fix(gpu): enable large integers other multi bit pbs #1565
Workflow file for this run
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
# Compile and test tfhe-cuda-backend on an H100 VM on hyperstack | |
name: Cuda - Fast tests on H100 | |
env: | |
CARGO_TERM_COLOR: always | |
ACTION_RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }} | |
RUSTFLAGS: "-C target-cpu=native" | |
RUST_BACKTRACE: "full" | |
RUST_MIN_STACK: "8388608" | |
SLACK_CHANNEL: ${{ secrets.SLACK_CHANNEL }} | |
SLACK_ICON: https://pbs.twimg.com/profile_images/1274014582265298945/OjBKP9kn_400x400.png | |
SLACK_USERNAME: ${{ secrets.BOT_USERNAME }} | |
SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }} | |
IS_PULL_REQUEST: ${{ github.event_name == 'pull_request' }} | |
CHECKOUT_TOKEN: ${{ secrets.REPO_CHECKOUT_TOKEN || secrets.GITHUB_TOKEN }} | |
# Secrets will be available only to zama-ai organization members | |
SECRETS_AVAILABLE: ${{ secrets.JOB_SECRET != '' }} | |
EXTERNAL_CONTRIBUTION_RUNNER: "gpu_ubuntu-22.04" | |
on: | |
# Allows you to run this workflow manually from the Actions tab as an alternative. | |
workflow_dispatch: | |
pull_request: | |
types: [ labeled ] | |
jobs: | |
should-run: | |
runs-on: ubuntu-latest | |
permissions: | |
pull-requests: read | |
outputs: | |
gpu_test: ${{ env.IS_PULL_REQUEST == 'false' || steps.changed-files.outputs.gpu_any_changed }} | |
steps: | |
- name: Checkout tfhe-rs | |
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 | |
with: | |
fetch-depth: 0 | |
persist-credentials: 'false' | |
token: ${{ env.CHECKOUT_TOKEN }} | |
- name: Check for file changes | |
id: changed-files | |
uses: tj-actions/changed-files@dcc7a0cba800f454d79fff4b993e8c3555bcc0a8 | |
with: | |
files_yaml: | | |
gpu: | |
- tfhe/Cargo.toml | |
- tfhe/build.rs | |
- backends/tfhe-cuda-backend/** | |
- tfhe/src/core_crypto/gpu/** | |
- tfhe/src/integer/gpu/** | |
- tfhe/src/integer/server_key/radix_parallel/tests_unsigned/** | |
- tfhe/src/integer/server_key/radix_parallel/tests_signed/** | |
- tfhe/src/integer/server_key/radix_parallel/tests_cases_unsigned.rs | |
- tfhe/src/shortint/parameters/** | |
- tfhe/src/high_level_api/** | |
- tfhe/src/c_api/** | |
- 'tfhe/docs/**/**.md' | |
- '.github/workflows/gpu_fast_h100_tests.yml' | |
- scripts/integer-tests.sh | |
- ci/slab.toml | |
setup-instance: | |
name: Setup instance (cuda-h100-tests) | |
needs: should-run | |
if: github.event_name != 'pull_request' || | |
(github.event.action != 'labeled' && needs.should-run.outputs.gpu_test == 'true') || | |
(github.event.action == 'labeled' && github.event.label.name == 'approved' && needs.should-run.outputs.gpu_test == 'true') | |
runs-on: ubuntu-latest | |
outputs: | |
runner-name: ${{ steps.start-remote-instance.outputs.label || steps.start-github-instance.outputs.runner_group }} | |
steps: | |
- name: Start remote instance | |
id: start-remote-instance | |
if: env.SECRETS_AVAILABLE == 'true' | |
uses: zama-ai/slab-github-runner@79939325c3c429837c10d6041e4fd8589d328bac | |
with: | |
mode: start | |
github-token: ${{ secrets.SLAB_ACTION_TOKEN }} | |
slab-url: ${{ secrets.SLAB_BASE_URL }} | |
job-secret: ${{ secrets.JOB_SECRET }} | |
backend: hyperstack | |
profile: single-h100 | |
# This instance will be spawned especially for pull-request from forked repository | |
- name: Start GitHub instance | |
id: start-github-instance | |
if: env.SECRETS_AVAILABLE == 'false' | |
run: | | |
echo "runner_group=${{ env.EXTERNAL_CONTRIBUTION_RUNNER }}" >> "$GITHUB_OUTPUT" | |
cuda-tests-linux: | |
name: CUDA H100 tests | |
needs: [ should-run, setup-instance ] | |
if: github.event_name != 'pull_request' || | |
(github.event_name == 'pull_request' && needs.setup-instance.result != 'skipped') | |
concurrency: | |
group: ${{ github.workflow }}_${{ github.head_ref || github.ref }} | |
cancel-in-progress: ${{ github.ref != 'refs/heads/main' }} | |
runs-on: ${{ needs.setup-instance.outputs.runner-name }} | |
strategy: | |
fail-fast: false | |
# explicit include-based build matrix, of known valid options | |
matrix: | |
include: | |
- os: ubuntu-22.04 | |
cuda: "12.2" | |
gcc: 11 | |
steps: | |
- name: Checkout tfhe-rs | |
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 | |
with: | |
persist-credentials: 'false' | |
token: ${{ env.CHECKOUT_TOKEN }} | |
- name: Setup Hyperstack dependencies | |
uses: ./.github/actions/gpu_setup | |
with: | |
cuda-version: ${{ matrix.cuda }} | |
gcc-version: ${{ matrix.gcc }} | |
github-instance: ${{ env.SECRETS_AVAILABLE == 'false' }} | |
- name: Install latest stable | |
uses: dtolnay/rust-toolchain@a54c7afa936fefeb4456b2dd8068152669aa8203 | |
with: | |
toolchain: stable | |
- name: Run core crypto and internal CUDA backend tests | |
run: | | |
BIG_TESTS_INSTANCE=TRUE make test_core_crypto_gpu | |
BIG_TESTS_INSTANCE=TRUE make test_integer_compression_gpu | |
BIG_TESTS_INSTANCE=TRUE make test_cuda_backend | |
- name: Run user docs tests | |
run: | | |
BIG_TESTS_INSTANCE=TRUE make test_user_doc_gpu | |
- name: Test C API | |
run: | | |
BIG_TESTS_INSTANCE=TRUE make test_c_api_gpu | |
- name: Run High Level API Tests | |
run: | | |
BIG_TESTS_INSTANCE=TRUE make test_high_level_api_gpu | |
slack-notify: | |
name: Slack Notification | |
needs: [ setup-instance, cuda-tests-linux ] | |
runs-on: ubuntu-latest | |
if: ${{ always() && needs.cuda-tests-linux.result != 'skipped' && failure() }} | |
continue-on-error: true | |
steps: | |
- name: Send message | |
if: env.SECRETS_AVAILABLE == 'true' | |
uses: rtCamp/action-slack-notify@c33737706dea87cd7784c687dadc9adf1be59990 | |
env: | |
SLACK_COLOR: ${{ needs.cuda-tests-linux.result }} | |
SLACK_MESSAGE: "Fast H100 tests finished with status: ${{ needs.cuda-tests-linux.result }} on '${{ env.BRANCH }}'. (${{ env.ACTION_RUN_URL }})" | |
teardown-instance: | |
name: Teardown instance (cuda-h100-tests) | |
if: ${{ always() && needs.setup-instance.result == 'success' }} | |
needs: [ setup-instance, cuda-tests-linux ] | |
runs-on: ubuntu-latest | |
steps: | |
- name: Stop remote instance | |
id: stop-instance | |
if: env.SECRETS_AVAILABLE == 'true' | |
uses: zama-ai/slab-github-runner@79939325c3c429837c10d6041e4fd8589d328bac | |
with: | |
mode: stop | |
github-token: ${{ secrets.SLAB_ACTION_TOKEN }} | |
slab-url: ${{ secrets.SLAB_BASE_URL }} | |
job-secret: ${{ secrets.JOB_SECRET }} | |
label: ${{ needs.setup-instance.outputs.runner-name }} | |
- name: Slack Notification | |
if: ${{ failure() }} | |
continue-on-error: true | |
uses: rtCamp/action-slack-notify@c33737706dea87cd7784c687dadc9adf1be59990 | |
env: | |
SLACK_COLOR: ${{ job.status }} | |
SLACK_MESSAGE: "Instance teardown (cuda-h100-tests) finished with status: ${{ job.status }} on '${{ env.BRANCH }}'. (${{ env.ACTION_RUN_URL }})" |