Cxj

[FEATURE] Enhance pack model speed

1

**Is your feature request related to a problem? Please describe.** when I quant [Qwen1.5-4B](https://huggingface.co/Qwen/Qwen1.5-4B) packing model take much time for some layer ```python os.environ['OMP_NUM_THREADS'] = '12' os.environ['OPENBLAS_NUM_THREADS'] = '12' os.environ['MKL_NUM_THREADS']...