Unified memory for NPP library

HI.
Before call funtion in NPP library,i need to call “cudamemcpy” to translate host to device.Can i use unified memory by calling “cudaMallocManaged” instead of “cudamemcpy”?Is there any code sample?

Hi,

Yes, you can.
Please find an example blow:

/usr/local/cuda-10.2/samples/7_CUDALibraries/conjugateGradientUM/

Thanks.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.