According to my GPU, which is GTX 1660 + WDDM + driver version 510…, I think I should use the Next-gen version. However, when I try to use the Nsight cuda debugger (next-gen) it reports as follow,
Does anyone encounter the same situation? Any reply will be thanksed.
OK, I have solved this question. This error report doesn’t have any relationship with the CUDA debugger. This report is caused by the cudaErrorInvalidConfiguration = 9, which means the thread block size in the code has surpassed the load of local GPU, so what you need to do is to decrease the size of your thread block size. You can use deviceQuery from %CUDA_PATH%/extras/demo_suite to see the necessary information.
By the way, if you do not feel comfortable with the Nsight VS cuda Debugger, you can use Nsight compute, which has the similar function and can generate the run process log.