minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 14 Packages Projects Releases Wiki Activity

[NPU] Nightly CI refactor and enhancement (#27433)

Browse Source
This commit is contained in:
hhhh1252023
2026-06-26 08:52:06 +08:00
committed by GitHub
parent f83cbc2516
commit 781537b61d
58 changed files with 10919 additions and 584 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/nightly-test-npu.yml
+316 -343
View File
@@ -1,5 +1,4 @@
name: Nightly Test (NPU)
on:
schedule:
- cron: '0 18 * * *' # Execute at 2:00 a.m. Beijing Time every day
@@ -9,6 +8,32 @@ on:
paths:
- ".github/workflows/nightly-test-npu.yml"
workflow_dispatch:
inputs:
ref:
description: 'Git ref (branch, tag, or SHA) to test. If not provided, uses the default branch.'
required: false
type: string
default: ''
job_filter:
description: 'Select which job to run (leave empty or "all" to run all jobs)'
required: false
type: string
default: 'all'
image_a2:
description: 'The a2 running docker image of the test task.'
required: false
type: string
default: 'swr.cn-southwest-2.myhuaweicloud.com/base_image/dockerhub/lmsysorg/sglang:main-cann9.0.0-910b'
image_a3:
description: 'The a3 running docker image of the test task.'
required: false
type: string
default: 'swr.cn-southwest-2.myhuaweicloud.com/base_image/dockerhub/lmsysorg/sglang:main-cann9.0.0-a3'
skip_install_flag:
description: 'Indicates whether to skip the installation of sglang, defaulting to false.'
required: false
type: string
default: 'true'
workflow_call:
inputs:
ref:
@@ -21,28 +46,31 @@ on:
required: false
type: string
default: 'all'
image_a2:
description: 'The a2 running docker image of the test task.'
required: false
type: string
default: ''
image_a3:
description: 'The a3 running docker image of the test task.'
required: false
type: string
default: 'swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/cann:9.0.0-a3-ubuntu22.04-py3.11'
default: ''
skip_install_flag:
description: 'Indicates whether to skip the installation of sglang, defaulting to false.'
required: false
type: string
default: 'false'
default: ''
concurrency:
group: nightly-test-npu-${{ inputs.ref || github.ref }}
cancel-in-progress: ${{ github.event_name != 'workflow_call' }}
jobs:
set-image-config:
runs-on: ubuntu-latest
outputs:
ref: ${{ steps.set-vars.outputs.ref }}
job_filter: ${{ steps.set-vars.outputs.job_filter }}
image_a2: ${{ steps.set-vars.outputs.image_a2 }}
image_a3: ${{ steps.set-vars.outputs.image_a3 }}
skip_install_flag: ${{ steps.set-vars.outputs.skip_install_flag }}
steps:
@@ -55,379 +83,324 @@ jobs:
else
echo "ref=${{ inputs.ref }}" >> $GITHUB_OUTPUT
fi
if [ -z "${{ inputs.job_filter }}" ]; then
echo "job_filter=all" >> $GITHUB_OUTPUT
else
echo "job_filter=${{ inputs.job_filter }}" >> $GITHUB_OUTPUT
fi
if [ -z "${{ inputs.image_a2 }}" ]; then
echo "image_a2=swr.cn-southwest-2.myhuaweicloud.com/base_image/dockerhub/lmsysorg/sglang:main-cann9.0.0-910b" >> $GITHUB_OUTPUT
else
echo "image_a2=${{ inputs.image_a2 }}" >> $GITHUB_OUTPUT
fi
if [ -z "${{ inputs.image_a3 }}" ]; then
echo "image_a3=swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/cann:9.0.0-a3-ubuntu22.04-py3.11" >> $GITHUB_OUTPUT
echo "image_a3=swr.cn-southwest-2.myhuaweicloud.com/base_image/dockerhub/lmsysorg/sglang:main-cann9.0.0-a3" >> $GITHUB_OUTPUT
else
echo "image_a3=${{ inputs.image_a3 }}" >> $GITHUB_OUTPUT
fi
if [ -z "${{ inputs.skip_install_flag }}" ]; then
echo "skip_install_flag=false" >> $GITHUB_OUTPUT
else
echo "skip_install_flag=${{ inputs.skip_install_flag }}" >> $GITHUB_OUTPUT
fi
nightly-1-npu-a3:
needs: [set-image-config]
if: ${{ (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') }}
runs-on: linux-aarch64-a3-2
nightly-poc-single-node-a2-tests:
name: single-node-poc-a2
if: ${{ !cancelled() }}
needs: [ set-image-config ]
strategy:
fail-fast: false
max-parallel: 6
matrix:
part: [0, 1]
container:
image: ${{ needs.set-image-config.outputs.image_a3 }}
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ needs.set-image-config.outputs.ref || github.ref }}
test_config:
- name: qwen3_32b_w8a8_2p_in3k5_out1k5_50ms_a2
runner: linux-aarch64-a2-4
test_case: test/registered/ascend/performance/qwen3_32b/test_npu_qwen3_32b_w8a8_2p_in3k5_out1k5_50ms_a2.py
test_type: 'perf'
uses: ./.github/workflows/nightly-test-npu-e2e-single-node.yml
with:
runner: ${{ matrix.test_config.runner }}
test_type: ${{ matrix.test_config.test_type }}
test_config_name: ${{ matrix.test_config.name }}
test_case: ${{ matrix.test_config.test_case }}
image: ${{ needs.set-image-config.outputs.image_a2 }}
install_sglang_from_source: false
transformers_version: ''
- name: Install dependencies
env:
TORCH_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/whl/cpu"
PYPI_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
UV_INDEX_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
GITHUB_PROXY_URL: "https://gh-proxy.test.osinfra.cn/"
run: |
# speed up by using infra cache services
CACHING_URL="cache-service.nginx-pypi-cache.svc.cluster.local"
sed -Ei "s@(ports|archive).ubuntu.com@${CACHING_URL}:8081@g" /etc/apt/sources.list
pip config set global.index-url http://${CACHING_URL}/pypi/simple
pip config set global.trusted-host "${CACHING_URL}"
if [ ${{ needs.set-image-config.outputs.skip_install_flag }} != "true" ];then
bash scripts/ci/npu/npu_ci_install_dependency.sh a3
fi
# copy required file from our daily cache
cp ~/.cache/modelscope/hub/datasets/otavia/ShareGPT_Vicuna_unfiltered/ShareGPT_V3_unfiltered_cleaned_split.json /tmp
# copy gsm8k dataset
cp ~/.cache/modelscope/hub/datasets/tmp/test.jsonl /tmp
- name: Print Log Information
run: |
bash scripts/ci/npu/npu_log_print.sh
- name: Run test
timeout-minutes: 240
env:
SGLANG_USE_MODELSCOPE: true
SGLANG_IS_IN_CI: true
HF_ENDPOINT: https://hf-mirror.com
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
run: |
pip install sglang_router
hf download lmms-lab/MMMU --repo-type dataset
pip install sentence_transformers torchaudio==2.10.0
pip install protobuf==6.31.1 zss pre-commit wandb>=0.16.0 tenacity==8.3.0 loguru openpyxl latex2sympy2 zstandard transformers-stream-generator tqdm-multiprocess pycocoevalcap
pip install yt-dlp sentencepiece==0.1.99 nltk av ftfy sqlitedict==2.1.0 sacrebleu>=1.5.0 pytablewriter black==24.1.0 isort==5.13.2 peft>=0.2.0 accelerate>=0.29.1
pip install jsonlines httpx==0.25.0 evaluate>=0.4.0 datasets==2.16.1 numexpr xgrammar==0.2.1 numpy==1.26.4 dotenv
git clone --branch v0.3.3 --depth 1 https://github.com/EvolvingLMMs-Lab/lmms-eval.git
cd ./lmms-eval
nohup pip install . > lmmslog.txt 2>&1 &
sleep 120
export PYTHONPATH=$PYTHONPATH:$(pwd)
cd ../
cd test
python3 run_suite.py --hw npu --suite nightly-1-npu-a3 --nightly --continue-on-error --timeout-per-file 3600 --auto-partition-id ${{ matrix.part }} --auto-partition-size 2
nightly-2-npu-a3:
nightly-poc-single-node-tests:
name: single-node-poc
if: ${{ !cancelled() }}
needs: [set-image-config]
if: ${{ (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') }}
runs-on: linux-aarch64-a3-2
strategy:
fail-fast: false
max-parallel: 6
matrix:
part: [0]
container:
test_config:
# qwen3_6_35b_a3b performance tests
- name: qwen3_6_35b_a3b_1p_in3k5_out1k5_50ms
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3_6_35b_a3b/test_npu_qwen3_6_35b_a3b_1p_in3k5_out1k5_50ms.py
test_type: 'perf'
- name: qwen3_6_35b_a3b_1p_aime26
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/accuracy/qwen3_6_35b_a3b/test_npu_qwen3_6_35b_a3b_1p_aime26.py
- name: qwen3_6_35b_a3b_1p_in64k_out1k_50ms
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3_6_35b_a3b/test_npu_qwen3_6_35b_a3b_1p_in64k_out1k_50ms.py
test_type: 'perf'
- name: qwen3_6_35b_a3b_1p_in128k_out1k_50ms
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3_6_35b_a3b/test_npu_qwen3_6_35b_a3b_1p_in128k_out1k_50ms.py
test_type: 'perf'
- name: qwen3_6_35b_a3b_1p_in64k_out1k_prefix90_50ms_aime26
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3_6_35b_a3b/test_npu_qwen3_6_35b_a3b_1p_in64k_out1k_prefix90_50ms_aime26.py
test_type: 'perf'
- name: qwen3_6_35b_a3b_1p_in128k_out1k_prefix90_50ms
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3_6_35b_a3b/test_npu_qwen3_6_35b_a3b_1p_in128k_out1k_prefix90_50ms.py
test_type: 'perf'
# qwen3_6_27b performance tests
- name: qwen3_6_27b_w8a8_1p_in3k5_out1k5_50ms_gpqa
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3_6_27b/test_npu_qwen3_6_27b_w8a8_1p_in3k5_out1k5_50ms_gpqa.py
test_type: 'perf'
- name: qwen3_6_27b_w8a8_2p_in16k_out1k_50ms
runner: linux-aarch64-a3-4
test_case: test/registered/ascend/performance/qwen3_6_27b/test_npu_qwen3_6_27b_w8a8_2p_in16k_out1k_50ms.py
test_type: 'perf'
- name: qwen3_6_27b_w8a8_2p_in64k_out1k_50ms
runner: linux-aarch64-a3-4
test_case: test/registered/ascend/performance/qwen3_6_27b/test_npu_qwen3_6_27b_w8a8_2p_in64k_out1k_50ms.py
test_type: 'perf'
- name: qwen3_6_27b_w8a8_2p_in128k_out1k_50ms
runner: linux-aarch64-a3-4
test_case: test/registered/ascend/performance/qwen3_6_27b/test_npu_qwen3_6_27b_w8a8_2p_in128k_out1k_50ms.py
test_type: 'perf'
- name: qwen3_6_27b_2p_in64k_out1k_prefix90_50ms
runner: linux-aarch64-a3-4
test_case: test/registered/ascend/performance/qwen3_6_27b/test_npu_qwen3_6_27b_2p_in64k_out1k_prefix90_50ms.py
test_type: 'perf'
- name: qwen3_6_27b_1p_in1024x1024_30_out1024_50ms
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3_6_27b/test_npu_qwen3_6_27b_1p_in1024x1024_30_out1024_50ms.py
test_type: 'perf'
- name: qwen3_6_27b_1p_in1080p_30_out256_50ms
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3_6_27b/test_npu_qwen3_6_27b_1p_in1080p_30_out256_50ms.py
test_type: 'perf'
- name: qwen3_6_27b_1p_gpqa
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/accuracy/qwen3_6_27b/test_npu_qwen3_6_27b_1p_gpqa.py
# qwen3_32b performance tests
- name: qwen3_32b_w8a8_2p_in3k5_out1k5_50ms_gpqa
runner: linux-aarch64-a3-4
test_case: test/registered/ascend/performance/qwen3_32b/test_npu_qwen3_32b_w8a8_2p_in3k5_out1k5_50ms_gpqa.py
test_type: 'perf'
- name: qwen3_32b_bf16_8p_in18k_out4k_6ms
runner: linux-aarch64-a3-16
test_case: test/registered/ascend/performance/qwen3_32b/test_npu_qwen3_32b_bf16_8p_in18k_out4k_6ms.py
test_type: 'perf'
- name: qwen3_32b_bf16_8p_gpqa
runner: linux-aarch64-a3-16
test_case: test/registered/ascend/accuracy/qwen3_32b/test_npu_qwen3_32b_bf16_8p_gpqa.py
# qwen3_30b_a3b performance tests
- name: qwen3_30b_w8a8_1p_in3k5_out1k5_50ms_aime25
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3_30b_a3b/test_npu_qwen3_30b_w8a8_1p_in3k5_out1k5_50ms_aime25.py
test_type: 'perf'
# qwen3-8b performance tests
- name: qwen3_8b_w8a8_1p_in3k5_out1k5_50ms_gpqa
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3-8b/test_npu_qwen3_8b_w8a8_1p_in3k5_out1k5_50ms_gpqa.py
test_type: 'perf'
- name: qwen3_8b_w8a8_1p_in6k_out1k5_bs16_gpqa
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/performance/qwen3-8b/test_npu_qwen3_8b_w8a8_1p_in6k_out1k5_bs16_gpqa.py
test_type: 'perf'
# qwen3_next_80b_a3b_instruct performance tests
- name: qwen3_next_80b_w8a8_2p_in6k_out1k5_bs16_aime25
runner: linux-aarch64-a3-4
test_case: test/registered/ascend/performance/qwen3_next_80b_a3b_instruct/test_npu_qwen3_next_80b_w8a8_2p_in6k_out1k5_bs16_aime25.py
test_type: 'perf'
# minimax_m2_5 performance tests
- name: minimax_m2_5_w8a8_8p_in3k5_out1k5_50ms_gpqa
runner: linux-aarch64-a3-16
test_case: test/registered/ascend/performance/minimax_m2_5/test_npu_minimax_m2_5_w8a8_8p_in3k5_out1k5_50ms_gpqa.py
test_type: 'perf'
- name: minimax_m2_5_w8a8_4p_in64k_out1k_prefix90_50ms_gpqa
runner: linux-aarch64-a3-16
test_case: test/registered/ascend/performance/minimax_m2_5/test_npu_minimax_m2_5_w8a8_4p_in64k_out1k_prefix90_50ms_gpqa.py
test_type: 'perf'
# deepseek_v3_2 accuracy tests
- name: deepseek_v3_2_8p_aime25
runner: linux-aarch64-a3-16
test_case: test/registered/ascend/accuracy/deepseek_v3_2/test_npu_deepseek_v3_2_8p_aime25.py
# glm4_7_flash accuracy tests
- name: glm4_7_flash_1p_aime25
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/accuracy/glm4_7_flash/test_npu_glm4_7_flash_1p_aime25.py
# glm4_6v_flash accuracy tests
- name: glm4_6v_flash_1p_mmmu
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/accuracy/glm4_6v_flash/test_npu_glm4_6v_flash_1p_mmmu.py
# qwen3_vl_8b_thinking accuracy tests
- name: qwen3_vl_8b_thinking_1p_mmmu
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/accuracy/qwen3_vl_8b_thinking/test_npu_qwen3_vl_8b_thinking_1p_mmmu.py
# qwen3_vl_30b_a3b_thinking accuracy tests
- name: qwen3_vl_30b_a3b_thinking_1p_mmmu
runner: linux-aarch64-a3-2
test_case: test/registered/ascend/accuracy/qwen3_vl_30b_a3b_thinking/test_npu_qwen3_vl_30b_a3b_thinking_1p_mmmu.py
uses: ./.github/workflows/nightly-test-npu-e2e-single-node.yml
with:
runner: ${{ matrix.test_config.runner }}
test_type: ${{ matrix.test_config.test_type }}
test_config_name: ${{ matrix.test_config.name }}
test_case: ${{ matrix.test_config.test_case }}
image: ${{ needs.set-image-config.outputs.image_a3 }}
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ needs.set-image-config.outputs.ref || github.ref }}
install_sglang_from_source: false
transformers_version: ''
- name: Install dependencies
env:
TORCH_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/whl/cpu"
PYPI_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
UV_INDEX_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
GITHUB_PROXY_URL: "https://gh-proxy.test.osinfra.cn/"
run: |
# speed up by using infra cache services
CACHING_URL="cache-service.nginx-pypi-cache.svc.cluster.local"
sed -Ei "s@(ports|archive).ubuntu.com@${CACHING_URL}:8081@g" /etc/apt/sources.list
pip config set global.index-url http://${CACHING_URL}/pypi/simple
pip config set global.trusted-host "${CACHING_URL}"
if [ ${{ needs.set-image-config.outputs.skip_install_flag }} != "true" ];then
bash scripts/ci/npu/npu_ci_install_dependency.sh a3
fi
# copy required file from our daily cache
cp ~/.cache/modelscope/hub/datasets/otavia/ShareGPT_Vicuna_unfiltered/ShareGPT_V3_unfiltered_cleaned_split.json /tmp
# copy gsm8k dataset
cp ~/.cache/modelscope/hub/datasets/tmp/test.jsonl /tmp
- name: Print Log Information
run: |
bash scripts/ci/npu/npu_log_print.sh
- name: Run test
timeout-minutes: 240
env:
SGLANG_USE_MODELSCOPE: true
SGLANG_IS_IN_CI: true
HF_ENDPOINT: https://hf-mirror.com
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
run: |
pip install sglang_router
hf download lmms-lab/MMMU --repo-type dataset
pip install sentence_transformers torchaudio==2.10.0
pip install protobuf==6.31.1 zss pre-commit wandb>=0.16.0 tenacity==8.3.0 loguru openpyxl latex2sympy2 zstandard transformers-stream-generator tqdm-multiprocess pycocoevalcap
pip install yt-dlp sentencepiece==0.1.99 nltk av ftfy sqlitedict==2.1.0 sacrebleu>=1.5.0 pytablewriter black==24.1.0 isort==5.13.2 peft>=0.2.0 accelerate>=0.29.1
pip install jsonlines httpx==0.25.0 evaluate>=0.4.0 datasets==2.16.1 numexpr xgrammar==0.2.1 numpy==1.26.4 dotenv
git clone --branch v0.3.3 --depth 1 https://github.com/EvolvingLMMs-Lab/lmms-eval.git
cd ./lmms-eval
nohup pip install . > lmmslog.txt 2>&1 &
sleep 120
export PYTHONPATH=$PYTHONPATH:$(pwd)
cd ../
cd test
python3 run_suite.py --hw npu --suite nightly-2-npu-a3 --nightly --continue-on-error --timeout-per-file 3600 --auto-partition-id ${{ matrix.part }} --auto-partition-size 1
nightly-4-npu-a3:
needs: [set-image-config]
if: ${{ (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') }}
runs-on: linux-aarch64-a3-4
nightly-poc-multi-node-tests:
name: multi-node-poc
if: ${{ !cancelled() }}
needs: [set-image-config, nightly-poc-single-node-tests]
strategy:
fail-fast: false
max-parallel: 1
matrix:
part: [0]
container:
test_config:
# glm5_1 performance tests
- name: glm5_1_w4a8_1p1d_32p_in64k_out1k_50ms_aime26
prefill_size: 2
decode_size: 2
router_size: 1
test_case: test/registered/ascend/performance/glm5_1/test_npu_glm5_1_w4a8_1p1d_32p_in64k_out1k_50ms_aime26.py
test_type: 'perf'
prefill_decode_deployment: 'separation'
uses: ./.github/workflows/nightly-test-npu-e2e-multi-node.yml
with:
runner: linux-amd64-cpu-8
test_type: ${{ matrix.test_config.test_type }}
test_config_name: ${{ matrix.test_config.name }}
prefill_size: ${{ matrix.test_config.prefill_size }}
decode_size: ${{ matrix.test_config.decode_size }}
router_size: ${{ matrix.test_config.router_size }}
test_case: ${{ matrix.test_config.test_case }}
image: ${{ needs.set-image-config.outputs.image_a3 }}
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ needs.set-image-config.outputs.ref|| github.ref }}
install_sglang_from_source: false
prefill_decode_deployment: ${{ matrix.test_config.prefill_decode_deployment }}
transformers_version: ''
- name: Install dependencies
env:
TORCH_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/whl/cpu"
PYPI_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
UV_INDEX_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
GITHUB_PROXY_URL: "https://gh-proxy.test.osinfra.cn/"
run: |
# speed up by using infra cache services
CACHING_URL="cache-service.nginx-pypi-cache.svc.cluster.local"
sed -Ei "s@(ports|archive).ubuntu.com@${CACHING_URL}:8081@g" /etc/apt/sources.list
pip config set global.index-url http://${CACHING_URL}/pypi/simple
pip config set global.trusted-host "${CACHING_URL}"
if [ ${{ needs.set-image-config.outputs.skip_install_flag }} != "true" ];then
bash scripts/ci/npu/npu_ci_install_dependency.sh a3
fi
# copy required file from our daily cache
cp ~/.cache/modelscope/hub/datasets/otavia/ShareGPT_Vicuna_unfiltered/ShareGPT_V3_unfiltered_cleaned_split.json /tmp
# copy gsm8k dataset
cp ~/.cache/modelscope/hub/datasets/tmp/test.jsonl /tmp
- name: Print Log Information
run: |
bash scripts/ci/npu/npu_log_print.sh
- name: Run test
timeout-minutes: 240
env:
SGLANG_USE_MODELSCOPE: true
SGLANG_IS_IN_CI: true
HF_ENDPOINT: https://hf-mirror.com
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
run: |
pip install sglang_router
hf download lmms-lab/MMMU --repo-type dataset
pip install sentence_transformers torchaudio==2.10.0
pip install protobuf==6.31.1 zss pre-commit wandb>=0.16.0 tenacity==8.3.0 loguru openpyxl latex2sympy2 zstandard transformers-stream-generator tqdm-multiprocess pycocoevalcap
pip install yt-dlp sentencepiece==0.1.99 nltk av ftfy sqlitedict==2.1.0 sacrebleu>=1.5.0 pytablewriter black==24.1.0 isort==5.13.2 peft>=0.2.0 accelerate>=0.29.1
pip install jsonlines httpx==0.25.0 evaluate>=0.4.0 datasets==2.16.1 numexpr xgrammar==0.2.1 numpy==1.26.4 dotenv
git clone --branch v0.3.3 --depth 1 https://github.com/EvolvingLMMs-Lab/lmms-eval.git
cd ./lmms-eval
nohup pip install . > lmmslog.txt 2>&1 &
sleep 120
export PYTHONPATH=$PYTHONPATH:$(pwd)
cd ../
cd test
python3 run_suite.py --hw npu --suite nightly-4-npu-a3 --nightly --continue-on-error --timeout-per-file 3600 --auto-partition-id ${{ matrix.part }} --auto-partition-size 1
nightly-8-npu-a3:
needs: [set-image-config]
if: ${{ (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') }}
runs-on: linux-aarch64-a3-8
nightly-poc-multi-node-mix-tests:
name: multi-node-mix-poc
if: ${{ !cancelled() }}
needs: [set-image-config, nightly-poc-single-node-tests, nightly-poc-multi-node-tests]
strategy:
fail-fast: false
max-parallel: 1
matrix:
part: [0]
container:
test_config:
# kimi_k2_6 performance tests
- name: kimi_k2_6_w4a8_16p_in64k_out1k_100ms_aime25
node_size: 2
test_case: test/registered/ascend/performance/kimi_k2_6/test_npu_kimi_k2_6_w4a8_16p_in64k_out1k_100ms_aime25.py
test_type: 'perf'
uses: ./.github/workflows/nightly-test-npu-e2e-multi-node.yml
with:
runner: linux-amd64-cpu-8
test_type: ${{ matrix.test_config.test_type }}
test_config_name: ${{ matrix.test_config.name }}
node_size: ${{ matrix.test_config.node_size }}
test_case: ${{ matrix.test_config.test_case }}
image: ${{ needs.set-image-config.outputs.image_a3 }}
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ needs.set-image-config.outputs.ref || github.ref }}
- name: Install dependencies
env:
TORCH_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/whl/cpu"
PYPI_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
UV_INDEX_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
GITHUB_PROXY_URL: "https://gh-proxy.test.osinfra.cn/"
run: |
# speed up by using infra cache services
CACHING_URL="cache-service.nginx-pypi-cache.svc.cluster.local"
sed -Ei "s@(ports|archive).ubuntu.com@${CACHING_URL}:8081@g" /etc/apt/sources.list
pip config set global.index-url http://${CACHING_URL}/pypi/simple
pip config set global.trusted-host "${CACHING_URL}"
if [ ${{ needs.set-image-config.outputs.skip_install_flag }} != "true" ];then
bash scripts/ci/npu/npu_ci_install_dependency.sh a3
fi
# copy required file from our daily cache
cp ~/.cache/modelscope/hub/datasets/otavia/ShareGPT_Vicuna_unfiltered/ShareGPT_V3_unfiltered_cleaned_split.json /tmp
# copy gsm8k dataset
cp ~/.cache/modelscope/hub/datasets/tmp/test.jsonl /tmp
- name: Print Log Information
run: |
bash scripts/ci/npu/npu_log_print.sh
- name: Run test
timeout-minutes: 240
env:
SGLANG_USE_MODELSCOPE: true
SGLANG_IS_IN_CI: true
HF_ENDPOINT: https://hf-mirror.com
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
run: |
pip install sglang_router
hf download lmms-lab/MMMU --repo-type dataset
pip install sentence_transformers torchaudio==2.10.0
pip install protobuf==6.31.1 zss pre-commit wandb>=0.16.0 tenacity==8.3.0 loguru openpyxl latex2sympy2 zstandard transformers-stream-generator tqdm-multiprocess pycocoevalcap
pip install yt-dlp sentencepiece==0.1.99 nltk av ftfy sqlitedict==2.1.0 sacrebleu>=1.5.0 pytablewriter black==24.1.0 isort==5.13.2 peft>=0.2.0 accelerate>=0.29.1
pip install jsonlines httpx==0.25.0 evaluate>=0.4.0 datasets==2.16.1 numexpr xgrammar==0.2.1 numpy==1.26.4 dotenv
git clone --branch v0.3.3 --depth 1 https://github.com/EvolvingLMMs-Lab/lmms-eval.git
cd ./lmms-eval
nohup pip install . > lmmslog.txt 2>&1 &
sleep 120
export PYTHONPATH=$PYTHONPATH:$(pwd)
cd ../
cd test
python3 run_suite.py --hw npu --suite nightly-8-npu-a3 --nightly --continue-on-error --timeout-per-file 3600 --auto-partition-id ${{ matrix.part }} --auto-partition-size 1
nightly-16-npu-a3:
needs: [set-image-config]
if: ${{ (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') }}
runs-on: linux-aarch64-a3-16
strategy:
fail-fast: false
matrix:
part: [0, 1]
container:
image: ${{ needs.set-image-config.outputs.image_a3 }}
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ needs.set-image-config.outputs.ref || github.ref }}
- name: Install dependencies
env:
TORCH_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/whl/cpu"
PYPI_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
UV_INDEX_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
GITHUB_PROXY_URL: "https://gh-proxy.test.osinfra.cn/"
run: |
# speed up by using infra cache services
CACHING_URL="cache-service.nginx-pypi-cache.svc.cluster.local"
sed -Ei "s@(ports|archive).ubuntu.com@${CACHING_URL}:8081@g" /etc/apt/sources.list
pip config set global.index-url http://${CACHING_URL}/pypi/simple
pip config set global.trusted-host "${CACHING_URL}"
if [ ${{ needs.set-image-config.outputs.skip_install_flag }} != "true" ];then
bash scripts/ci/npu/npu_ci_install_dependency.sh a3
fi
# copy required file from our daily cache
cp ~/.cache/modelscope/hub/datasets/otavia/ShareGPT_Vicuna_unfiltered/ShareGPT_V3_unfiltered_cleaned_split.json /tmp
# copy gsm8k dataset
cp ~/.cache/modelscope/hub/datasets/tmp/test.jsonl /tmp
- name: Print Log Information
run: |
bash scripts/ci/npu/npu_log_print.sh
- name: Run test
timeout-minutes: 240
env:
SGLANG_USE_MODELSCOPE: true
SGLANG_IS_IN_CI: true
HF_ENDPOINT: https://hf-mirror.com
TORCH_EXTENSIONS_DIR: /tmp/torch_extensions
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
STREAMS_PER_DEVICE: 32
run: |
pip install sglang_router
hf download lmms-lab/MMMU --repo-type dataset
pip install sentence_transformers torchaudio==2.10.0
pip install protobuf==6.31.1 zss pre-commit wandb>=0.16.0 tenacity==8.3.0 loguru openpyxl latex2sympy2 zstandard transformers-stream-generator tqdm-multiprocess pycocoevalcap
pip install yt-dlp sentencepiece==0.1.99 nltk av ftfy sqlitedict==2.1.0 sacrebleu>=1.5.0 pytablewriter black==24.1.0 isort==5.13.2 peft>=0.2.0 accelerate>=0.29.1
pip install jsonlines httpx==0.25.0 evaluate>=0.4.0 datasets==2.16.1 numexpr xgrammar==0.2.1 numpy==1.26.4 dotenv
git clone --branch v0.3.3 --depth 1 https://github.com/EvolvingLMMs-Lab/lmms-eval.git
cd ./lmms-eval
nohup pip install . > lmmslog.txt 2>&1 &
sleep 120
export PYTHONPATH=$PYTHONPATH:$(pwd)
cd ../
cd test
python3 run_suite.py --hw npu --suite nightly-16-npu-a3 --nightly --continue-on-error --timeout-per-file 3600 --auto-partition-id ${{ matrix.part }} --auto-partition-size 2
install_sglang_from_source: false
prefill_decode_deployment: 'mix'
transformers_version: ''
check-all-jobs:
if: github.repository == 'sgl-project/sglang' && always()
if: ${{ !cancelled() }}
needs:
- nightly-1-npu-a3
- nightly-2-npu-a3
- nightly-4-npu-a3
- nightly-8-npu-a3
- nightly-16-npu-a3
- nightly-poc-single-node-a2-tests
- nightly-poc-single-node-tests
- nightly-poc-multi-node-tests
- nightly-poc-multi-node-mix-tests
runs-on: ubuntu-latest
container:
image: docker.m.daocloud.io/ubuntu:22.04
steps:
- name: Check if any job failed
- name: Download all metrics
uses: actions/download-artifact@v4
with:
pattern: metrics-*
path: /tmp/metrics
merge-multiple: false
- name: Generate results table
run: |
if [[ "${{ contains(needs.*.result, 'failure') }}" == "true" ]]; then
echo "One or more nightly test jobs failed"
exit 1
status_emoji() {
case "$1" in
pass) echo "✅" ;;
fail) echo "❌" ;;
*) echo "❓" ;;
esac
}
single_result_a2="${{ needs.nightly-poc-single-node-a2-tests.result }}"
single_result="${{ needs.nightly-poc-single-node-tests.result }}"
multi_result="${{ needs.nightly-poc-multi-node-tests.result }}"
mix_result="${{ needs.nightly-poc-multi-node-mix-tests.result }}"
group_icon() {
case "$1" in
success) echo "✅" ;;
failure) echo "❌" ;;
cancelled) echo "⏭️" ;;
skipped) echo "⏭️" ;;
*) echo "❓" ;;
esac
}
echo "## Nightly Test Results" >> $GITHUB_STEP_SUMMARY
echo "" >> $GITHUB_STEP_SUMMARY
echo "| Group | Status |" >> $GITHUB_STEP_SUMMARY
echo "|-------|--------|" >> $GITHUB_STEP_SUMMARY
echo "| single-node-poc-a2 | $(group_icon ${single_result_a2}) ${single_result_a2} |" >> $GITHUB_STEP_SUMMARY
echo "| single-node-poc | $(group_icon ${single_result}) ${single_result} |" >> $GITHUB_STEP_SUMMARY
echo "| multi-node-poc | $(group_icon ${multi_result}) ${multi_result} |" >> $GITHUB_STEP_SUMMARY
echo "| multi-node-mix-poc | $(group_icon ${mix_result}) ${mix_result} |" >> $GITHUB_STEP_SUMMARY
echo "" >> $GITHUB_STEP_SUMMARY
echo "## Per-Test Metrics" >> $GITHUB_STEP_SUMMARY
echo "" >> $GITHUB_STEP_SUMMARY
has_metrics=false
for dir in /tmp/metrics/metrics-*/; do
file="${dir}metrics.json"
if [ -f "$file" ]; then
has_metrics=true
echo "import json" > /tmp/parse_metrics.py
echo "with open('$file') as f:" >> /tmp/parse_metrics.py
echo " d = json.load(f)" >> /tmp/parse_metrics.py
echo "tc = d.get('test_case', '-')" >> /tmp/parse_metrics.py
echo "tp = d.get('test_type', '-')" >> /tmp/parse_metrics.py
echo "st = d.get('status', 'unknown')" >> /tmp/parse_metrics.py
echo "icon = '✅' if st == 'pass' else '❌'" >> /tmp/parse_metrics.py
echo "metrics = d.get('metrics', {})" >> /tmp/parse_metrics.py
echo "baselines = d.get('baselines', {})" >> /tmp/parse_metrics.py
echo "mstr = ', '.join(f'{k}={v}' for k,v in metrics.items()) if metrics else '-'" >> /tmp/parse_metrics.py
echo "bstr = ', '.join(f'{k}={v}' for k,v in baselines.items()) if baselines else '-'" >> /tmp/parse_metrics.py
echo "print(f'| {tc} | {tp} | {icon} {st} | {mstr} | {bstr} |')" >> /tmp/parse_metrics.py
python3 /tmp/parse_metrics.py
fi
done > /tmp/metrics_table.txt
if [ "${has_metrics}" = "true" ]; then
echo "| Test Case | Type | Status | Metrics | Baseline |" >> $GITHUB_STEP_SUMMARY
echo "|-----------|------|--------|---------|----------|" >> $GITHUB_STEP_SUMMARY
cat /tmp/metrics_table.txt >> $GITHUB_STEP_SUMMARY
else
echo "No per-test metrics available (artifacts not found)." >> $GITHUB_STEP_SUMMARY
fi
if [[ "${{ contains(needs.*.result, 'cancelled') }}" == "true" ]]; then
echo "One or more nightly test jobs were cancelled"
exit 1
fi
echo "All nightly test jobs passed"
echo "" >> $GITHUB_STEP_SUMMARY
FAIL=0
if [ "${single_result}" != "success" ] && [ "${single_result}" != "skipped" ]; then FAIL=1; fi
exit $FAIL