#include "mat_mul.h"

#include <cstdlib>
#include <cstdio>
#include <pthread.h>

#define BLOCKSIZE 32
#define min(a,b)  (((a) < (b)) ? (a) : (b))

static float *A, *B, *C;
static int M, N, K;
static int num_threads;

static void* mat_mul_thread(void *data) {
  // TODO: parallelize & optimize matrix multiplication
  int num_slice = *((int *)data);

  int slice = M/num_threads;  
  int start = num_slice * slice;
  int end;

  if (num_slice < num_threads - 1) end = (num_slice + 1) * slice; 
  else end = M;

  float Aik;
  int bs = BLOCKSIZE;
  for (int ii = start; ii < end; ii += bs){
    for (int kk = 0; kk < K; kk += bs) {
      //for (int jj = 0; jj < N; jj += bs) {
        for (int i = ii; i < min(ii + bs, end); i++) {
          for (int k = kk; k < min(kk + bs, K); k++) {
            Aik = A[i * K + k];
	    //int size = min(jj + bs, N);
	    int ssize = (N / 8) * 8;
            for (int j = 0; j < ssize; j += 8) {
              C[i * N + j] += Aik * B[k * N + j];
              C[i * N + j+1] += Aik * B[k * N + j+1];
              C[i * N + j+2] += Aik * B[k * N + j+2];
              C[i * N + j+3] += Aik * B[k * N + j+3];
              C[i * N + j+4] += Aik * B[k * N + j+4];
              C[i * N + j+5] += Aik * B[k * N + j+5];
              C[i * N + j+6] += Aik * B[k * N + j+6];
              C[i * N + j+7] += Aik * B[k * N + j+7];
            }
	    for (int j = ssize; j < N; j++){
	      C[i * N + j] += Aik * B[k * N + j];
	    }
          }
        }
      //}
    }
  }

  return NULL;
}

void mat_mul(float *_A, float *_B, float *_C, int _M, int _N, int _K, int _num_threads) {
  A = _A, B = _B, C = _C;
  M = _M, N = _N, K = _K;
  num_threads = _num_threads;

  pthread_t threads[num_threads];
  int thread_args[num_threads];
  int i;


  // TODO: create '_num_threads' pthreads
  for (i=0; i<num_threads; i++){
	thread_args[i] = i;
  	pthread_create(&threads[i], NULL, mat_mul_thread, (void *)&thread_args[i]);
  }

  for (i=0; i<num_threads; i++){
  	pthread_join(threads[i], NULL);
  }
}