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

Install DeepEP from release wheels (#33932)

Browse Source
This commit is contained in:
Baizhou Zhang
2026-08-07 15:38:44 -07:00
committed by GitHub
parent 115cd7bde1
commit eb3cc879e0
31 changed files with 141 additions and 353 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/_pr-test-sgl-kernel-build.yml
-1
View File
@@ -40,7 +40,6 @@ env:
SGLANG_CUDA_COREDUMP: "1"
SGLANG_JIT_DEEPGEMM_FAST_WARMUP: true
SKIP_PR_TEST_HEALTH_CHECK: ${{ inputs.skip_pr_test_health_check && 'true' || 'false' }}
FORCE_REBUILD_DEEPEP: '1'
PR_TEST_BYPASS_MAINTENANCE_ON_MAIN: ${{ github.ref == 'refs/heads/main' && 'true' || 'false' }}
USE_VENV: false
.github/workflows/_pr-test-stage.yml
-3
View File
@@ -75,7 +75,6 @@ env:
SGLANG_CUDA_COREDUMP: "1"
SGLANG_JIT_DEEPGEMM_FAST_WARMUP: true
SKIP_PR_TEST_HEALTH_CHECK: ${{ (fromJson(inputs.caller_inputs).skip_pr_test_health_check || fromJson(inputs.caller_inputs).test_parallel_dispatch || fromJson(inputs.caller_inputs).run_all_tests) && 'true' || 'false' }}
FORCE_REBUILD_DEEPEP: '1'
PR_TEST_BYPASS_MAINTENANCE_ON_MAIN: ${{ github.ref == 'refs/heads/main' && 'true' || 'false' }}
USE_VENV: false
@@ -136,8 +135,6 @@ jobs:
- name: Install dependencies
timeout-minutes: ${{ fromJson(steps.rc.outputs.install_timeout) }}
env:
GRACE_BLACKWELL: ${{ steps.rc.outputs.grace_blackwell || '0' }}
run: |
CUSTOM_BUILD_SGL_KERNEL=${{ fromJson(inputs.check_changes).sgl_kernel }} bash ${{ steps.rc.outputs.install }}
.github/workflows/nightly-test-nvidia.yml
+1 -3
View File
@@ -486,10 +486,8 @@ jobs:
- uses: ./.github/actions/check-maintenance
- name: Install dependencies
env:
GRACE_BLACKWELL: "1"
run: |
bash scripts/ci/cuda/ci_install_deepep.sh
bash scripts/ci/cuda/ci_install_dependency.sh
- name: Run test
timeout-minutes: 600
.github/workflows/pr-test-extra.yml
-47
View File
@@ -64,7 +64,6 @@ env:
SGLANG_CUDA_COREDUMP: "1"
SGLANG_JIT_DEEPGEMM_FAST_WARMUP: true
SKIP_PR_TEST_HEALTH_CHECK: ${{ (inputs.skip_pr_test_health_check == true || inputs.run_all_tests == true) && 'true' || 'false' }}
FORCE_REBUILD_DEEPEP: '1'
PR_TEST_BYPASS_MAINTENANCE_ON_MAIN: ${{ github.ref == 'refs/heads/main' && 'true' || 'false' }}
USE_VENV: false
@@ -238,49 +237,6 @@ jobs:
rust_ext_artifact: ${{ needs.rust-ext-build.outputs.artifact_name }}
secrets: inherit
extra-b-test-deepep-4-gpu-h100:
needs: [check-changes, call-gate, sgl-kernel-build-wheels, rust-ext-build]
if: ${{ !failure() && !cancelled() && needs.check-changes.result == 'success' && (needs.call-gate.result == 'success' || needs.call-gate.result == 'skipped') }}
uses: ./.github/workflows/_pr-test-stage.yml
with:
self_name: extra-b-test-deepep-4-gpu-h100
runner_config: deepep-4-gpu-h100
check_changes: ${{ toJson(needs.check-changes.outputs) }}
caller_inputs: ${{ toJson(inputs) }}
partitions: ${{ needs.check-changes.outputs.partitions }}
run_timeout_minutes: '60'
rust_ext_artifact: ${{ needs.rust-ext-build.outputs.artifact_name }}
secrets: inherit
extra-b-test-deepep-4-gpu-b200:
needs: [check-changes, call-gate, sgl-kernel-build-wheels, rust-ext-build]
if: ${{ !failure() && !cancelled() && needs.check-changes.result == 'success' && (needs.call-gate.result == 'success' || needs.call-gate.result == 'skipped') }}
uses: ./.github/workflows/_pr-test-stage.yml
with:
self_name: extra-b-test-deepep-4-gpu-b200
runner_config: deepep-4-gpu-b200
check_changes: ${{ toJson(needs.check-changes.outputs) }}
caller_inputs: ${{ toJson(inputs) }}
partitions: ${{ needs.check-changes.outputs.partitions }}
run_timeout_minutes: '60'
timeout_per_file: '1800'
rust_ext_artifact: ${{ needs.rust-ext-build.outputs.artifact_name }}
secrets: inherit
extra-b-test-deepep-8-gpu-h200:
needs: [check-changes, call-gate, sgl-kernel-build-wheels, rust-ext-build]
if: ${{ !failure() && !cancelled() && needs.check-changes.result == 'success' && (needs.call-gate.result == 'success' || needs.call-gate.result == 'skipped') }}
uses: ./.github/workflows/_pr-test-stage.yml
with:
self_name: extra-b-test-deepep-8-gpu-h200
runner_config: deepep-8-gpu-h200
check_changes: ${{ toJson(needs.check-changes.outputs) }}
caller_inputs: ${{ toJson(inputs) }}
partitions: ${{ needs.check-changes.outputs.partitions }}
run_timeout_minutes: '60'
rust_ext_artifact: ${{ needs.rust-ext-build.outputs.artifact_name }}
secrets: inherit
# =============================================== aggregator ====================================================
# Mirrors pr-test.yml's `pr-test-finish` so notify-pr-states below only
# depends on one job rather than re-listing every stage. Fails if any
@@ -298,9 +254,6 @@ jobs:
extra-b-test-4-gpu-h100,
extra-b-test-4-gpu-b200,
extra-b-test-8-gpu-h200,
extra-b-test-deepep-4-gpu-h100,
extra-b-test-deepep-4-gpu-b200,
extra-b-test-deepep-8-gpu-h200,
]
if: always()
runs-on: ubuntu-latest
.github/workflows/pr-test.yml
+3 -51
View File
@@ -59,8 +59,6 @@ env:
SGLANG_CUDA_COREDUMP: "1"
SGLANG_JIT_DEEPGEMM_FAST_WARMUP: true
SKIP_PR_TEST_HEALTH_CHECK: ${{ (inputs.skip_pr_test_health_check == true || inputs.test_parallel_dispatch == true || inputs.run_all_tests == true) && 'true' || 'false' }}
# TEMP: force DeepEP rebuilds during the initial Torch 2.13 rollout so every CI runner replaces its Torch 2.11 build; remove after the runner fleet has cycled.
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
@@ -400,6 +398,8 @@ jobs:
caller_inputs: ${{ toJson(inputs) }}
partitions: ${{ needs.check-changes.outputs.partitions }}
run_timeout_minutes: '30'
warmup_deep_gemm_models: 'lmsys/sglang-ci-dsv3-test:4'
warmup_server_models: 'lmsys/sglang-ci-dsv3-test:4'
rust_ext_artifact: ${{ needs.rust-ext-build.outputs.artifact_name }}
secrets: inherit
@@ -414,6 +414,7 @@ jobs:
caller_inputs: ${{ toJson(inputs) }}
partitions: ${{ needs.check-changes.outputs.partitions }}
run_timeout_minutes: '30'
timeout_per_file: '1800'
# Per-model TP must match the test's launch in test/registered/ -- see
# FALLBACK_ARGS in scripts/ci/cuda/warmup_deep_gemm.py for extra dp/ep
# flags. Only models that actually invoke DeepGEMM kernels at runtime
@@ -438,52 +439,6 @@ jobs:
rust_ext_artifact: ${{ needs.rust-ext-build.outputs.artifact_name }}
secrets: inherit
base-c-test-deepep-4-gpu-h100:
needs: [check-changes, call-gate, wait-for-base-b, sgl-kernel-build-wheels, rust-ext-build]
if: ${{ !failure() && !cancelled() }}
uses: ./.github/workflows/_pr-test-stage.yml
with:
self_name: base-c-test-deepep-4-gpu-h100
runner_config: deepep-4-gpu-h100
check_changes: ${{ toJson(needs.check-changes.outputs) }}
caller_inputs: ${{ toJson(inputs) }}
partitions: ${{ needs.check-changes.outputs.partitions }}
run_timeout_minutes: '30'
warmup_deep_gemm_models: 'lmsys/sglang-ci-dsv3-test:4'
warmup_server_models: 'lmsys/sglang-ci-dsv3-test:4'
rust_ext_artifact: ${{ needs.rust-ext-build.outputs.artifact_name }}
secrets: inherit
base-c-test-deepep-4-gpu-b200:
needs: [check-changes, call-gate, wait-for-base-b, sgl-kernel-build-wheels, rust-ext-build]
if: ${{ !failure() && !cancelled() }}
uses: ./.github/workflows/_pr-test-stage.yml
with:
self_name: base-c-test-deepep-4-gpu-b200
runner_config: deepep-4-gpu-b200
check_changes: ${{ toJson(needs.check-changes.outputs) }}
caller_inputs: ${{ toJson(inputs) }}
partitions: ${{ needs.check-changes.outputs.partitions }}
run_timeout_minutes: '30'
timeout_per_file: '1800'
rust_ext_artifact: ${{ needs.rust-ext-build.outputs.artifact_name }}
secrets: inherit
base-c-test-deepep-8-gpu-h200:
needs: [check-changes, call-gate, wait-for-base-b, sgl-kernel-build-wheels, rust-ext-build]
if: ${{ !failure() && !cancelled() }}
uses: ./.github/workflows/_pr-test-stage.yml
with:
self_name: base-c-test-deepep-8-gpu-h200
runner_config: deepep-8-gpu-h200
check_changes: ${{ toJson(needs.check-changes.outputs) }}
caller_inputs: ${{ toJson(inputs) }}
partitions: ${{ needs.check-changes.outputs.partitions }}
run_timeout_minutes: '30'
timeout_per_file: '1800'
rust_ext_artifact: ${{ needs.rust-ext-build.outputs.artifact_name }}
secrets: inherit
base-c-test-4-gpu-b200:
needs: [check-changes, call-gate, wait-for-base-b, sgl-kernel-build-wheels, rust-ext-build]
if: ${{ !failure() && !cancelled() }}
@@ -561,9 +516,6 @@ jobs:
base-c-test-4-gpu-h100,
base-c-test-8-gpu-h20,
base-c-test-8-gpu-h200,
base-c-test-deepep-4-gpu-h100,
base-c-test-deepep-4-gpu-b200,
base-c-test-deepep-8-gpu-h200,
base-c-test-4-gpu-b200,
base-c-test-4-gpu-gb300,
base-c-test-8-gpu-b300,
.github/workflows/rerun-test.yml
+1 -10
View File
@@ -31,11 +31,6 @@ on:
required: false
type: string
default: "20"
grace_blackwell:
description: "Set GRACE_BLACKWELL for the install step (cuda only)"
required: false
type: string
default: "0"
rdma_devices:
description: "SGLANG_CI_RDMA_ALL_DEVICES csv (cuda only; empty = unset)"
required: false
@@ -62,8 +57,6 @@ env:
SGLANG_ENABLE_ASYNC_ASSERT: true
SGLANG_CUDA_COREDUMP: "1"
SGLANG_JIT_DEEPGEMM_FAST_WARMUP: true
# TEMP: force DeepEP rebuilds during the initial Torch 2.13 rollout so every CI runner replaces its Torch 2.11 build; remove after the runner fleet has cycled.
FORCE_REBUILD_DEEPEP: '1'
permissions:
actions: write
@@ -109,8 +102,6 @@ jobs:
- name: Install dependencies
timeout-minutes: ${{ fromJson(inputs.install_timeout) }}
env:
GRACE_BLACKWELL: ${{ inputs.grace_blackwell }}
run: |
if [[ "${{ inputs.runs_on }}" == "1-gpu-5090" ]]; then
source /etc/profile.d/sglang-ci.sh
@@ -120,7 +111,7 @@ jobs:
# then dies with `ModuleNotFoundError: No module named 'sglang'`.
# /rerun-test resolves install_script from runner_configs.yml; a manual
# workflow_dispatch must pass it explicitly (it can't be derived from
# runs_on, which is shared across configs, e.g. default vs deepep).
# runs_on, which can be shared across runner configs).
if [ -z "${{ inputs.install_script }}" ]; then
echo "::error::install_script is required for cuda mode (empty would silently skip installing sglang). Pass e.g. scripts/ci/cuda/ci_install_dependency.sh"
exit 1