I still try to study about cusolverMG library.
My workstation has 4 K80C cards. nvidia-smi command tells me that 11441MB memory is available for each GPU core. I’ve tried the example code in https://docs.nvidia.com/cuda/cusolver/index.html#mgsyevd-example1. I’ve watched the memory usage while running the code and found that core 0 uses 907MB and the other 7 cores use 112MB memory after run cusolverMgDeviceSelect routine. I didn’t allocate any memory at the moment. My question is as follows:
What does the cusolverMgDeviceSelect do internally?
Why does it allocate memory inhomogeneously?