I am exploring memory access profiling of C/C++/CUDA programs.
On the CPU side, I can check memory access using perf tool. I tried exploring nvprof tool for GPU for the same purpose, but it does not provide me timestamp value of memory accesses. It only provides the total count of memory accesses. Are there any other profiling tools or any known proven techniques to get the memory access profiling of CUDA programs?
Thank you in advance.
Please check if Nsight Systems can meet your requirement:
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.