How to judge whether IMMA kernel is chosed by cublasLt?

AI & Data Science Deep Learning (Training & Inference) Frameworks
,
1

I want to use cublasLt to execute high-performance int8 GEMM. Each descriptor which is needed by IMMA kernel is constructed strictly, but i am not sure the kernel chosen as follows is IMMA kernel.

24a9796aebe4f630823702f7cb7ff8a0
24a9796aebe4f630823702f7cb7ff8a01699×468 153 KB