pu-zhe 5df450bca4 [Feat] [310p] Support w8a8sc quantization method (#7075) ...

### What this PR does / why we need it?
New Quantization Method: Introduced support for the W8A8SC static linear
quantization scheme specifically for 310P hardware, enabling more
efficient model compression.
Refactored the save_sharded_state_310.py to avoid multi-process issue.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
W8A8SC quant E2E test.

- vLLM version: v0.16.0
- vLLM main:
4034c3d32e

---------

Signed-off-by: pu-zhe <zpuaa@outlook.com>

2026-03-10 16:13:20 +08:00

..

attention

[300I] support decode-only aclgraph mode (#6849)

2026-03-02 14:15:14 +08:00

fused_moe

clean 0.15.0 support (#6852)

2026-02-28 09:20:57 +08:00

ops

[300I][Bugfix] fix unquant model weight nd2nz error (#6851)

2026-03-03 15:57:26 +08:00

quantization

[Feat] [310p] Support w8a8sc quantization method (#7075)

2026-03-10 16:13:20 +08:00

__init__.py

[Feature]: Support 310P device run qwen2.5/3 dense and qwen2.5vl models (#5776)

2026-01-17 11:49:18 +08:00

model_runner_310p.py

[misc] move mxfp_compat into device to decouple from quantization init chain (#6918)

2026-03-02 18:17:01 +08:00

sharded_state_loader_310p.py

[Feat][310p] 310P support w8a8s quantization and saving w8a8sc state (#6878)

2026-03-02 20:09:15 +08:00

worker_310p.py

[Feat][310p] 310P support w8a8s quantization and saving w8a8sc state (#6878)

2026-03-02 20:09:15 +08:00