A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 4, 22:30 UTC
[bugfix] Use only onlines CPUs in lscpu (#40161)
Signed-off-by: kse <kevin.sejourne@cloud-temple.com> Co-authored-by: kse <kevin.sejourne@cloud-temple.com>
Galigator committed
Apr 20, 2026 at 15:19 UTC
a023edfa5bc1982ba5c6365ea2e36c6dfc48a9ef
1 file changed
+1
-1
vllm/utils/cpu_resource_utils.py
+1
-1
@@ -155,7 +155,7 @@ def _get_cpu_list() -> list[LogicalCPUInfo]:
155
return [LogicalCPUInfo(i, i, 0) for i in range(cpu_count)]
156
157
lscpu_output = subprocess.check_output(
158
- "lscpu -J -e=CPU,CORE,NODE", shell=True, text=True
158
+ "lscpu --json --extended=CPU,CORE,NODE --online", shell=True, text=True
159
)
160
161
# For platform without NUMA, replace '-' to '0'