[Graph][Fusion] Integrating inductor pass and npugraph ex pass (#6354)

### What this PR does / why we need it?
Integrating inductor pass and npugraph ex pass, see RFC:
https://github.com/vllm-project/vllm-ascend/issues/6347

### Does this PR introduce _any_ user-facing change?
N/A

### How was this patch tested?
all tests passed.

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

---------

Signed-off-by: wxsIcey <1790571317@qq.com>
This commit is contained in:
Icey
2026-02-13 15:34:55 +08:00
committed by GitHub
parent 87a0b7b7c7
commit 7164990904
16 changed files with 220 additions and 909 deletions

View File

@@ -30,7 +30,7 @@ from vllm.platforms import Platform, PlatformEnum
# todo: please remove it when solve cuda hard code in vllm
os.environ["VLLM_DISABLE_SHARED_EXPERTS_STREAM"] = "1"
from vllm_ascend.ascend_config import get_ascend_config, init_ascend_config
from vllm_ascend.ascend_config import init_ascend_config
# isort: off
from vllm_ascend.utils import (
@@ -120,11 +120,7 @@ class NPUPlatform(Platform):
Get the pass manager class for this platform.
It will be registered as a custom pass under the current_platform.pass_key.
"""
npugraph_ex_config = get_ascend_config().npugraph_ex_config
if npugraph_ex_config.enable:
return "vllm_ascend.compilation.npu_graph_ex_pass_manager.NpuGraphEXPassManager"
else:
return "vllm_ascend.compilation.graph_fusion_pass_manager.GraphFusionPassManager"
return "vllm_ascend.compilation.graph_fusion_pass_manager.GraphFusionPassManager"
@classmethod
def get_compile_backend(self) -> str: