minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

Revert "bugfix: prioritize init_npu_backend to fix various initialization bugs" (#19343)

Browse Source
This commit is contained in:
Even Zhou
2026-02-25 23:04:30 +08:00
committed by GitHub
parent c7c4a1cbbd
commit 2fb239450e
2 changed files with 6 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/srt/model_executor/model_runner.py
+5
View File
@@ -194,6 +194,11 @@ _is_npu = is_npu()
_is_cpu_amx_available = cpu_has_amx_support() _is_cpu_amx_available = cpu_has_amx_support()
_is_cpu_arm64 = is_host_cpu_arm64() _is_cpu_arm64 = is_host_cpu_arm64()
if _is_npu:
from sglang.srt.hardware_backend.npu.utils import init_npu_backend
init_npu_backend()
MLA_ATTENTION_BACKENDS = [ MLA_ATTENTION_BACKENDS = [
"aiter", "aiter",
"flashinfer", "flashinfer",
python/sglang/srt/server_args.py
+1 -6
View File
@@ -891,12 +891,7 @@ class ServerArgs:
def _handle_npu_backends(self): def _handle_npu_backends(self):
if self.device == "npu": if self.device == "npu":
from sglang.srt.hardware_backend.npu.utils import ( from sglang.srt.hardware_backend.npu.utils import set_default_server_args
init_npu_backend,
set_default_server_args,
)
init_npu_backend()
set_default_server_args(self) set_default_server_args(self)