Nsight graphics tracing memory failure?

I am using latest nsight graphics (2022.3.1) and 3050PC. And I am profiling a kernel compiled from cublas.(which should be correct!) But I received this:

Thank you!!!

Thank you for using Nsight for your profiling needs. Were you using Nsight Compute or Nsight Graphics? Since you were profiling a Cuda kernel I’m wondering if you were using our Nsight Compute product which is targeted specifically for Cuda.

1 Like

I am actually using graphics…
But you mentioned this is not for CUDA??
I write a CUDA kernel and nvcc compile it into a.exe in my 3050PCwin10, and then put it here in graphics…
What should be the correct way?

Please see the overview flowchart here of how to use our Nsight Tools, NVIDIA Developer Tools Overview | NVIDIA Developer.

All developers should start with Nsight Systems to identify the largest optimization opportunities. Nsight Systems provides developers a system-wide visualization of an applications performance. Developers can optimize bottlenecks to scale efficiently across any number or size of CPUs and GPUs; from large servers to our smallest SoC. For further optimizations to compute kernels developers should use Nsight Compute or to further optimize a graphics workloads, use Nsight Graphics.

Oh…Thank you. So I go to document and see it just now…Does not find any problem…Still not sure why my measuring fail…

Would you tell me? Thank you!!!

Is that only me? I guess many others will also meet similar question?