ahennequ

#include <stdio.h>

// Check tensor core's warp register layout
// nvcc -arch=sm_75 tensorcore_mapping.cu -o mapping
// ./mapping

// Define some error checking macros.
#define cudaErrCheck(stat) { cudaErrCheck_((stat), __FILE__, __LINE__); }
void cudaErrCheck_(cudaError_t stat, const char *file, int line) {
   if (stat != cudaSuccess) {