minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[Dependency] Upgrade to Torch 2.11.0 (#21247)

Browse Source 
Co-authored-by: Kangyan Zhou <zky314343421@gmail.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: Baizhou Zhang <sobereddiezhang@gmail.com>
Co-authored-by: b8zhong <b8zhong@users.noreply.github.com>
Co-authored-by: Mick <mickjagger19@icloud.com>
This commit is contained in:
Brayden Zhong
2026-05-02 12:25:36 -07:00
committed by GitHub
co-authored by Kangyan Zhou Claude Opus 4.7 Baizhou Zhang b8zhong Mick
parent 24a6b3084d
commit 88bb5dffe4
21 changed files with 658 additions and 211 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/pr-test-amd-rocm720.yml
+20
View File
@@ -638,6 +638,7 @@ jobs:
-e SGLANG_DENOISE_AGG_TOLERANCE=0.3 \
-e SGLANG_SKIP_CONSISTENCY=1 \
-e SGLANG_TEST_NUM_INFERENCE_STEPS=5 \
-e SGLANG_DIFFUSION_ARTIFACT_DIR=/sglang-checkout/diffusion-failures \
-e AITER_JIT_DIR=/sgl-data/aiter-kernels \
-e MIOPEN_USER_DB_PATH=/sgl-data/miopen-cache \
-e HF_HUB_ENABLE_HF_TRANSFER=1 \
@@ -653,6 +654,15 @@ jobs:
echo "=== Post-test System Memory Status ==="
free -h
- name: Upload diffusion failure artifacts
if: always()
uses: actions/upload-artifact@v4
with:
name: diffusion-failures-amd-rocm720-1gpu-${{ matrix.part }}-${{ github.run_attempt }}
path: diffusion-failures/
if-no-files-found: ignore
retention-days: 7
multimodal-gen-test-2-gpu-amd-rocm720:
needs: [check-changes]
if: |
@@ -768,6 +778,7 @@ jobs:
-e SGLANG_DENOISE_AGG_TOLERANCE=0.3 \
-e SGLANG_SKIP_CONSISTENCY=1 \
-e SGLANG_TEST_NUM_INFERENCE_STEPS=5 \
-e SGLANG_DIFFUSION_ARTIFACT_DIR=/sglang-checkout/diffusion-failures \
-e AITER_JIT_DIR=/sgl-data/aiter-kernels \
-e MIOPEN_USER_DB_PATH=/sgl-data/miopen-cache \
-e HF_HUB_ENABLE_HF_TRANSFER=1 \
@@ -782,6 +793,15 @@ jobs:
echo "=== Post-test System Memory Status ==="
free -h
- name: Upload diffusion failure artifacts
if: always()
uses: actions/upload-artifact@v4
with:
name: diffusion-failures-amd-rocm720-2gpu-${{ matrix.part }}-${{ github.run_attempt }}
path: diffusion-failures/
if-no-files-found: ignore
retention-days: 7
stage-c-test-4-gpu-amd-rocm720:
needs: [check-changes, stage-b-test-1-gpu-small-amd-rocm720, stage-b-test-2-gpu-large-amd-rocm720]
.github/workflows/pr-test-amd.yml
+20
View File
@@ -693,6 +693,7 @@ jobs:
-e SGLANG_DENOISE_AGG_TOLERANCE=0.3 \
-e SGLANG_SKIP_CONSISTENCY=1 \
-e SGLANG_TEST_NUM_INFERENCE_STEPS=5 \
-e SGLANG_DIFFUSION_ARTIFACT_DIR=/sglang-checkout/diffusion-failures \
-e AITER_JIT_DIR=/sgl-data/aiter-kernels \
-e MIOPEN_USER_DB_PATH=/sgl-data/miopen-cache \
-e HF_HUB_ENABLE_HF_TRANSFER=1 \
@@ -709,6 +710,15 @@ jobs:
echo "=== Post-test System Memory Status ==="
free -h
- name: Upload diffusion failure artifacts
if: always()
uses: actions/upload-artifact@v4
with:
name: diffusion-failures-amd-1gpu-${{ matrix.part }}-${{ github.run_attempt }}
path: diffusion-failures/
if-no-files-found: ignore
retention-days: 7
multimodal-gen-test-2-gpu-amd:
name: ${{ format('multimodal-gen-test-2-gpu-amd (linux-{0}-2gpu-sglang, {1})', inputs.runner_arch || (github.event_name == 'pull_request' && 'mi300' || 'mi325'), matrix.part) }}
needs: [check-changes, call-gate]
@@ -823,6 +833,7 @@ jobs:
-e SGLANG_DENOISE_AGG_TOLERANCE=0.3 \
-e SGLANG_SKIP_CONSISTENCY=1 \
-e SGLANG_TEST_NUM_INFERENCE_STEPS=5 \
-e SGLANG_DIFFUSION_ARTIFACT_DIR=/sglang-checkout/diffusion-failures \
-e AITER_JIT_DIR=/sgl-data/aiter-kernels \
-e MIOPEN_USER_DB_PATH=/sgl-data/miopen-cache \
-e HF_HUB_ENABLE_HF_TRANSFER=1 \
@@ -838,6 +849,15 @@ jobs:
echo "=== Post-test System Memory Status ==="
free -h
- name: Upload diffusion failure artifacts
if: always()
uses: actions/upload-artifact@v4
with:
name: diffusion-failures-amd-2gpu-${{ matrix.part }}-${{ github.run_attempt }}
path: diffusion-failures/
if-no-files-found: ignore
retention-days: 7
wait-for-stage-b-amd:
needs: [check-changes, call-gate, wait-for-stage-a-amd]
.github/workflows/pr-test-jit-kernel.yml
+63 -4
View File
@@ -6,6 +6,9 @@ on:
jit_kernel:
required: true
type: string
sgl_kernel:
required: true
type: string
b200_runner:
required: true
type: string
@@ -56,10 +59,24 @@ jobs:
- uses: ./.github/actions/check-maintenance
- name: Cleanup
if: inputs.sgl_kernel == 'true'
run: |
ls -alh sgl-kernel/dist || true
rm -rf sgl-kernel/dist/* || true
- name: Download artifacts
if: inputs.sgl_kernel == 'true'
uses: actions/download-artifact@v4
with:
path: sgl-kernel/dist/
merge-multiple: true
pattern: wheel-python3.10-cuda13.0
- name: Install dependencies
timeout-minutes: 20
run: |
bash scripts/ci/cuda/ci_install_dependency.sh diffusion
CUSTOM_BUILD_SGL_KERNEL=${{ inputs.sgl_kernel }} bash scripts/ci/cuda/ci_install_dependency.sh diffusion
- name: Run test
timeout-minutes: 30
@@ -81,10 +98,24 @@ jobs:
- uses: ./.github/actions/check-maintenance
- name: Cleanup
if: inputs.sgl_kernel == 'true'
run: |
ls -alh sgl-kernel/dist || true
rm -rf sgl-kernel/dist/* || true
- name: Download artifacts
if: inputs.sgl_kernel == 'true'
uses: actions/download-artifact@v4
with:
path: sgl-kernel/dist/
merge-multiple: true
pattern: wheel-python3.10-cuda13.0
- name: Install dependencies
timeout-minutes: 20
run: |
bash scripts/ci/cuda/ci_install_dependency.sh diffusion
CUSTOM_BUILD_SGL_KERNEL=${{ inputs.sgl_kernel }} bash scripts/ci/cuda/ci_install_dependency.sh diffusion
- name: Run multi-GPU test
timeout-minutes: 45
@@ -108,10 +139,24 @@ jobs:
- uses: ./.github/actions/check-maintenance
- name: Cleanup
if: inputs.sgl_kernel == 'true'
run: |
ls -alh sgl-kernel/dist || true
rm -rf sgl-kernel/dist/* || true
- name: Download artifacts
if: inputs.sgl_kernel == 'true'
uses: actions/download-artifact@v4
with:
path: sgl-kernel/dist/
merge-multiple: true
pattern: wheel-python3.10-cuda13.0
- name: Install dependencies
timeout-minutes: 20
run: |
bash scripts/ci/cuda/ci_install_dependency.sh diffusion
CUSTOM_BUILD_SGL_KERNEL=${{ inputs.sgl_kernel }} bash scripts/ci/cuda/ci_install_dependency.sh diffusion
- name: Run benchmark tests
timeout-minutes: 45
@@ -135,10 +180,24 @@ jobs:
- uses: ./.github/actions/check-maintenance
- name: Cleanup
if: inputs.sgl_kernel == 'true'
run: |
ls -alh sgl-kernel/dist || true
rm -rf sgl-kernel/dist/* || true
- name: Download artifacts
if: inputs.sgl_kernel == 'true'
uses: actions/download-artifact@v4
with:
path: sgl-kernel/dist/
merge-multiple: true
pattern: wheel-python3.10-cuda13.0
- name: Install dependencies
timeout-minutes: 20
run: |
bash scripts/ci/cuda/ci_install_dependency.sh diffusion
CUSTOM_BUILD_SGL_KERNEL=${{ inputs.sgl_kernel }} bash scripts/ci/cuda/ci_install_dependency.sh diffusion
- name: Run B200 diffusion test
timeout-minutes: 30
.github/workflows/pr-test-multimodal-gen.yml
+24 -4
View File
@@ -129,6 +129,7 @@ jobs:
RUNAI_STREAMER_MEMORY_LIMIT: 0
CONTINUE_ON_ERROR_FLAG: ${{ inputs.continue_on_error == 'true' && '--continue-on-error' || '' }}
PARTITION_PLAN_JSON: ${{ needs.compute-diffusion-partitions.outputs.plan-1gpu }}
SGLANG_DIFFUSION_ARTIFACT_DIR: ${{ github.workspace }}/diffusion-failures
run: |
cd python
python3 sglang/multimodal_gen/test/run_suite.py \
@@ -146,6 +147,15 @@ jobs:
path: python/sglang/multimodal_gen/test/execution_report_*.json
retention-days: 1
- name: Upload diffusion failure artifacts
if: always()
uses: actions/upload-artifact@v4
with:
name: diffusion-failures-1gpu-${{ matrix.part }}-${{ github.run_attempt }}
path: diffusion-failures/
if-no-files-found: ignore
retention-days: 7
- uses: ./.github/actions/upload-cuda-coredumps
if: failure()
with:
@@ -199,6 +209,7 @@ jobs:
RUNAI_STREAMER_MEMORY_LIMIT: 0
CONTINUE_ON_ERROR_FLAG: ${{ inputs.continue_on_error == 'true' && '--continue-on-error' || '' }}
PARTITION_PLAN_JSON: ${{ needs.compute-diffusion-partitions.outputs.plan-2gpu }}
SGLANG_DIFFUSION_ARTIFACT_DIR: ${{ github.workspace }}/diffusion-failures
run: |
cd python
python3 sglang/multimodal_gen/test/run_suite.py \
@@ -216,6 +227,15 @@ jobs:
path: python/sglang/multimodal_gen/test/execution_report_*.json
retention-days: 1
- name: Upload diffusion failure artifacts
if: always()
uses: actions/upload-artifact@v4
with:
name: diffusion-failures-2gpu-${{ matrix.part }}-${{ github.run_attempt }}
path: diffusion-failures/
if-no-files-found: ignore
retention-days: 7
- uses: ./.github/actions/upload-cuda-coredumps
if: failure()
with:
@@ -312,19 +332,19 @@ jobs:
env:
RUNAI_STREAMER_MEMORY_LIMIT: 0
CONTINUE_ON_ERROR_FLAG: ${{ inputs.continue_on_error == 'true' && '--continue-on-error' || '' }}
SGLANG_DIFFUSION_ARTIFACT_DIR: ${{ github.workspace }}/diffusion-artifacts
SGLANG_DIFFUSION_ARTIFACT_DIR: ${{ github.workspace }}/diffusion-failures
run: |
cd python
python3 sglang/multimodal_gen/test/run_suite.py \
--suite 1-gpu-b200 \
$CONTINUE_ON_ERROR_FLAG
- name: Upload diffusion artifacts
- name: Upload diffusion failure artifacts
if: always()
uses: actions/upload-artifact@v4
with:
name: diffusion-artifacts-${{ github.job }}-${{ github.run_attempt }}
path: diffusion-artifacts/
name: diffusion-failures-${{ github.job }}-${{ github.run_attempt }}
path: diffusion-failures/
if-no-files-found: ignore
- uses: ./.github/actions/upload-cuda-coredumps
.github/workflows/pr-test-npu.yml
+56 -26
View File
@@ -327,15 +327,25 @@ jobs:
SGLANG_USE_MODELSCOPE: true
SGLANG_IS_IN_CI: true
HF_ENDPOINT: https://hf-mirror.com
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
run: |
export PATH="/usr/local/Ascend/8.3.RC1/compiler/bishengir/bin:${PATH}"
cd python
python3 sglang/multimodal_gen/test/run_suite_npu.py --suite 1-npu
multimodal-gen-test-2-npu-a3:
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
SGLANG_DIFFUSION_ARTIFACT_DIR: ${{ github.workspace }}/diffusion-failures
run: |
export PATH="/usr/local/Ascend/8.3.RC1/compiler/bishengir/bin:${PATH}"
cd python
python3 sglang/multimodal_gen/test/run_suite_npu.py --suite 1-npu
- name: Upload diffusion failure artifacts
if: always()
uses: actions/upload-artifact@v4
with:
name: diffusion-failures-npu-1-${{ github.run_attempt }}
path: diffusion-failures/
if-no-files-found: ignore
retention-days: 7
multimodal-gen-test-2-npu-a3:
needs: [check-changes, pr-gate]
if: needs.check-changes.outputs.multimodal_gen == 'true'
runs-on: linux-aarch64-a3-16
@@ -376,15 +386,25 @@ jobs:
SGLANG_USE_MODELSCOPE: true
SGLANG_IS_IN_CI: true
HF_ENDPOINT: https://hf-mirror.com
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
run: |
export PATH="/usr/local/Ascend/8.3.RC1/compiler/bishengir/bin:${PATH}"
cd python
python3 sglang/multimodal_gen/test/run_suite_npu.py --suite 2-npu
multimodal-gen-test-8-npu-a3:
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
SGLANG_DIFFUSION_ARTIFACT_DIR: ${{ github.workspace }}/diffusion-failures
run: |
export PATH="/usr/local/Ascend/8.3.RC1/compiler/bishengir/bin:${PATH}"
cd python
python3 sglang/multimodal_gen/test/run_suite_npu.py --suite 2-npu
- name: Upload diffusion failure artifacts
if: always()
uses: actions/upload-artifact@v4
with:
name: diffusion-failures-npu-2-${{ github.run_attempt }}
path: diffusion-failures/
if-no-files-found: ignore
retention-days: 7
multimodal-gen-test-8-npu-a3:
needs: [check-changes, pr-gate]
if: needs.check-changes.outputs.multimodal_gen == 'true'
runs-on: linux-aarch64-a3-8
@@ -425,14 +445,24 @@ jobs:
SGLANG_USE_MODELSCOPE: true
SGLANG_IS_IN_CI: true
HF_ENDPOINT: https://hf-mirror.com
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
run: |
cd python
python3 sglang/multimodal_gen/test/run_suite_npu.py --suite 8-npu
pr-test-npu-finish:
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
SGLANG_DIFFUSION_ARTIFACT_DIR: ${{ github.workspace }}/diffusion-failures
run: |
cd python
python3 sglang/multimodal_gen/test/run_suite_npu.py --suite 8-npu
- name: Upload diffusion failure artifacts
if: always()
uses: actions/upload-artifact@v4
with:
name: diffusion-failures-npu-8-${{ github.run_attempt }}
path: diffusion-failures/
if-no-files-found: ignore
retention-days: 7
pr-test-npu-finish:
needs:
[
check-changes,
.github/workflows/pr-test.yml
+13 -10
View File
@@ -68,6 +68,8 @@ env:
SGLANG_CUDA_COREDUMP: "1"
SGLANG_JIT_DEEPGEMM_FAST_WARMUP: true
SKIP_STAGE_HEALTH_CHECK: ${{ inputs.skip_stage_health_check == true && 'true' || 'false' }}
# TEMP: rebuild deepep against the new torch for torch-211-merge PR only — revert before merging to main.
FORCE_REBUILD_DEEPEP: '1'
# Schedule / main-branch dispatch / workflow_call from main use refs/heads/main; PR events use refs/pull/*/merge
PR_TEST_BYPASS_MAINTENANCE_ON_MAIN: ${{ github.ref == 'refs/heads/main' && 'true' || 'false' }}
USE_VENV: false
@@ -610,11 +612,18 @@ jobs:
# =============================================== jit-kernel ====================================================
call-jit-kernel-tests:
needs: [check-changes, call-gate]
if: needs.check-changes.outputs.jit_kernel == 'true'
needs: [check-changes, call-gate, sgl-kernel-build-wheels]
if: |
always() &&
!failure() && !cancelled() &&
github.event_name != 'schedule' &&
inputs.test_parallel_dispatch != true &&
!inputs.target_stage &&
needs.check-changes.outputs.jit_kernel == 'true'
uses: ./.github/workflows/pr-test-jit-kernel.yml
with:
jit_kernel: ${{ needs.check-changes.outputs.jit_kernel }}
sgl_kernel: ${{ needs.check-changes.outputs.sgl_kernel }}
b200_runner: ${{ needs.check-changes.outputs.b200_runner }}
pr_head_sha: ${{ inputs.pr_head_sha || '' }}
git_ref: ${{ inputs.git_ref || '' }}
@@ -971,7 +980,7 @@ jobs:
CUSTOM_BUILD_SGL_KERNEL=${{needs.check-changes.outputs.sgl_kernel}} bash scripts/ci/cuda/ci_install_dependency.sh
- name: Run test
timeout-minutes: 30
timeout-minutes: 40
env:
CONTINUE_ON_ERROR_FLAG: ${{ needs.check-changes.outputs.continue_on_error == 'true' && '--continue-on-error' || '' }}
run: |
@@ -1175,10 +1184,6 @@ jobs:
env:
SGLANG_CI_RDMA_ALL_DEVICES: "mlx5_1,mlx5_2,mlx5_3,mlx5_4"
CU_VERSION: cu129
strategy:
fail-fast: false
matrix:
part: [0, 1]
steps:
- name: Checkout code
uses: actions/checkout@v4
@@ -1208,12 +1213,10 @@ jobs:
CONTINUE_ON_ERROR_FLAG: ${{ needs.check-changes.outputs.continue_on_error == 'true' && '--continue-on-error' || '' }}
run: |
cd test
python3 run_suite.py --hw cuda --suite stage-c-test-8-gpu-h20 --auto-partition-id ${{ matrix.part }} --auto-partition-size 2 $CONTINUE_ON_ERROR_FLAG
python3 run_suite.py --hw cuda --suite stage-c-test-8-gpu-h20 $CONTINUE_ON_ERROR_FLAG
- uses: ./.github/actions/upload-cuda-coredumps
if: failure()
with:
artifact-suffix: ${{ matrix.part }}
- name: Cleanup venv
if: always()