mat_inverse_kernel.cu

#include "mat_inverse_kernel.h"

#define BLOCK_SIZE 32

__device__ __forceinline__ float mat_inverse (float a){
    return 1.0/(a+1e-8);
}


__global__ void mat_inverse_kernel (const float * __restrict__ src,
                                float * __restrict__ dst, int m, int n){
    int row = blockIdx.y*blockDim.y+threadIdx.y;
    int col = blockIdx.x*blockDim.x+threadIdx.x;

    if (row < m && col < n){
        dst[row * n + col] = mat_inverse(src[row * n + col]);
    }
}

void mat_inverse_kernel_exec(const float *src, float *dst, int m, int n){
    /* specified block and grid size */
    dim3 block(BLOCK_SIZE, BLOCK_SIZE);
    dim3 grid((n+block.x-1)/block.x, (m+block.y-1)/block.y);

    /* lunch kernel */
    mat_inverse_kernel<<<grid, block>>>(src, dst, m, n);
    cudaThreadSynchronize();

}