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

[E2E] Optimize the E2E test time. (#5294)

Browse Source 
### What this PR does / why we need it?
Add cudagraph_capture_sizes for E2E CI test.

- vLLM version: release/v0.13.0
- vLLM main:
ad32e3e19c

Signed-off-by: menogrey <1299267905@qq.com>
This commit is contained in:
zhangyiming
2025-12-26 14:17:50 +08:00
committed by GitHub
parent 29d2fe653d
commit 45c5bcd962
22 changed files with 57 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
tests/e2e/multicard/long_sequence/test_basic.py
View File

@@ -122,6 +122,7 @@ def test_models_pcp_dcp_piece_wise():
decode_context_parallel_size=2,
max_num_batched_tokens=1024,
enable_expert_parallel=True,
cudagraph_capture_sizes=[1, 2, 4, 8],
block_size=128) as runner:
runner.model.generate(prompts, sampling_params)
@@ -132,6 +133,7 @@ def test_models_pcp_dcp_piece_wise():
prefill_context_parallel_size=2,
decode_context_parallel_size=1,
enable_expert_parallel=True,
cudagraph_capture_sizes=[1, 2, 4, 8],
block_size=128,
quantization="ascend") as runner:
runner.model.generate(prompts, sampling_params)