EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[Test] Add deepseek v3.2 exp nightly test (#4191)

Browse Source 
### What this PR does / why we need it?

- skip the nightly image build when the github event is pull_request
- set imagepullpolicy as alway for multi_node test
- move multi_node tests ahead to have some resource clean first
- do not relevant nightly image build with nightly tests for tolerance

- vLLM version: v0.11.0
- vLLM main:
2918c1b49c

---------

Signed-off-by: hfadzxy <starmoon_zhang@163.com>
Signed-off-by: wangli <wangli858794774@gmail.com>
Co-authored-by: wangli <wangli858794774@gmail.com>
This commit is contained in:
zhangxinyuehfad
2025-11-14 15:46:10 +08:00
committed by GitHub
parent 1d0f13c1a3
commit 67f2b3a031
7 changed files with 113 additions and 106 deletions
Download Patch File Download Diff File Expand all files Collapse all files

96
.github/workflows/vllm_ascend_test_nightly_a3.yaml vendored
View File

@@ -41,18 +41,53 @@ concurrency:
cancel-in-progress: true
jobs:
image_build:
name: nightly image build
uses: ./.github/workflows/_nightly_image_build.yaml
multi-node-tests:
name: multi-node
if: always() && (github.event_name == 'schedule' || github.event_name == 'workflow_dispatch')
strategy:
fail-fast: false
max-parallel: 1
matrix:
test_config:
- name: multi-node-deepseek-pd
config_file_path: DeepSeek-V3.yaml
size: 2
- name: multi-node-qwen3-dp
config_file_path: Qwen3-235B-A3B.yaml
size: 2
- name: multi-node-dpsk-4node-pd
config_file_path: DeepSeek-R1-W8A8.yaml
size: 4
- name: multi-node-qwenw8a8-2node
config_file_path: Qwen3-235B-W8A8.yaml
size: 2
- name: multi-node-glm-2node
config_file_path: GLM-4_5.yaml
size: 2
- name: multi-node-dpsk3.2-exp-2node
config_file_path: DeepSeek-V3_2-Exp-bf16.yaml
size: 2
- name: multi-node-deepseek-r1-w8a8-eplb
config_file_path: DeepSeek-R1-W8A8-EPLB.yaml
size: 4
- name: multi-node-qwenw8a8-2node-eplb
config_file_path: Qwen3-235B-W8A8-EPLB.yaml
size: 2
uses: ./.github/workflows/_e2e_nightly_multi_node.yaml
with:
target: a3
soc_version: a3
runner: linux-aarch64-a3-0
image: 'swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:nightly-a3'
replicas: 1
size: ${{ matrix.test_config.size }}
config_file_path: ${{ matrix.test_config.config_file_path }}
secrets:
HW_USERNAME: ${{ secrets.HW_USERNAME }}
HW_TOKEN: ${{ secrets.HW_TOKEN }}
KUBECONFIG_B64: ${{ secrets.KUBECONFIG_B64 }}
single-node-tests:
name: single-node
if: always() && (github.event_name == 'schedule' || github.event_name == 'workflow_dispatch')
needs: image_build
needs: multi-node-tests
strategy:
fail-fast: false
matrix:
@@ -103,51 +138,6 @@ jobs:
with:
vllm: v0.11.0
runner: ${{ matrix.test_config.os }}
image: ${{ fromJSON(format('"{0}"', needs.image_build.outputs.image-tag || 'swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:nightly-a3')) }}
image: 'swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:nightly-a3'
tests: ${{ matrix.test_config.tests }}
name: ${{ matrix.test_config.name }}
multi-node-tests:
name: multi-node
needs: [single-node-tests, image_build]
if: always() && (github.event_name == 'schedule' || github.event_name == 'workflow_dispatch')
strategy:
fail-fast: false
max-parallel: 1
matrix:
test_config:
- name: multi-node-deepseek-pd
config_file_path: DeepSeek-V3.yaml
size: 2
- name: multi-node-qwen3-dp
config_file_path: Qwen3-235B-A3B.yaml
size: 2
- name: multi-node-dpsk-4node-pd
config_file_path: DeepSeek-R1-W8A8.yaml
size: 4
- name: multi-node-qwenw8a8-2node
config_file_path: Qwen3-235B-W8A8.yaml
size: 2
- name: multi-node-glm-2node
config_file_path: GLM-4_5.yaml
size: 2
- name: multi-node-dpsk3.2-exp-2node
config_file_path: DeepSeek-V3_2-Exp-bf16.yaml
size: 2
- name: multi-node-deepseek-r1-w8a8-eplb
config_file_path: DeepSeek-R1-W8A8-EPLB.yaml
size: 4
- name: multi-node-qwenw8a8-2node-eplb
config_file_path: Qwen3-235B-W8A8-EPLB.yaml
size: 2
uses: ./.github/workflows/_e2e_nightly_multi_node.yaml
with:
soc_version: a3
runner: linux-aarch64-a3-0
image: ${{ fromJSON(format('"{0}"', needs.image_build.outputs.image-tag || 'swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:nightly-a3')) }}
replicas: 1
size: ${{ matrix.test_config.size }}
config_file_path: ${{ matrix.test_config.config_file_path }}
secrets:
KUBECONFIG_B64: ${{ secrets.KUBECONFIG_B64 }}