string.h的strcmp的性能比较

string.h基于汇编实现strcmp ,和普通strcmp，针对循环调用次数和字符串查找长度2个纬度做了一次性能对比效测试。

#include <sys/socket.h>
#include <sys/epoll.h>
#include <netinet/in.h>
#include <arpa/inet.h>
#include <fcntl.h>
#include <unistd.h>
#include <stdio.h>
#include <errno.h>
#include <iostream>
#include <strings.h>
#include <string.h>
#include <stdlib.h>
#include <sys/time.h>

using namespace std;


static inline int strcmp_asm(const char * cs,const char * ct)
{
    int d0, d1;
    register int __res;
    __asm__ __volatile__(
            "1:\tlodsb\n\t"
            "scasb\n\t"
            "jne 2f\n\t"
            "testb %%al,%%al\n\t"
            "jne 1b\n\t"
            "xorl %%eax,%%eax\n\t"
            "jmp 3f\n"
            "2:\tsbbl %%eax,%%eax\n\t"
            "orb $1,%%al\n"
            "3:"
            :"=a" (__res), "=&S" (d0), "=&D" (d1)
            :"1" (cs),"2" (ct));
    return __res;
}


static inline int strcmp_normal(const char * cs,const char * ct)
{
    while(*cs!='\0' && *ct!='\0')
    {
        if(*cs != *ct)
            return *cs<*ct?-1:1;
        cs++;
        ct++;
    }
    if(*cs != *ct)
        return *cs<*ct?-1:1;
    return 0;
}

static inline int generator_data(string &s,int iSamelen=1,const char *sepc=NULL)
{
    s.clear();
    for(int i=0;i<iSamelen;++i)
        s+="1";
    s+=sepc;
    return 0;
}

static inline void strcmp_perfom(int loop=1,int iSamelen=1)
{
    //loop asm
    string s1,s2;
    generator_data(s1,iSamelen,"s1");
    generator_data(s2,iSamelen,"s2");


    struct timeval start_tv,loop1_tv;
    gettimeofday(&start_tv, NULL);

    for(int i=0;i<loop;++i)
    {
        strcmp_asm(s1.c_str(),s2.c_str());
    }
    gettimeofday(&loop1_tv, NULL);
    int iMs = (loop1_tv.tv_sec - start_tv.tv_sec)*1000 + (loop1_tv.tv_usec - start_tv.tv_usec)/1000;

    //loop normal
    string ss1,ss2;
    generator_data(ss1,iSamelen,"s1");
    generator_data(ss2,iSamelen,"s2");

    gettimeofday(&start_tv, NULL);
    for(int i=0;i<loop;++i)
    {
        strcmp_normal(ss1.c_str(),ss2.c_str());
    }
    gettimeofday(&loop1_tv, NULL);
    int iMs2 = (loop1_tv.tv_sec - start_tv.tv_sec)*1000 + (loop1_tv.tv_usec - start_tv.tv_usec)/1000;

    printf("loop:%d,strcmp_asm,time:%d ms,strcmp_normal,time:%d ms\n",loop,iMs,iMs2);
}

int main(int argc,char** argv)
{
    int loop=10000;
    int iSamelen=1000;
    if(argc>1)
    {
        loop = atoll(argv[1]);
    }
    if(argc>2)
    {
        iSamelen = atoll(argv[2]);
    }
    printf("input loop:%d,iSamelen:%d\n",loop,iSamelen);

    strcmp_perfom(loop,iSamelen);

    return 0;
}

测试结果如下，在匹配串长度足够的情况下，性能差了3~4倍。
input loop:5000,iSamelen:100000
loop:5000,strcmp_asm,time:468 ms,strcmp_normal,time:1747 ms


input loop:50000,iSamelen:10000
loop:50000,strcmp_asm,time:501 ms,strcmp_normal,time:1749 ms


input loop:500000,iSamelen:1000
loop:500000,strcmp_asm,time:494 ms,strcmp_normal,time:1787 ms


input loop:5000000,iSamelen:100
loop:5000000,strcmp_asm,time:693 ms,strcmp_normal,time:1922 ms


input loop:50000000,iSamelen:10
loop:50000000,strcmp_asm,time:1337 ms,strcmp_normal,time:2424 ms


在匹配串极短的场景，1，2，4字节的情况，性能也有1.5~2倍左右的差距
input loop:50000000,iSamelen:4
loop:50000000,strcmp_asm,time:849 ms,strcmp_normal,time:1425 ms


input loop:50000000,iSamelen:2
loop:50000000,strcmp_asm,time:753 ms,strcmp_normal,time:1130 ms


input loop:50000000,iSamelen:1
loop:50000000,strcmp_asm,time:670 ms,strcmp_normal,time:1040 ms