Question for "CUDA"

Hi,

I installed CUDA 10.1, my visual studio is 2015 version.
graphic card is GT650.

after I opened CUDA 10.1 and checked value of cudaStatus,
an error has occured as cudaErrorNoKernelImageForDevice(209).
this problem is as below.

// Check for any errors launching the kernel
cudaStatus = cudaGetLastError();
if (cudaStatus != cudaSuccess) {
    fprintf(stderr, "addKernel launch failed: %s\n", cudaGetErrorString(cudaStatus));
    goto Error;
}

I don’t know why it is. pls check it.
thanks in advance.

Hi, that error means that you haven’t compiled eith support for sm_53 (Nano GPU arch). Please add these flags to your nvcc compile options:

-gencode arch=compute_53,code=sm_53 \
             -gencode arch=compute_53,code=compute_53

thanks for your kindly reply,

however, I still don’t know where should I add “nvcc compile options”.

Visual Studio 2015\Projects\CudaSumArrayDLL

as I know, the path route is project path.

.cu type file is only “kernel.cu” in this path, do you means that should I change it?

Hi,

May I know more detail about your environment first?
Do you want to cross-compile an application for Jetson Nano?

I don’t think visual studio supports Jetson environment.
Thanks.