代码优化--loop unrolling 更新版

原文地址：http://www.aixchina.net/Article/38453

Loop unwinding, also known as loop unrolling, is a loop transformation technique that attempts to optimize a program's execution speed at the expense of its binary size (space-time tradeoff). The transformation can be undertaken manually by the programmer or by an optimizing compiler.

    The goal of loop unwinding is to increase a program's speed by reducing (or eliminating) instructions that control the loop, such as pointer arithmetic and "end of loop" tests on each iteration;reducing branch penalties; as well as "hiding latencies, in particular, the delay in reading data from memory".

    Loops can be re-written instead as a repeated sequence of similar independent statements eliminating this overhead.

    gcc 优化指令：-funroll-loops

 减少循环比较，提高效率

A simple manual example in C Language

相同功能代码比较

int i=0;

int n=10;

int arr[10]={0};

代码一：

for（i=0;i<n;i++）

{

arr[i]=i;

}

常规代码

代码二：

arr[0]=0;

arr[1]=1;

arr[2]=2;

......

arr[9]=9;

最直接，没有循环条件比较，可以充分利用CPU的并行性(流水线)，效率最高，但是代码冗余度高，损失空间换取时间；

代码三：

for（i=0;i<n;i+=k）

{

arr[i]=i;

arr[i+1]=i+1;

arr[i+2]=i+2;

.......

arr[i+k-1]=i+k-1;

}

缺点：n必须是k的倍数，否则遗落(n%k)个赋值；

代码四：

for（i=0;i<(n%k);i+=(n%k)）

{

arr[i]=i;

arr[i+1]=i+1;

arr[i+2]=i+2;

.......

arr[i+(n%k)-1]=i+(n%k)-1;

}

先对数组首余数个赋值

改进代码：

不需要循环对数组首余数个赋值

if(n%k)

{

arr[i]=i;

arr[i+1]=i+1;

...

arr[i+n%k-1]=i+n%k-1;

}

//

for（i=n%k;i<n;i+=k）

{

arr[i]=i;

arr[i+1]=i+1;

arr[i+2]=i+2;

.......

arr[i+k-1]=i+k-1;

}

通用代码，倍数与否都可以

对于循环下标问题，考虑是否为同一起点，如都是从0或1开始