Apply forwarding in pjit linearization rule to avoid intermediate copies #1277
Workflow file for this run
This file contains hidden or 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
| name: CI - Bazel Optional CUDA tests | |
| on: | |
| workflow_dispatch: | |
| inputs: | |
| halt-for-connection: | |
| description: 'Should this workflow run wait for a remote connection?' | |
| type: choice | |
| required: true | |
| default: 'no' | |
| options: | |
| - 'yes' | |
| - 'no' | |
| pull_request: | |
| branches: | |
| - main | |
| types: [ labeled, synchronize ] | |
| schedule: | |
| - cron: "0 */2 * * *" # Run once every 2 hours | |
| permissions: | |
| contents: read | |
| concurrency: | |
| group: ${{ github.workflow }}-${{ github.head_ref || github.ref }} | |
| # Don't cancel in-progress jobs for main/release branches. | |
| cancel-in-progress: ${{ !contains(github.ref, 'release/') && github.ref != 'main' }} | |
| jobs: | |
| run_tests: | |
| if: ${{ github.event.repository.fork == false && (github.event_name == 'schedule' || contains(github.event.pull_request.labels.*.name, 'CI Optional GPU Presubmit')) }} | |
| runs-on: ${{ matrix.runner }} | |
| container: 'us-central1-docker.pkg.dev/tensorflow-sigs/tensorflow/ml-build-cuda12.8-cudnn9.8:latest' | |
| strategy: | |
| matrix: | |
| # Optional gpus to run against | |
| runner: ["linux-x86-a4-224-b200-1gpu"] | |
| name: "Bazel single accelerator CUDA tests (${{ matrix.runner }})" | |
| # End Presubmit Naming Check github-cuda-presubmits | |
| steps: | |
| - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2 | |
| - name: Wait For Connection | |
| uses: google-ml-infra/actions/ci_connection@main | |
| with: | |
| halt-dispatch-input: ${{ inputs.halt-for-connection }} | |
| - name: Run Bazel CUDA Tests | |
| run: | | |
| nvidia-smi | |
| bazel test --config=ci_linux_x86_64_cuda \ | |
| --config=resultstore \ | |
| --config=rbe_cache \ | |
| --repo_env=HERMETIC_CUDA_VERSION="12.8.0" \ | |
| --repo_env=HERMETIC_CUDNN_VERSION="9.8.0" \ | |
| --repo_env=HERMETIC_PYTHON_VERSION="3.13" \ | |
| --test_env=XLA_PYTHON_CLIENT_ALLOCATOR=platform \ | |
| --run_under "$(pwd)/build/parallel_accelerator_execute.sh" \ | |
| --test_output=errors \ | |
| --test_env=JAX_ACCELERATOR_COUNT=1 \ | |
| --test_env=JAX_TESTS_PER_ACCELERATOR=32 \ | |
| --local_test_jobs=32 \ | |
| --test_env=JAX_EXCLUDE_TEST_TARGETS=PmapTest.testSizeOverflow \ | |
| --test_tag_filters=-multiaccelerator \ | |
| --test_env=TF_CPP_MIN_LOG_LEVEL=0 \ | |
| --test_env=JAX_SKIP_SLOW_TESTS=true \ | |
| --action_env=JAX_ENABLE_X64="1" \ | |
| --action_env=NCCL_DEBUG=WARN \ | |
| --color=yes \ | |
| //tests:gpu_tests //tests:backend_independent_tests \ | |
| //tests/pallas:gpu_tests //tests/pallas:backend_independent_tests |