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

migrate NPU PR/nightly test cases to a3-560T (#33346)

Browse Source 
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: sglang-npu-bot <sglangnpu@163.com>
This commit is contained in:
Sugar920
2026-08-05 01:17:58 +08:00
committed by GitHub
co-authored by Claude sglang-npu-bot
parent 95d0e57e83
commit e76d0acdc9
33 changed files with 502 additions and 487 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/pr-test-npu.yml
+90 -196
View File
@@ -130,116 +130,10 @@ jobs:
cd test
python3 run_suite.py --hw npu --suite stage-a-unit-test-npu
stage-b-test-1-npu-a2:
needs: [check-changes, pr-gate, set-image-config]
if: needs.check-changes.outputs.main_package == 'true'
runs-on: linux-aarch64-a2-1
strategy:
fail-fast: false
matrix:
part: [ 0, 1 ]
container:
image: ${{ needs.set-image-config.outputs.CANN_image_910b }}
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ inputs.ref || github.ref }}
- name: Mark repository safe
run: |
git config --system --add safe.directory ${GITHUB_WORKSPACE}
- name: Install dependencies
env:
TORCH_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/whl/cpu"
PYPI_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
UV_INDEX_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
GITHUB_PROXY_URL: "https://gh-proxy.test.osinfra.cn/"
RUSTUP_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local:8082"
run: |
# speed up by using infra cache services
CACHING_URL="cache-service.nginx-pypi-cache.svc.cluster.local"
sed -Ei "s@(ports|archive).ubuntu.com@${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}"
bash scripts/ci/npu/npu_ci_install_dependency.sh 910b
# 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: Run test
timeout-minutes: 60
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: |
cd test
python3 run_suite.py --hw npu --suite stage-b-test-1-npu-a2 --auto-partition-id ${{ matrix.part }} --auto-partition-size 2
stage-b-test-2-npu-a2:
needs: [check-changes, pr-gate, set-image-config]
if: needs.check-changes.outputs.main_package == 'true'
runs-on: linux-aarch64-a2-2
strategy:
fail-fast: true
matrix:
part: [0, 1]
container:
image: ${{ needs.set-image-config.outputs.CANN_image_910b }}
steps:
- name: Checkout code
uses: actions/checkout@v4
with:
ref: ${{ inputs.ref || github.ref }}
- name: Mark repository safe
run: |
git config --system --add safe.directory ${GITHUB_WORKSPACE}
- name: Install dependencies
env:
TORCH_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/whl/cpu"
PYPI_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
UV_INDEX_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local/pypi/simple"
GITHUB_PROXY_URL: "https://gh-proxy.test.osinfra.cn/"
RUSTUP_CACHE_URL: "http://cache-service.nginx-pypi-cache.svc.cluster.local:8082"
run: |
# speed up by using infra cache services
CACHING_URL="cache-service.nginx-pypi-cache.svc.cluster.local"
sed -Ei "s@(ports|archive).ubuntu.com@${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}"
bash scripts/ci/npu/npu_ci_install_dependency.sh 910b
# 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: Run test
timeout-minutes: 60
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: |
cd test
python3 run_suite.py --hw npu --suite stage-b-test-2-npu-a2 --auto-partition-id ${{ matrix.part }} --auto-partition-size 2
stage-b-test-1-npu-a3:
needs: [check-changes, pr-gate, set-image-config]
if: needs.check-changes.outputs.main_package == 'true'
runs-on: linux-aarch64-a3-2
runs-on: linux-aarch64-a3-2-
container:
image: ${{ needs.set-image-config.outputs.CANN_image_a3 }}
steps:
@@ -289,7 +183,7 @@ jobs:
stage-b-test-2-npu-a3:
needs: [check-changes, pr-gate, set-image-config]
if: needs.check-changes.outputs.main_package == 'true'
runs-on: linux-aarch64-a3-2
runs-on: linux-aarch64-a3-2-
container:
image: ${{ needs.set-image-config.outputs.CANN_image_a3 }}
steps:
@@ -339,7 +233,7 @@ jobs:
stage-b-test-4-npu-a3:
needs: [check-changes, pr-gate, set-image-config]
if: needs.check-changes.outputs.main_package == 'true'
runs-on: linux-aarch64-a3-4
runs-on: linux-aarch64-a3-4-
strategy:
fail-fast: false
matrix:
@@ -393,7 +287,7 @@ jobs:
stage-b-test-8-npu-a3:
needs: [check-changes, pr-gate, set-image-config]
if: needs.check-changes.outputs.main_package == 'true'
runs-on: linux-aarch64-a3-8
runs-on: linux-aarch64-a3-8-
container:
image: ${{ needs.set-image-config.outputs.CANN_image_a3 }}
steps:
@@ -442,7 +336,7 @@ jobs:
stage-b-test-16-npu-a3:
needs: [check-changes, pr-gate, set-image-config]
if: needs.check-changes.outputs.main_package == 'true'
runs-on: linux-aarch64-a3-16
runs-on: linux-aarch64-a3-16-
container:
image: ${{ needs.set-image-config.outputs.CANN_image_a3 }}
steps:
@@ -493,7 +387,7 @@ jobs:
multimodal-gen-test-1-npu-a3:
needs: [check-changes, pr-gate, set-image-config]
if: needs.check-changes.outputs.multimodal_gen == 'true'
runs-on: linux-aarch64-a3-2
runs-on: linux-aarch64-a3-800t-2
container:
image: ${{ needs.set-image-config.outputs.CANN_image_a3 }}
steps:
@@ -550,7 +444,7 @@ jobs:
multimodal-gen-test-2-npu-a3:
needs: [check-changes, pr-gate, set-image-config]
if: needs.check-changes.outputs.multimodal_gen == 'true'
runs-on: linux-aarch64-a3-16
runs-on: linux-aarch64-a3-800t-2
container:
image: ${{ needs.set-image-config.outputs.CANN_image_a3 }}
steps:
@@ -615,105 +509,107 @@ jobs:
test_config:
# qwen3_6_27b performance tests
- name: qwen3_6_27b_w8a8_1p_in64k_out1k_50ms
runner: linux-aarch64-a3-2
runner: linux-aarch64-a3-800t-2
test_case: test/registered/npu/performance/qwen3_6_27b/test_npu_qwen3_6_27b_w8a8_1p_in64k_out1k_50ms.py
test_type: 'perf'
- name: qwen3_6_27b_1p_in1024x1024_30_out1024_50ms
runner: linux-aarch64-a3-2
test_case: test/registered/npu/performance/qwen3_6_27b/test_npu_qwen3_6_27b_1p_in1024x1024_30_out1024_50ms.py
test_type: 'perf'
# - name: qwen3_6_27b_1p_in1024x1024_30_out1024_50ms
# runner: linux-aarch64-a3-800t-2
# test_case: test/registered/npu/performance/qwen3_6_27b/test_npu_qwen3_6_27b_1p_in1024x1024_30_out1024_50ms.py
# test_type: 'perf'
# qwen3_8b performance tests
- name: qwen3_8b_w8a8_1p_in3k5_out1k5_50ms
runner: linux-aarch64-a3-2
test_case: test/registered/npu/performance/qwen3-8b/test_npu_qwen3_8b_w8a8_1p_in3k5_out1k5_50ms.py
test_type: 'perf'
# # qwen3_8b performance tests
# - name: qwen3_8b_w8a8_1p_in3k5_out1k5_50ms
# runner: linux-aarch64-a3-800t-2
# test_case: test/registered/npu/performance/qwen3-8b/test_npu_qwen3_8b_w8a8_1p_in3k5_out1k5_50ms.py
# test_type: 'perf'
# qwen3_30b_a3b performance tests
- name: qwen3_30b_w8a8_1p_in3k5_out1k5_50ms
runner: linux-aarch64-a3-2
test_case: test/registered/npu/performance/qwen3_30b_a3b/test_npu_qwen3_30b_w8a8_1p_in3k5_out1k5_50ms.py
test_type: 'perf'
# # qwen3_30b_a3b performance tests
# - name: qwen3_30b_w8a8_1p_in3k5_out1k5_50ms
# runner: linux-aarch64-a3-800t-2
# test_case: test/registered/npu/performance/qwen3_30b_a3b/test_npu_qwen3_30b_w8a8_1p_in3k5_out1k5_50ms.py
# test_type: 'perf'
# qwen3_6_35b_a3b performance tests
- name: qwen3_6_35b_a3b_1p_in64k_out1k_prefix90_50ms
runner: linux-aarch64-a3-2
test_case: test/registered/npu/performance/qwen3_6_35b_a3b/test_npu_qwen3_6_35b_a3b_1p_in64k_out1k_prefix90_50ms.py
test_type: 'perf'
# # qwen3_6_35b_a3b performance tests
# - name: qwen3_6_35b_a3b_1p_in64k_out1k_prefix90_50ms
# runner: linux-aarch64-a3-800t-2
# test_case: test/registered/npu/performance/qwen3_6_35b_a3b/test_npu_qwen3_6_35b_a3b_1p_in64k_out1k_prefix90_50ms.py
# test_type: 'perf'
# qwen3_vl_8b_thinking accuracy tests
- name: qwen3_vl_8b_thinking_1p_mmmu
runner: linux-aarch64-a3-2
test_case: test/registered/npu/accuracy/qwen3_vl_8b_thinking/test_npu_qwen3_vl_8b_thinking_1p_mmmu.py
test_type: 'accuracy'
# # qwen3_vl_8b_thinking accuracy tests
# - name: qwen3_vl_8b_thinking_1p_mmmu
# runner: linux-aarch64-a3-2-
# test_case: test/registered/npu/accuracy/qwen3_vl_8b_thinking/test_npu_qwen3_vl_8b_thinking_1p_mmmu.py
# test_type: 'accuracy'
# qwen3_32b performance tests
- name: qwen3_32b_w8a8_2p_in3k5_out1k5_50ms
runner: linux-aarch64-a3-4
test_case: test/registered/npu/performance/qwen3_32b/test_npu_qwen3_32b_w8a8_2p_in3k5_out1k5_50ms.py
test_type: 'perf'
# # qwen3_32b performance tests
# - name: qwen3_32b_w8a8_2p_in3k5_out1k5_50ms
# runner: linux-aarch64-a3-800t-4
# test_case: test/registered/npu/performance/qwen3_32b/test_npu_qwen3_32b_w8a8_2p_in3k5_out1k5_50ms.py
# test_type: 'perf'
# qwen3_next_80b_a3b performance tests
- name: qwen3_next_80b_w8a8_2p_in6k_out1k5_bs16
runner: linux-aarch64-a3-4
test_case: test/registered/npu/performance/qwen3_next_80b_a3b_instruct/test_npu_qwen3_next_80b_w8a8_2p_in6k_out1k5_bs16.py
test_type: 'perf'
# # qwen3_next_80b_a3b performance tests
# - name: qwen3_next_80b_w8a8_2p_in6k_out1k5_bs16
# runner: linux-aarch64-a3-800t-4
# test_case: test/registered/npu/performance/qwen3_next_80b_a3b_instruct/test_npu_qwen3_next_80b_w8a8_2p_in6k_out1k5_bs16.py
# test_type: 'perf'
# minimax_m2_5 performance tests
- name: minimax_m2_5_w8a8_4p_in64k_out1k_prefix90_50ms
runner: linux-aarch64-a3-8
test_case: test/registered/npu/performance/minimax_m2_5/test_npu_minimax_m2_5_w8a8_4p_in64k_out1k_prefix90_50ms.py
test_type: 'perf'
# # minimax_m2_5 performance tests
# - name: minimax_m2_5_w8a8_4p_in64k_out1k_prefix90_50ms
# runner: linux-aarch64-a3-800t-8
# test_case: test/registered/npu/performance/minimax_m2_5/test_npu_minimax_m2_5_w8a8_4p_in64k_out1k_prefix90_50ms.py
# test_type: 'perf'
# deepseek_v4_flash performance tests
- name: deepseek_v4_flash_w8a8_8p_in8k_out1k_50ms
runner: linux-aarch64-a3-16
test_case: test/registered/npu/performance/deepseek_v4_flash/test_npu_deepseek_v4_flash_w8a8_8p_in8k_out1k_50ms.py
test_type: 'perf'
# # deepseek_v4_flash performance tests
# - name: deepseek_v4_flash_w8a8_8p_in8k_out1k_50ms
# runner: linux-aarch64-a3-800t-16
# test_case: test/registered/npu/performance/deepseek_v4_flash/test_npu_deepseek_v4_flash_w8a8_8p_in8k_out1k_50ms.py
# test_type: 'perf'
# kimi_k2_6 performance tests
- name: kimi_k2_6_w4a8_8p_in3k5_out1k5_20ms
runner: linux-aarch64-a3-16
test_case: test/registered/npu/performance/kimi_k2_6/test_npu_kimi_k2_6_w4a8_8p_in3k5_out1k5_20ms.py
test_type: 'perf'
# # kimi_k2_6 performance tests
# - name: kimi_k2_6_w4a8_8p_in3k5_out1k5_20ms
# runner: linux-aarch64-a3-800t-16
# test_case: test/registered/npu/performance/kimi_k2_6/test_npu_kimi_k2_6_w4a8_8p_in3k5_out1k5_20ms.py
# test_type: 'perf'
# qwen3_235b performance tests
- name: qwen3_235b_w8a8_8p_in3k5_out1k5_50ms
runner: linux-aarch64-a3-16
test_case: test/registered/npu/performance/qwen3_235b_a22b/test_npu_qwen3_235b_w8a8_8p_in3k5_out1k5_50ms.py
test_type: 'perf'
# # qwen3_235b performance tests
# - name: qwen3_235b_w8a8_8p_in3k5_out1k5_50ms
# runner: linux-aarch64-a3-800t-16
# test_case: test/registered/npu/performance/qwen3_235b_a22b/test_npu_qwen3_235b_w8a8_8p_in3k5_out1k5_50ms.py
# test_type: 'perf'
# # qwen3_5_397b performance tests
# - name: qwen3_5_397b_w4a8_8p_in3k5_out1k5_50ms
# runner: linux-aarch64-a3-800t-16
# test_case: test/registered/npu/performance/qwen3_5_397b/test_npu_qwen3_5_397b_w4a8_8p_in3k5_out1k5_50ms.py
# test_type: 'perf'
# qwen3_5_397b performance tests
- name: qwen3_5_397b_w4a8_8p_in3k5_out1k5_50ms
runner: linux-aarch64-a3-16
test_case: test/registered/npu/performance/qwen3_5_397b/test_npu_qwen3_5_397b_w4a8_8p_in3k5_out1k5_50ms.py
test_type: 'perf'
# NPU accuracy tests
- name: glm4_7_flash_1p_gsm8k
runner: linux-aarch64-a3-2
test_case: test/registered/npu/accuracy/glm4_7_flash/test_npu_glm4_7_flash_1p_gsm8k.py
test_type: 'accuracy'
- name: qwen3_vl_30b_a3b_bf16_2p_gsm8k
runner: linux-aarch64-a3-4
test_case: test/registered/npu/accuracy/qwen3_vl_30b_a3b/test_npu_qwen3_vl_30b_a3b_bf16_2p_gsm8k.py
test_type: 'accuracy'
# - name: glm4_7_flash_1p_gsm8k
# runner: linux-aarch64-a3-2-
# test_case: test/registered/npu/accuracy/glm4_7_flash/test_npu_glm4_7_flash_1p_gsm8k.py
# test_type: 'accuracy'
# - name: qwen3_vl_30b_a3b_bf16_2p_gsm8k
# runner: linux-aarch64-a3-4-
# test_case: test/registered/npu/accuracy/qwen3_vl_30b_a3b/test_npu_qwen3_vl_30b_a3b_bf16_2p_gsm8k.py
# test_type: 'accuracy'
- name: glm5_top64_pruned_bf16_8p_gsm8k
runner: linux-aarch64-a3-16
runner: linux-aarch64-a3-16-
test_case: test/registered/npu/accuracy/glm5_top64_pruned/test_npu_glm5_top64_pruned_bf16_8p_gsm8k.py
test_type: 'accuracy'
- name: moonshotai_moonlight_16b_a3b_bf16_1p_gsm8k
runner: linux-aarch64-a3-2
test_case: test/registered/npu/accuracy/moonshotai_moonlight_16b_a3b/test_npu_moonlight_16b_a3b_bf16_1p_gsm8k.py
test_type: 'accuracy'
- name: qwen3_5_9b_bf16_1p_gsm8k
runner: linux-aarch64-a3-2
test_case: test/registered/npu/accuracy/qwen3_5_9b/test_npu_qwen3_5_9b_bf16_1p_gsm8k.py
test_type: 'accuracy'
- name: qwen3_vl_8b_bf16_2p_gsm8k
runner: linux-aarch64-a3-4
test_case: test/registered/npu/accuracy/qwen3_vl_8b/test_npu_qwen3_vl_8b_bf16_2p_gsm8k.py
test_type: 'accuracy'
# - name: moonshotai_moonlight_16b_a3b_bf16_1p_gsm8k
# runner: linux-aarch64-a3-2-
# test_case: test/registered/npu/accuracy/moonshotai_moonlight_16b_a3b/test_npu_moonlight_16b_a3b_bf16_1p_gsm8k.py
# test_type: 'accuracy'
# - name: qwen3_5_9b_bf16_1p_gsm8k
# runner: linux-aarch64-a3-2-
# test_case: test/registered/npu/accuracy/qwen3_5_9b/test_npu_qwen3_5_9b_bf16_1p_gsm8k.py
# test_type: 'accuracy'
# - name: qwen3_vl_8b_bf16_2p_gsm8k
# runner: linux-aarch64-a3-4-
# test_case: test/registered/npu/accuracy/qwen3_vl_8b/test_npu_qwen3_vl_8b_bf16_2p_gsm8k.py
# test_type: 'accuracy'
uses: ./.github/workflows/nightly-test-npu-e2e-single-node.yml
with:
runner: ${{ matrix.test_config.runner }}
@@ -732,9 +628,7 @@ jobs:
check-changes,
stage-a-unit-test-npu,
stage-b-test-1-npu-a2,
stage-b-test-1-npu-a3,
stage-b-test-2-npu-a2,
stage-b-test-2-npu-a3,
stage-b-test-4-npu-a3,
stage-b-test-8-npu-a3,