How to test ' spent on GEMM kernels ' ? what tool used ?

Tencent / TurboTransformers

a fast and user-friendly runtime for transformer inference (Bert, Albert, GPT2, Decoders, etc) on CPU and GPU.

Other

1.49k stars 198 forks source link

Closed hadoop2xu closed 4 years ago

hadoop2xu commented 4 years ago

请问下大佬这个占比是用什么工具分析出来的？

feifeibear commented 4 years ago

nvvp工具看time line得到

feifeibear commented 4 years ago

代码里也有手工计时函数，可以打印一些丑陋的日志。