扩展图形函数2，循环优化

上一篇的代码已经经过初步的优化了：

unsigned short srcColor, *destColor; for(j = 0; j < pDIBSrc->cy; ++j) { for(i = 0; i < pIDIBSrc->cx; ++i) { srcColor = *(unsigned short *)&pIDIBSrc->pBmp[i + j * pIDIBDest->nPitch]; destColor = (unsigned short *)&pIDIBDest->pBmp[i + j * pIDIBDest->nPitch]; srcColor = ((srcColor) >> 1) & 0x7BEF; *destColor = ((*destColor) >> 1) & 0x7BEF; *destColor = srcColor + (*destColor); } }

 当时项目并不紧，我有足够的时间来打磨和优化

比如这一行

srcColor = *(unsigned short *)&pIDIBSrc->pBmp[i + j * pIDIBDest->nPitch];

着一个运算

j * pIDIBDest->nPitch

会执行pIDIBSrc->cx * pIDIBSrc->cy次，而且乘法在没有协处理器的功能机上，效率非常低

 优化方式是加一个步进，将这一个乘法的cx * cy的复杂度变为cx复杂度

unsigned short srcColor, *destColor; int step = 0; for(j = 0; j < pDIBSrc->cy; ++j) { for(i = 0; i < pIDIBSrc->cx; ++i) { srcColor = *(unsigned short *)&pIDIBSrc->pBmp[i + step]; destColor = (unsigned short *)&pIDIBDest->pBmp[i + step]; srcColor = ((srcColor) >> 1) & 0x7BEF; *destColor = ((*destColor) >> 1) & 0x7BEF; *destColor = srcColor + (*destColor); } step += pDIBDest->nPitch; }

 又更进了一步：）