EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
1,740 Commits 3 Branches 0 Tags
cc23067f1ea0dad9d461aaaf816ea4ae2792767f
Commit Graph

10 Commits

Author SHA1 Message Date
SILONG ZENG
e56dba9b0d [CI]cleanup e2e test (#4800) 
### What this PR does / why we need it?
This PR refactors the E2E multicard test suite to improve test case
identification and maintainability. Specifically, it renames various
test functions to be more descriptive (explicitly indicating model
families like Qwen/DeepSeek and parallelism strategies like DP/TP/PP/EP)
and cleans up outdated or redundant test configurations in the offline
distributed inference tests.

**Key Changes:**
1. Test Function Renaming (Standardization): Renamed multiple test
functions across **`tests/e2e/multicard/`** to include clear
suffixes/prefixes regarding the model and parallel strategy. This helps
differentiate test cases in CI logs and prevents naming collisions.

**`test_aclgraph_capture_replay.py`:** 
- `test_aclgraph_capture_replay_dp2` ->
`test_aclgraph_capture_replay_metrics_dp2`

**`test_data_parallel.py`:**
- `test_data_parallel_inference` -> `test_qwen_inference_dp2`

**`test_data_parallel_tp2.py`:**
- `test_data_parallel_inference` -> `test_qwen_inference_dp2_tp2`

**`test_expert_parallel.py`:**
- `test_e2e_ep_correctness` -> `test_deepseek_correctness_ep`

**`test_external_launcher.py`:**
- `test_external_launcher` -> `test_qwen_external_launcher`
- `test_moe_external_launcher` -> `test_qwen_moe_external_launcher_ep`
- `test_external_launcher_and_sleepmode` ->
`test_qwen_external_launcher_with_sleepmode`
- `test_external_launcher_and_sleepmode_level2` ->
`test_qwen_external_launcher_with_sleepmode_level2`
- `test_mm_allreduce` ->
`test_qwen_external_launcher_with_matmul_allreduce`

**`test_full_graph_mode.py`:** 
- `test_models_distributed_Qwen3_MOE_TP2_WITH_FULL_DECODE_ONLY` ->
`test_qwen_moe_with_full_decode_only`
- `test_models_distributed_Qwen3_MOE_TP2_WITH_FULL` ->
`test_qwen_moe_with_full`

**`test_fused_moe_allgather_ep.py`:** 
- `test_generate_with_allgather `->
`test_deepseek_moe_fused_allgather_ep`
- `test_generate_with_alltoall` -> `test_deepseek_moe_fused_alltoall_ep`

**`test_offline_weight_load.py`:**
- `test_offline_weight_load_and_sleepmode` ->
`test_qwen_offline_weight_load_and_sleepmode`

**`test_pipeline_parallel.py`:**
- `test_models` -> `test_models_pp2`

2. Distributed Inference Cleanup
(**`test_offline_inference_distributed.py`**):

**model list changes:**
```
QWEN_DENSE_MODELS = [
-     "vllm-ascend/Qwen3-8B-W8A8", "vllm-ascend/Qwen2.5-0.5B-Instruct-W8A8"
+     "vllm-ascend/Qwen3-8B-W8A8",
]
```

```
- QWEN_W4A8_OLD_VERSION_MODELS = [
-    "vllm-ascend/Qwen3-8B-W4A8",
- ]

- QWEN_W4A8_NEW_VERSION_MODELS = [
-     "vllm-ascend/DeepSeek-V3-W4A8-Pruing",
-     "vllm-ascend/DeepSeek-V3.1-W4A8-puring",
- ]

+ DEEPSEEK_W4A8_MODELS = [
+      "vllm-ascend/DeepSeek-V3.1-W4A8-puring",
+ ]
```

**Test Function Changes:**
- removed `test_models_distributed_QwQ`
- removed `test_models_distributed_Qwen3_W8A8`
- removed `test_models_distributed_Qwen3_W4A8DYNAMIC_old_version`
- `test_models_distributed_Qwen3_W4A8DYNAMIC_new_version` ->
`test_models_distributed_Qwen3_W4A8DYNAMIC`

- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

---------

Signed-off-by: MrZ20 <2609716663@qq.com>
 2025-12-11 20:35:32 +08:00
wangxiyuan
27b09ca9b9 [CI] drop ascend scheduler test (#4582) 
let' drop ascend scheduler test first to ensure all function works
without it.


- vLLM version: v0.11.2
- vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.2

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
 2025-12-01 20:33:50 +08:00
Mengqing Cao
517fd9272d Revert "drop ascend scheduler" (#4580) 
Reverts vllm-project/vllm-ascend#4498
- vLLM version: v0.11.2
- vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.2
 2025-11-29 22:20:48 +08:00
wangxiyuan
f10acddb78 drop ascend scheduler (#4498) 
Ascend scheduler was added for non chunk prefill case before, since that
the npu ops didn't work well with chunked prefill.

Now the ops with chunked prefill work better, it's time to remove the
ascend scheduler to use vLLM default scheduler.

- vLLM version: v0.11.2

---------

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
 2025-11-29 16:18:34 +08:00
wangxiyuan
fef18b60bc Refactor e2e CI (#2276) 
Refactor E2E CI to make it clear and faster
1. remove some uesless e2e test
2. remove some uesless function
3. Make sure all test runs with VLLMRunner to avoid oom error
4. Make sure all ops test end with torch.empty_cache to avoid oom error
5. run the test one by one to avoid resource limit error


- vLLM version: v0.10.1.1
- vLLM main:
a344a5aa0a

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
 2025-09-02 09:02:22 +08:00
Li Wang
f60bb474f9 [CI] Enable linux-aarch64-a2 (64GB) and tp2 * 2 max-parallel to speed up CI (#2065) 
### What this PR does / why we need it?
Currently our workflow run time takes about 3 hours in total, which
seriously affects the developer experience, so it is urgent to have a
optimization, after this pr, It is expected that the running time of the
full CI can be shortened to 1h40min.

- Enable linux-aarch64-a2 (64GB) to replace linux-arm64-npu (32GB)
- Change TP4 ---> TP2 * 2 max-parallel
- Move DeepSeek-V2-Lite-W8A8 to single card test

### Does this PR introduce _any_ user-facing change?
No


- vLLM version: v0.10.0
- vLLM main:
a2480251ec

---------

Signed-off-by: wangli <wangli858794774@gmail.com>
 2025-07-29 18:59:05 +08:00
Pleaplusone
df0ec55162 Disaggregate prefill for kv cache register style (#950) 
### What this PR does / why we need it?
This PR adopt `LLMDataDist` for kv cache register and `pull_blocks`
style disaggregate prefill implementation. The interface implementation
mainly follows the design of NIXL PR
https://github.com/vllm-project/vllm/pull/17751/files#diff-7eaad0b7dee0626bf29d10081b0f0c5e3ea15a4af97e7b182a4e0d35f8346953
.

This PR can be test with the following step:
- Generate the rank table for all machine.
- execute`toy_proxy.py` to launch the disaggregate prefill proxy server,
specify the prefill ip, port and the decode ip, port
- Run the prefill server and decode server.
- send the request to the disaggregate prefill proxy

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?


- vLLM version: v0.9.2
- vLLM main:
8d0a01a5f2

---------

Signed-off-by: ganyi <pleaplusone.gy@gmail.com>
Signed-off-by: machenglong <machenglong_yewu@cmss.chinamobile.com>
Signed-off-by: liziyu179 <3475441767@qq.com>
Signed-off-by: underfitc <hucong24@huawei.com>
Signed-off-by: zouyida2052 <zouyida@huawei.com>
Signed-off-by: liziyu <liziyu16@huawei.com>
Signed-off-by: underfituu <hzhucong@163.com>
Co-authored-by: machenglong <machenglong_yewu@cmss.chinamobile.com>
Co-authored-by: liziyu179 <3475441767@qq.com>
Co-authored-by: underfitc <hucong24@huawei.com>
Co-authored-by: zouyida2052 <zouyida@huawei.com>
Co-authored-by: liziyu <liziyu16@huawei.com>
Co-authored-by: underfituu <hzhucong@163.com>
 2025-07-26 17:15:47 +08:00
Mengqing Cao
8cfd257992 [Dist][EP] Remove ETP/EP maintained in vllm-ascend (#1681) 
### What this PR does / why we need it?
Remove ETP/EP maintained in branch main. We drop this as there is no
relevant scenarios to use ETP now, and we may subsequently advocate
implementing expert tensor parallelism in vLLM to support scenarios
where the expert is needed to be sliced

This is a part of #1422 backport.

Fixes https://github.com/vllm-project/vllm-ascend/issues/1396
https://github.com/vllm-project/vllm-ascend/issues/1154

### Does this PR introduce _any_ user-facing change?
We'll not maintain etp/ep in vllm-ascend anymore, and use the tp/ep in
vllm instead.

### How was this patch tested?
CI passed with new added and existing test.


- vLLM version: v0.9.2
- vLLM main:
fe8a2c544a

Signed-off-by: MengqingCao <cmq0113@163.com>
 2025-07-21 09:08:04 +08:00
wangxiyuan
787010a637 [Test] Remove VLLM_USE_V1 in example and tests (#1733) 
V1 is enabled by default, no need to set it by hand now. This PR remove
the useless setting in example and tests

- vLLM version: v0.9.2
- vLLM main:
9ad0a4588b

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
 2025-07-15 12:49:57 +08:00
wangxiyuan
a054f0f4ca [CI] change to new ds model (#1513) 
Previous, the DeepSeek V3 Pruning weight is not correct, the moe layer
is not tested. We update a new Pruning model to enable moe layer
compute.

This PR fix the CI to address the new weight.

---------

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
 2025-06-30 19:02:29 +08:00