE_DISTRIBUTED=1 USE_NCCL=1 USE_SYSTEM_NCCL=1 UCC_HOME=${HPCX_UCC_DIR} UCX_HOME=${HPCX_UCX_DIR} USE_NCCL_WITH_UCC=1 USE_UCC=1 USE_SYSTEM_UCC=1; fi; } && USE_CUDNN=1 USE_OPENCV=1 BUILD_TORCH=ON BUILD_TEST=0 CUDA_HOST_COMPILER=cc USE_CUDA=1 FORCE_CUDA=1 USE_NNPACK=1 CC=cc CXX=c++ BUILD_VERSION=\"$(../version-string.sh \"$TORCH_AUDIO_VERSION\")\" TORCH_NVCC_FLAGS=\"-Xfatbin -compress-all\" python3 setup.py bdist_wheel --dist-dir ../dist" did not complete successfully: exit code: 1

Build torch:nccl (12.4.1, cudnn, ubuntu22.04, 2.23.4-1, 2ff05b2) / Build torch / Build Images

FailFast: cancelling since parallel instance has failed

Build torch:nccl (12.4.1, cudnn, ubuntu22.04, 2.23.4-1, 2ff05b2) / Build torch / Build Images

The operation was canceled.

Build torch:nccl (12.2.2, cudnn8, ubuntu22.04, 2.23.4-1, 2ff05b2) / Build torch / Build Images

FailFast: cancelling since parallel instance has failed

Build torch:nccl (12.2.2, cudnn8, ubuntu22.04, 2.23.4-1, 2ff05b2) / Build torch / Build Images

The operation was canceled.

Build torch:nccl (12.4.1, cudnn, ubuntu20.04, 2.23.4-1, 2ff05b2) / Build torch / Build Images

FailFast: cancelling since parallel instance has failed

Build torch:nccl (12.4.1, cudnn, ubuntu20.04, 2.23.4-1, 2ff05b2) / Build torch / Build Images

The operation was canceled.