[Refact.]: Refactor some leftover implementations of 300I DUO in the main branch. (#6425)

### What this PR does / why we need it?
- Replace the RoPE operator implementation.
- Refactor some leftover implementations of 300I DUO in the main branch.

### Does this PR introduce _any_ user-facing change?
NA
### How was this patch tested?

- vLLM version: v0.14.1
- vLLM main:
dc917cceb8

---------

Signed-off-by: Tflowers-0129 <2906339855@qq.com>
This commit is contained in:
Shaoxu Cheng
2026-02-02 16:12:04 +08:00
committed by GitHub
parent eeedf7c503
commit 460ea88276
7 changed files with 94 additions and 23 deletions

View File

@@ -18,13 +18,13 @@
import torch
import torch.nn.functional as F
from vllm_ascend.ops.activation import AscendSiluAndMul as _Base
from vllm_ascend.ops.activation import AscendSiluAndMul
class AscendSiluAndMul310(_Base):
class AscendSiluAndMul310(AscendSiluAndMul):
def forward(self, x: torch.Tensor) -> torch.Tensor:
torch.ops.vllm.maybe_prefetch_mlp_down_proj(x)
h = x.shape[-1] // 2
out = (F.silu(x[..., :h].to(torch.float32)) * x[..., h:].to(torch.float32)).to(torch.float16)
out = F.silu(x[..., :h]) * x[..., h:]
torch.ops.vllm.maybe_wait_prefetch_done(out)
return out