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

[bugfix] fix w8a8dynamic fused_moe trans nz (#5199)

Browse Source 
### What this PR does / why we need it?
Currently, `torch_npu.npu_grouped_matmul_swiglu_quant` can only support
weight nz, so we need to trans w13_weight, w2_weight to nz forcely.

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

### How was this patch tested?

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

Signed-off-by: zzzzwwjj <1183291235@qq.com>
This commit is contained in:
zzzzwwjj
2025-12-22 17:45:34 +08:00
committed by GitHub
parent 55beac9c91
commit 052e472453
2 changed files with 52 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
vllm_ascend/quantization/w8a8_dynamic.py
View File

@@ -29,7 +29,7 @@ from vllm_ascend.ascend_forward_context import MoECommType
from vllm_ascend.distributed.parallel_state import get_mc2_group
from vllm_ascend.flash_common3_context import get_flash_common3_context
from vllm_ascend.ops.fused_moe.experts_selector import select_experts
from vllm_ascend.utils import maybe_trans_nz
from vllm_ascend.utils import ACL_FORMAT_FRACTAL_NZ, maybe_trans_nz
class AscendW8A8DynamicLinearMethod:
@@ -276,8 +276,12 @@ class AscendW8A8DynamicFusedMoEMethod:
1, 2).contiguous()
layer.w2_weight.data = layer.w2_weight.data.transpose(1,
2).contiguous()
layer.w13_weight.data = maybe_trans_nz(layer.w13_weight.data)
layer.w2_weight.data = maybe_trans_nz(layer.w2_weight.data)
# TODO(zzzzwwjj): Currently, `torch_npu.npu_grouped_matmul_swiglu_quant`
# can only support weight nz.
layer.w13_weight.data = torch_npu.npu_format_cast(
layer.w13_weight.data, ACL_FORMAT_FRACTAL_NZ)
layer.w2_weight.data = torch_npu.npu_format_cast(
layer.w2_weight.data, ACL_FORMAT_FRACTAL_NZ)
layer.w13_weight_scale.data = layer.w13_weight_scale.data.view(
layer.w13_weight_scale.data.shape[0], -1)
layer.w13_weight_scale_fp32 = layer.w13_weight_scale.data.to(