realliujiaxu bedf223771 [Perf] move quant before allgather in Allgather EP (#3420) ...

### What this PR does / why we need it?
move quant before allgather in Allgather EP, rely on
https://github.com/vllm-project/vllm-ascend/pull/3334

Deepseek R1 W8A8 performance on A2 with
`HCCL_ALGO="level0:NA;level1:pipeline"`:
| Seq length | Mean TTFT (ms) main | Mean TTFT (ms)  this PR |
|----------|----------|----------|
| 4k   |  375.21  | 364.99   |
| 16k  | 1465.23   | 1421.75  |
### Does this PR introduce _any_ user-facing change?

### How was this patch tested?


- vLLM version: v0.11.0
- vLLM main:
83f478bb19

---------

Signed-off-by: realliujiaxu <realliujiaxu@163.com>

2025-11-04 16:49:58 +08:00

..

__init__.py

[Core] Cherry pick from 0.7.1 to keep the main code newest (#127)

2025-02-21 17:07:37 +08:00

quant_config.py

【Bugfix】bugfix for weight load of kimi-k2 (#3798)

2025-10-27 21:18:35 +08:00

utils.py

[Feature] Add W4A4 Flat Quantization support (#3427)

2025-10-13 23:20:16 +08:00

w4a4_flatquant_dynamic.py

[main][bugfix] Add 'layer_type' param to get_pergroup_param() for compatibility (#3682)

2025-10-23 21:26:33 +08:00

w4a8_dynamic.py

[Perf] move quant before allgather in Allgather EP (#3420)

2025-11-04 16:49:58 +08:00

w8a8_dynamic.py

[Perf] move quant before allgather in Allgather EP (#3420)

2025-11-04 16:49:58 +08:00

w8a8.py

[Refactor] [MoE] Rename moe-related classes & files (#3646)

2025-10-25 11:22:03 +08:00