[CI] Reorganize stage-b 1-GPU tests for 5090 compatibility (#16826)

This commit is contained in:
Alison Shao
2026-01-15 15:23:35 -08:00
committed by GitHub
parent 8b22deef5b
commit 146b5fcc84
136 changed files with 277 additions and 404 deletions
+56 -99
View File
@@ -595,110 +595,13 @@ jobs:
((needs.check-changes.outputs.main_package == 'true') || (needs.check-changes.outputs.sgl_kernel == 'true'))
)
)
runs-on: 1-gpu-runner
env:
RUNNER_LABELS: 1-gpu-runner
strategy:
fail-fast: false
max-parallel: ${{ fromJson(needs.check-changes.outputs.max_parallel) }}
matrix:
partition: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13]
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ inputs.pr_head_sha || inputs.ref || github.sha }}
- name: Download artifacts
if: needs.check-changes.outputs.sgl_kernel == 'true'
uses: actions/download-artifact@v4
with:
path: sgl-kernel/dist/
merge-multiple: true
pattern: wheel-python3.10-cuda12.9
- name: Install dependencies
timeout-minutes: 10
run: |
CUSTOM_BUILD_SGL_KERNEL=${{needs.check-changes.outputs.sgl_kernel}} bash scripts/ci/ci_install_dependency.sh
- name: Run test
timeout-minutes: 30
run: |
cd test/
CONTINUE_ON_ERROR_FLAG=""
if [[ "${{ needs.check-changes.outputs.continue_on_error }}" == "true" ]]; then
CONTINUE_ON_ERROR_FLAG="--continue-on-error"
fi
python3 run_suite.py --hw cuda --suite stage-b-test-small-1-gpu --auto-partition-id ${{ matrix.partition }} --auto-partition-size 14 $CONTINUE_ON_ERROR_FLAG
stage-b-test-large-1-gpu:
needs: [check-changes, call-gate, stage-a-test-1, sgl-kernel-build-wheels]
if: |
always() &&
(
(inputs.target_stage == 'stage-b-test-large-1-gpu') ||
(
!inputs.target_stage &&
(github.event_name == 'schedule' || (!failure() && !cancelled())) &&
((needs.check-changes.outputs.main_package == 'true') || (needs.check-changes.outputs.sgl_kernel == 'true'))
)
)
runs-on: 1-gpu-runner
env:
RUNNER_LABELS: 1-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ inputs.pr_head_sha || inputs.ref || github.sha }}
- name: Download artifacts
if: needs.check-changes.outputs.sgl_kernel == 'true'
uses: actions/download-artifact@v4
with:
path: sgl-kernel/dist/
merge-multiple: true
pattern: wheel-python3.10-cuda12.9
- name: Install dependencies
timeout-minutes: 10
run: |
CUSTOM_BUILD_SGL_KERNEL=${{needs.check-changes.outputs.sgl_kernel}} bash scripts/ci/ci_install_dependency.sh
- name: Run test
timeout-minutes: 30
run: |
cd test/
CONTINUE_ON_ERROR_FLAG=""
if [[ "${{ needs.check-changes.outputs.continue_on_error }}" == "true" ]]; then
CONTINUE_ON_ERROR_FLAG="--continue-on-error"
fi
python3 run_suite.py --hw cuda --suite stage-b-test-large-1-gpu $CONTINUE_ON_ERROR_FLAG
# 5090 dry run stage - only runs on scheduled CI or when explicitly targeted via /rerun-stage
# Used to validate 5090 compatibility before full integration
stage-b-test-small-1-gpu-5090:
needs: [check-changes, call-gate, stage-a-test-1, sgl-kernel-build-wheels]
if: |
always() &&
(
(inputs.target_stage == 'stage-b-test-small-1-gpu-5090') ||
(
!inputs.target_stage &&
github.event_name == 'schedule' &&
(!failure() && !cancelled()) &&
((needs.check-changes.outputs.main_package == 'true') || (needs.check-changes.outputs.sgl_kernel == 'true'))
)
)
runs-on: 1-gpu-5090
continue-on-error: true
env:
RUNNER_LABELS: 1-gpu-5090
IS_BLACKWELL: "1"
strategy:
fail-fast: false
max-parallel: 8
max-parallel: 4
matrix:
partition: [0, 1, 2, 3, 4, 5, 6, 7]
steps:
@@ -726,7 +629,61 @@ jobs:
run: |
source /etc/profile.d/sglang-ci.sh
cd test/
python3 run_suite.py --hw cuda --suite stage-b-test-small-1-gpu-5090 --auto-partition-id ${{ matrix.partition }} --auto-partition-size 8 --continue-on-error
CONTINUE_ON_ERROR_FLAG=""
if [[ "${{ needs.check-changes.outputs.continue_on_error }}" == "true" ]]; then
CONTINUE_ON_ERROR_FLAG="--continue-on-error"
fi
python3 run_suite.py --hw cuda --suite stage-b-test-small-1-gpu --auto-partition-id ${{ matrix.partition }} --auto-partition-size 8 $CONTINUE_ON_ERROR_FLAG
# Runs on H200 (80GB, SM90) - tests that don't pass on 5090 (FA3, FP8, high VRAM, etc.)
stage-b-test-large-1-gpu:
needs: [check-changes, call-gate, stage-a-test-1, sgl-kernel-build-wheels]
if: |
always() &&
(
(inputs.target_stage == 'stage-b-test-large-1-gpu') ||
(
!inputs.target_stage &&
(github.event_name == 'schedule' || (!failure() && !cancelled())) &&
((needs.check-changes.outputs.main_package == 'true') || (needs.check-changes.outputs.sgl_kernel == 'true'))
)
)
runs-on: 1-gpu-runner
env:
RUNNER_LABELS: 1-gpu-runner
strategy:
fail-fast: false
max-parallel: ${{ fromJson(needs.check-changes.outputs.max_parallel) }}
matrix:
partition: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ inputs.pr_head_sha || inputs.ref || github.sha }}
- name: Download artifacts
if: needs.check-changes.outputs.sgl_kernel == 'true'
uses: actions/download-artifact@v4
with:
path: sgl-kernel/dist/
merge-multiple: true
pattern: wheel-python3.10-cuda12.9
- name: Install dependencies
timeout-minutes: 10
run: |
CUSTOM_BUILD_SGL_KERNEL=${{needs.check-changes.outputs.sgl_kernel}} bash scripts/ci/ci_install_dependency.sh
- name: Run test
timeout-minutes: 30
run: |
cd test/
CONTINUE_ON_ERROR_FLAG=""
if [[ "${{ needs.check-changes.outputs.continue_on_error }}" == "true" ]]; then
CONTINUE_ON_ERROR_FLAG="--continue-on-error"
fi
python3 run_suite.py --hw cuda --suite stage-b-test-large-1-gpu --auto-partition-id ${{ matrix.partition }} --auto-partition-size 10 $CONTINUE_ON_ERROR_FLAG
stage-b-test-large-2-gpu:
needs: [check-changes, call-gate, stage-a-test-1, sgl-kernel-build-wheels]