Skip to content

[Performance] Make _to_consolidated compatible with compile #2190

[Performance] Make _to_consolidated compatible with compile

[Performance] Make _to_consolidated compatible with compile #2190

Workflow file for this run

name: Unit-tests (RL) on Linux
on:
pull_request:
push:
branches:
- nightly
- main
- release/*
workflow_dispatch:
env:
CHANNEL: "nightly"
concurrency:
# Documentation suggests ${{ github.head_ref }}, but that's only available on pull_request/pull_request_target triggers, so using ${{ github.ref }}.
# On master, we want all builds to complete even if merging happens faster to make it easier to discover at which point something broke.
group: ${{ github.workflow }}-${{ github.ref == 'refs/heads/main' && format('ci-master-{0}', github.sha) || format('ci-{0}', github.ref) }}
cancel-in-progress: true
jobs:
test-gpu:
strategy:
matrix:
python_version: ["3.10"]
cuda_arch_version: ["12.1"]
fail-fast: false
uses: pytorch/test-infra/.github/workflows/linux_job.yml@main
with:
runner: linux.g5.4xlarge.nvidia.gpu
repository: pytorch/tensordict
gpu-arch-type: cuda
gpu-arch-version: ${{ matrix.cuda_arch_version }}
timeout: 120
script: |
# Set env vars from matrix
export PYTHON_VERSION=${{ matrix.python_version }}
# Commenting these out for now because the GPU test are not working inside docker
export CUDA_ARCH_VERSION=${{ matrix.cuda_arch_version }}
export CU_VERSION="cu${CUDA_ARCH_VERSION:0:2}${CUDA_ARCH_VERSION:3:1}"
export TORCH_VERSION=nightly
# Remove the following line when the GPU tests are working inside docker, and uncomment the above lines
#export CU_VERSION="cpu"
echo "PYTHON_VERSION: $PYTHON_VERSION"
echo "CU_VERSION: $CU_VERSION"
## setup_env.sh
bash .github/unittest/rl_linux_optdeps/scripts/setup_env.sh
bash .github/unittest/rl_linux_optdeps/scripts/install.sh
bash .github/unittest/rl_linux_optdeps/scripts/run_test.sh
bash .github/unittest/rl_linux_optdeps/scripts/post_process.sh