[XPU CI] Expand stage-a and consolidate stage-b tests into stage-a (#27156)

Co-authored-by: vshekhawat-hlab <vshekhawat@habana.ai>
Co-authored-by: Ma Mingfei <mingfei.ma@intel.com>
This commit is contained in:
ashwini rathi
2026-06-04 12:00:25 +08:00
committed by GitHub
co-authored by vshekhawat-hlab Ma Mingfei
parent 10ab7c919f
commit 5c8a04ac4e
9 changed files with 68 additions and 69 deletions
+57 -62
View File
@@ -132,79 +132,74 @@ jobs:
steps:
- run: echo "stage-a passed"
# ==================== Stage B ==================== #
stage-b-test-1-gpu-xpu:
needs: [check-changes, pr-gate, wait-for-stage-a]
if: needs.check-changes.outputs.main_package == 'true'
runs-on: intel-bmg
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
fetch-depth: 0
ref: ${{ inputs.ref || github.ref }}
- name: Build Docker image
run: |
PR_REPO=${{ github.event.pull_request.head.repo.clone_url }}
PR_HEAD_REF=${{ github.head_ref }}
docker build \
${PR_REPO:+--build-arg SG_LANG_REPO=$PR_REPO} \
${PR_HEAD_REF:+--build-arg SG_LANG_BRANCH=$PR_HEAD_REF} \
--no-cache --progress=plain -f docker/xpu.Dockerfile -t xpu_sglang_main:bmg .
- name: Run container
id: start_container
run: |
container_id=$(docker run -dt \
--group-add 992 \
--group-add $(getent group video | cut -d: -f3) \
--group-add $(getent group render | cut -d: -f3) \
-v $HOME/.cache/huggingface:/root/.cache/huggingface \
--device /dev/dri \
-v /dev/dri/by-path:/dev/dri/by-path \
-e HF_TOKEN="$(cat ~/huggingface_token.txt)" \
xpu_sglang_main:bmg)
echo "Started container: $container_id"
echo "container_id=$container_id" >> "$GITHUB_OUTPUT"
- name: Install Dependency
timeout-minutes: 20
run: |
cid="${{ steps.start_container.outputs.container_id }}"
docker exec "$cid" /opt/venv/bin/python3 -m pip install --upgrade pip
docker exec "$cid" /opt/venv/bin/python3 -m pip install pytest expecttest ray huggingface_hub tabulate
docker exec "$cid" /opt/venv/bin/python3 -m pip uninstall -y flashinfer-python
docker exec "$cid" /bin/bash -c '/opt/venv/bin/hf auth login --token ${HF_TOKEN} '
- name: Run stage-b tests
timeout-minutes: 60
run: |
cid="${{ steps.start_container.outputs.container_id }}"
docker exec "$cid" bash -c "source /opt/venv/bin/activate && cd /sgl-workspace/sglang/test && python3 run_suite.py --hw xpu --suite stage-b-test-1-gpu-xpu"
- name: Cleanup container
if: always()
run: |
cid="${{ steps.start_container.outputs.container_id }}"
docker rm -f "$cid" || true
# ==================== Stage B (disabled — tests moved to Stage A) ==================== #
# stage-b-test-1-gpu-xpu:
# needs: [check-changes, pr-gate, wait-for-stage-a]
# if: needs.check-changes.outputs.main_package == 'true'
# runs-on: intel-bmg
# steps:
# - name: Checkout code
# uses: actions/checkout@v4
# with:
# fetch-depth: 0
# ref: ${{ inputs.ref || github.ref }}
#
# - name: Build Docker image
# run: |
# PR_REPO=${{ github.event.pull_request.head.repo.clone_url }}
# PR_HEAD_REF=${{ github.head_ref }}
# docker build \
# ${PR_REPO:+--build-arg SG_LANG_REPO=$PR_REPO} \
# ${PR_HEAD_REF:+--build-arg SG_LANG_BRANCH=$PR_HEAD_REF} \
# --no-cache --progress=plain -f docker/xpu.Dockerfile -t xpu_sglang_main:bmg .
#
# - name: Run container
# id: start_container
# run: |
# container_id=$(docker run -dt \
# --group-add 992 \
# --group-add $(getent group video | cut -d: -f3) \
# --group-add $(getent group render | cut -d: -f3) \
# -v $HOME/.cache/huggingface:/root/.cache/huggingface \
# --device /dev/dri \
# -v /dev/dri/by-path:/dev/dri/by-path \
# -e HF_TOKEN="$(cat ~/huggingface_token.txt)" \
# xpu_sglang_main:bmg)
# echo "Started container: $container_id"
# echo "container_id=$container_id" >> "$GITHUB_OUTPUT"
#
# - name: Install Dependency
# timeout-minutes: 20
# run: |
# cid="${{ steps.start_container.outputs.container_id }}"
# docker exec "$cid" /opt/venv/bin/python3 -m pip install --upgrade pip
# docker exec "$cid" /opt/venv/bin/python3 -m pip install pytest expecttest ray huggingface_hub tabulate
# docker exec "$cid" /opt/venv/bin/python3 -m pip uninstall -y flashinfer-python
# docker exec "$cid" /bin/bash -c '/opt/venv/bin/hf auth login --token ${HF_TOKEN} '
#
# - name: Run stage-b tests
# timeout-minutes: 60
# run: |
# cid="${{ steps.start_container.outputs.container_id }}"
# docker exec "$cid" bash -c "source /opt/venv/bin/activate && cd /sgl-workspace/sglang/test && python3 run_suite.py --hw xpu --suite stage-b-test-1-gpu-xpu"
#
# - name: Cleanup container
# if: always()
# run: |
# cid="${{ steps.start_container.outputs.container_id }}"
# docker rm -f "$cid" || true
finish:
if: always()
needs: [stage-a-test-1-gpu-xpu, stage-b-test-1-gpu-xpu, pr-gate]
needs: [stage-a-test-1-gpu-xpu, pr-gate]
runs-on: ubuntu-latest
steps:
- name: Check job status
run: |
stage_a="${{ needs.stage-a-test-1-gpu-xpu.result }}"
stage_b="${{ needs.stage-b-test-1-gpu-xpu.result }}"
if [ "$stage_a" != "success" ] && [ "$stage_a" != "skipped" ]; then
echo "stage-a failed with result: $stage_a"
exit 1
fi
if [ "$stage_b" != "success" ] && [ "$stage_b" != "skipped" ]; then
echo "stage-b failed with result: $stage_b"
exit 1
fi
echo "All jobs completed successfully"
exit 0
@@ -10,7 +10,7 @@ from sglang.srt.utils import get_device
from sglang.test.ci.ci_register import register_cuda_ci, register_xpu_ci
register_cuda_ci(est_time=11, stage="base-b", runner_config="1-gpu-large")
register_xpu_ci(est_time=900, suite="stage-b-test-1-gpu-xpu")
register_xpu_ci(est_time=900, suite="stage-a-test-1-gpu-xpu")
@unittest.skipIf(
@@ -24,11 +24,13 @@ from sglang.test.ci.ci_register import (
register_amd_ci,
register_cpu_ci,
register_cuda_ci,
register_xpu_ci,
)
register_cuda_ci(est_time=200, suite="nightly-1-gpu", nightly=True)
register_amd_ci(est_time=200, suite="nightly-amd-1-gpu", nightly=True)
register_cpu_ci(est_time=6, suite="base-b-test-cpu")
register_xpu_ci(est_time=10, suite="stage-a-test-1-gpu-xpu")
class TestLoRAEvictionPolicy(unittest.TestCase):
@@ -1,9 +1,10 @@
"""Unit tests for srt/sampling/sampling_params.py — no server, no model loading."""
from sglang.test.ci.ci_register import register_cpu_ci
from sglang.test.ci.ci_register import register_cpu_ci, register_xpu_ci
register_cpu_ci(est_time=7, suite="base-a-test-cpu")
register_cpu_ci(est_time=7, suite="base-b-test-cpu")
register_xpu_ci(est_time=10, suite="stage-a-test-1-gpu-xpu")
import unittest
from unittest.mock import MagicMock
@@ -5,9 +5,10 @@ import unittest
from sglang.srt.speculative.adaptive_spec_params import (
AdaptiveSpeculativeParams,
)
from sglang.test.ci.ci_register import register_cpu_ci
from sglang.test.ci.ci_register import register_cpu_ci, register_xpu_ci
register_cpu_ci(est_time=6, suite="base-a-test-cpu")
register_xpu_ci(est_time=10, suite="stage-a-test-1-gpu-xpu")
class TestAdaptiveSpeculativeParams(unittest.TestCase):
+1 -1
View File
@@ -19,7 +19,7 @@ from sglang.test.test_utils import (
popen_launch_server,
)
register_xpu_ci(est_time=360, suite="stage-b-test-1-gpu-xpu")
register_xpu_ci(est_time=360, suite="stage-a-test-1-gpu-xpu")
class TestDeepSeekOCR(CustomTestCase):
@@ -18,7 +18,7 @@ from sglang.test.test_utils import (
register_xpu_ci(
est_time=360,
suite="stage-b-test-1-gpu-xpu",
suite="stage-a-test-1-gpu-xpu",
disabled="Temporarily disabled until Triton-XPU upgrade",
)
@@ -16,7 +16,7 @@ from sglang.test.test_utils import (
run_bench_one_batch,
)
register_xpu_ci(est_time=600, suite="stage-b-test-1-gpu-xpu")
register_xpu_ci(est_time=600, suite="stage-a-test-1-gpu-xpu")
def intel_xpu_benchmark(
+1 -1
View File
@@ -11,7 +11,7 @@ from sglang.srt.layers.moe.topk import (
from sglang.test.ci.ci_register import register_xpu_ci
from sglang.test.test_utils import CustomTestCase
register_xpu_ci(est_time=5, suite="stage-b-test-1-gpu-xpu")
register_xpu_ci(est_time=5, suite="stage-a-test-1-gpu-xpu")
# Nemotron-3 uses biased_grouped_topk