Currently, I can’t get cuda kernel profiling for RTX30 graphics card on WSL2 based on nsight system, but it’s possible for RTX 20 graphics card. I’s very inconvenient to debug kernel performance.
Will you support cuda kernel profiling on WSL2 for RTX30 graphics card?
Currently Nsight Systems does not support WSL2 (across all GPU cards including RTX 20 and RTX 30). It will be supported in a future release.
Note that Nsight Compute supports cuda kernel profiling. But WSL2 is still not supported.