whx a58ff9e92f [Cherry-pick] Port MoE multi-stream fix to v0.11.0-dev (#3753) ...

This PR moves the communication operation of shared experts out of extra
stream because I found that this might cause rtMemcpy related errors
when running shared experts multistream with aclgraph.

Furthermore, I utilize a global variable as extra stream object to avoid
allocating streams for each layer in full-graph mode.

Signed-off-by: whx-sjtu <2952154980@qq.com>

2025-10-25 15:51:43 +08:00

..

310p

Refactor e2e CI (#2276)

2025-09-02 09:02:22 +08:00

doctests

[CI]: Fix doctest ci for main release (#3451)

2025-10-16 14:38:11 +08:00

models

ACLgraph enable: Test cases revisions for all features (#3388)

2025-10-17 17:15:19 +08:00

multicard

[Feat][quantization] Support new version w4a8 dynamic quantization for Linear layers (#3311)

2025-10-21 20:18:39 +08:00

nightly

[CI]Add EPLB CI. (#3568)

2025-10-21 22:58:02 +08:00

pd_disaggreate

ACLgraph enable: Test cases revisions for all features (#3388)

2025-10-17 17:15:19 +08:00

prompts

[CI/UT] Add test for chunk prefill and prefix cache on v1/AscendScheduler (#1505)

2025-07-02 16:57:03 +08:00

singlecard

[Cherry-pick] Port MoE multi-stream fix to v0.11.0-dev (#3753)

2025-10-25 15:51:43 +08:00

vllm_interface

[bugFix] Correct the vllm interface e2e test Base container image name (#3179)

2025-09-25 16:03:09 +08:00

__init__.py

[Test] Clean up duplicate test for ascend scheduler (#1819)

2025-07-16 17:57:48 +08:00

common.sh

Increase doctest timeout to 300s and time print (#3041)

2025-09-19 20:26:00 +08:00

conftest.py

[CI] Refator multi-node CI (#3487)

2025-10-17 09:04:31 +08:00

model_utils.py

[Misc] Clean up useless patch (#3320)

2025-10-09 14:07:26 +08:00

run_disagg_pd.sh

[CI] Fix PD job (#1129)

2025-06-09 16:34:41 +08:00

run_doctests.sh

Increase doctest timeout to 300s and time print (#3041)

2025-09-19 20:26:00 +08:00

utils.py

[Test] Remove VLLM_USE_V1 in example and tests (#1733)

2025-07-15 12:49:57 +08:00