Quantize matmul in CPU avx2 have effect？ #189
Description
I want to know quantize matmul use gemmlowp have any effect? can improve the performance of inference？so can I export the python interfaces by .cc？
{{ message }}
I want to know quantize matmul use gemmlowp have any effect? can improve the performance of inference？so can I export the python interfaces by .cc？