CookieApril/enginex-ascend-910-vllm
1
0
Fork 0
forked from EngineX-Ascend/enginex-ascend-910-vllm
Code Pull Requests Activity

init v0.11.0rc0

Browse Source
This commit is contained in:
luopingyi
2025-10-14 10:38:28 +08:00
parent 67afd0ea78
commit 66dc16f966
278 changed files with 28130 additions and 11708 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
vllm_ascend/ops/__init__.py
View File

@@ -20,6 +20,7 @@ import torch
import vllm_ascend.ops.common_fused_moe # noqa
import vllm_ascend.ops.fused_moe # noqa
import vllm_ascend.ops.layernorm # noqa
import vllm_ascend.ops.register_custom_ops # noqa
import vllm_ascend.ops.vocab_parallel_embedding # noqa
from vllm_ascend.ops.activation import AscendQuickGELU, AscendSiluAndMul
from vllm_ascend.ops.rotary_embedding import (
@@ -34,19 +35,20 @@ class dummyFusionOp:
def register_dummy_fusion_op() -> None:
torch.ops._C.rms_norm = dummyFusionOp(name="rms_norm")
torch.ops._C.fused_add_rms_norm = dummyFusionOp(name="fused_add_rms_norm")
torch.ops._C.static_scaled_fp8_quant = dummyFusionOp(
torch.ops._C_ascend.rms_norm = dummyFusionOp(name="rms_norm")
torch.ops._C_ascend.fused_add_rms_norm = dummyFusionOp(
name="fused_add_rms_norm")
torch.ops._C_ascend.static_scaled_fp8_quant = dummyFusionOp(
name="static_scaled_fp8_quant")
torch.ops._C.dynamic_scaled_fp8_quant = dummyFusionOp(
torch.ops._C_ascend.dynamic_scaled_fp8_quant = dummyFusionOp(
name="dynamic_scaled_fp8_quant")
torch.ops._C.dynamic_per_token_scaled_fp8_quant = dummyFusionOp(
torch.ops._C_ascend.dynamic_per_token_scaled_fp8_quant = dummyFusionOp(
name="dynamic_per_token_scaled_fp8_quant")
torch.ops._C.rms_norm_static_fp8_quant = dummyFusionOp(
torch.ops._C_ascend.rms_norm_static_fp8_quant = dummyFusionOp(
name="rms_norm_static_fp8_quant")
torch.ops._C.fused_add_rms_norm_static_fp8_quant = dummyFusionOp(
torch.ops._C_ascend.fused_add_rms_norm_static_fp8_quant = dummyFusionOp(
name="fused_add_rms_norm_static_fp8_quant")
torch.ops._C.rms_norm_dynamic_per_token_quant = dummyFusionOp(
torch.ops._C_ascend.rms_norm_dynamic_per_token_quant = dummyFusionOp(
name="rms_norm_dynamic_per_token_quant")