Hello! I am a undergrad student currently working on a project. I am trying to implement NVTX, and I am running into an error when I attempt to add the NVTX code. I have successfully implemented NVTX with CPU code and I can see it within Nsight Systems, but when I try to implement in the GPU section I receive the error “too many arguments in function call”. Has anyone seen this before? Thank you.
Hi,
NVTX is a CPU side API. You mark the ranges in your CPU code and then tools (like Nsight Systems) do the projection to show what all CUDA kernels were inside of those ranges.