News for hackers. Stuff that matters.

A comment on How to Optimize a CUDA Matmul Kernel for CuBLAS-Like Performance: A Worklog. Parent comment

Re:How to Optimize a CUDA Matmul Kernel for CuBLAS-Like Perf... by david-gpu on Friday July 26, 2024 @03:33PM (#41081537)
I would bet actual money that they are not doing an apples-to-apples comparison.

I have seen how those high-performance libraries are made and I'm still in awe at the quality and quantity of the staffing involved. Those were the smartest and most knowledgeable engineers I met in my career.

[ Parent | on HN ]
Re:How to Optimize a CUDA Matmul Kernel for CuBLAS-Like Perf... by ap4 on Saturday July 27, 2024 @12:15PM (#41087477)
It is apples-to-apples. The code from the article runs a chosen kernel 50 times in a loop. In the case of CuBLAS it runs the function cublasGemmEx() 50 times.
[ Parent | on HN ]
Re:How to Optimize a CUDA Matmul Kernel for CuBLAS-Like Perf... by david-gpu on Sunday July 28, 2024 @07:52AM (#41092636)
Does your implementation support the complete feature set of cublasGemmEx()? E.g. varying matrix sizes, etc. Does your implementation beat cublas on a wide range of cases, or only cherry-picked examples? Does the code from the article follow the best coding practices for cublas in the first place?

Generalizing from a micro benchmark is typically hubris.

[ Parent | on HN ]