Hello. I would like to have my function output directly to LabView IMAQ image buffers, which are in page locked memory. I have confirmed this by doing
cudaMemcpyAsync() to and from the buffers and reading back the results without problem.
But cudaHostGetDevicePointer() errors with “invalid parameter”. My conclusion is that cudaHostAlloc sets up a mapping on the GPU that isn’t done with other page locked memory. How can I set up such a mapping?
Please Tim, tell me this can be done.