新手上路，想请教大神，下面这段程序怎么优化。如果要用到汇编的话，希望大神能说详细点哈，我不太懂汇编


#include<stdio.h>
#include<stdlib.h>

#define BLOCK_SIZE             4

int a[16][16] ;

void forward4x4(int (*block)[16], int (*tblock)[16], int pos_y, int pos_x)
{
  int i, ii;  
  int tmp[16];
  int *pTmp = tmp, *pblock;
  int p0,p1,p2,p3;
  int t0,t1,t2,t3;

  // Horizontal
  for (i=pos_y; i < pos_y + BLOCK_SIZE; i++)
  {
    pblock = &block[i][pos_x];
    p0 = *(pblock++);
    p1 = *(pblock++);
    p2 = *(pblock++);
    p3 = *(pblock  );

    t0 = p0 + p3;
    t1 = p1 + p2;
    t2 = p1 - p2;
    t3 = p0 - p3;

    *(pTmp++) =  t0 + t1;
    *(pTmp++) = (t3 << 1) + t2;
    *(pTmp++) =  t0 - t1;   
    *(pTmp++) =  t3 - (t2 << 1);
  }
   
  // Vertical
   for (i=0; i < BLOCK_SIZE; i++)
  {
    pTmp = tmp + i;
    p0 = *pTmp;
    p1 = *(pTmp += BLOCK_SIZE);
    p2 = *(pTmp += BLOCK_SIZE);
    p3 = *(pTmp += BLOCK_SIZE);

    t0 = p0 + p3;
    t1 = p1 + p2;
    t2 = p1 - p2;
    t3 = p0 - p3;

    ii = pos_x + i;
    tblock[pos_y    ][ii] = t0 +  t1;
    tblock[pos_y + 1][ii] = t2 + (t3 << 1);
    tblock[pos_y + 2][ii] = t0 -  t1;
    tblock[pos_y + 3][ii] = t3 - (t2 << 1);
  }
}

int main(){
   
    int (*b)[16] = a;
    int j,k;

    //input
    freopen("D:/BISHE/input.txt","r",stdin);
    for(j=0;j<16;j++){
        for(k=0;k<16;k++){
            scanf("%d ",&b[j][k]);   
        }
    }
    freopen("CON","r",stdin);
   
     for (j = 0; j < 16; j+=4)
    {
        for (k = 0;k < 16; k+=4)
        {
            forward4x4( b,  b, j, k);
        }
    }
   
    //output
    freopen("D:/BISHE/output1.txt","w",stdout);
    for (j = 0; j < 16; j++)
    {
        for (k = 0;k < 16; k++)
        {
            printf("%d ",a[j][k]);
        }
        printf("\n");
    }
    freopen("CON","w",stdout);
   
    system("pause");
    return 0;   
}

这是一个离散余弦变换算法，我说的优化是想能不能改写的让这段代码运行的时间更短一点，计算的快一点

嗯嗯，谢谢