google/gemmlowp

Quantize matmul in CPU avx2 have effect?

zhoujianqian opened this issue · 0 comments

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?