=== env ===
python 3.12.14
torch 2.13.0+cpu
torch_npu 2.13.0.rc1
vllm 0.28.1.post1.dev143+gf18cf803c.empty
vllm-ascend 0.25.1rc2.dev125+hust.20260903.4.g74f0c0a27
vllm-ascend-split-batch 0.1.0.dev0
vllm-hust-ext 0.2.0.dev0
ascend-kernel 2026.3.9
triton-ascend 3.2.2
CANN n/a
=== kernel wheel probe ===
INFO 09-09 15:44:29 [__init__.py:52] Available plugins for group vllm.platform_plugins:
INFO 09-09 15:44:29 [__init__.py:54] - ascend -> vllm_ascend:register
INFO 09-09 15:44:29 [__init__.py:57] All plugins in this group will be loaded. Set `VLLM_PLUGINS` to control which plugins to load.
INFO 09-09 15:44:29 [__init__.py:272] Platform plugin ascend is activated
INFO 09-09 15:44:29 [platform.py:70] Breakable CUDAGraph on Ascend is opt-in; using VLLM_USE_BREAKABLE_CUDAGRAPH=0.
probe_ok= True reason= ''
has fa_fp32_stage1: True
has lse_merge: True
=== card ===
	HBM Usage Rate(%)              : 5
