#include "kernels.h"

__global__ void reorg_kernel(int N, float *x, int w, int h, int c, int batch, int stride, int forward, float *out)
{
    int i = (blockIdx.x + blockIdx.y*gridDim.x) * blockDim.x + threadIdx.x;
    if(i >= N) return;
    int in_index = i;
    int in_w = i%w;
    i = i/w;
    int in_h = i%h;
    i = i/h;
    int in_c = i%c;
    i = i/c;
    int b = i%batch;

    int out_c = c/(stride*stride);

    int c2 = in_c % out_c;
    int offset = in_c / out_c;
    int w2 = in_w*stride + offset % stride;
    int h2 = in_h*stride + offset / stride;
    //printf("%d\n", offset);
    int out_index = w2 + w*stride*(h2 + h*stride*(c2 + out_c*b));

   // printf("%d %d %d\n", w2, h2, c2);
    //printf("%d %d\n", in_index, out_index);
    //if(out_index >= N || out_index < 0) printf("bad bad bad \n");

    if(forward) out[out_index] = x[in_index];
    else out[in_index] = x[out_index];
    //if(forward) out[1] = x[1];
    //else out[0] = x[0];
}

/**
    reorg function function
*/
void reorgForward(dnnType* srcData, dnnType* dstData, 
                  int n, int c, int h, int w, int stride, cudaStream_t stream) {

    int size = n*c*h*w;

    int blocks = (size+255)/256;
    int threads = 256;
    
    reorg_kernel<<<blocks, threads, 0, stream>>>(size, srcData, w, h, c, n, stride, false, dstData);
}