gettimeofday和clock_gettime的不同

时间：2010-12-01 来源：dufangst

clock_gettime比gettimeofday更加精确
简单做了一下测试先用clock_gettime做一下测试
test.c

#include<time.h>
#include<stdio.h>

#define MILLION 1000000

int main(void)
{
        struct timespec tpstart;
        struct timespec tpend;
        long timedif;

        clock_gettime(CLOCK_MONOTONIC, &tpstart);
        clock_gettime(CLOCK_MONOTONIC, &tpend);
        timedif = MILLION*(tpend.tv_sec-tpstart.tv_sec)+(tpend.tv_nsec-tpstart.tv_nsec)/1000;
        fprintf(stdout, "it took %ld microseconds\n", timedif);

        return 0;
}

在linux 2.6内核下面
gcc -o test test.c -lrt
./test
得到结果：
it took 2 microseconds
现在用gettimeofday来做一下测试

#include<time.h>
#include<stdio.h>

#define MILLION 1000000

int main(void)
{
        struct timespec tpstart;
        struct timespec tpend;
        long timedif;

        gettimeofday(&tpstart, NULL);
        gettimeofday(&tpend, NULL);
        timedif = MILLION*(tpend.tv_sec-tpstart.tv_sec)+(tpend.tv_nsec-tpstart.tv_nsec)/1000;
        fprintf(stdout, "it took %ld microseconds\n", timedif);

        return 0;
}

gcc -o test test.c
./test
得到结果：
it took 0 microseconds AIX下面也得到同样的结果

而且可以通过clock_gettime这个函数测试出64位程序要比32为程序运行快
在AIX P570，16个CPU 15.5G内存机器上测试了一把
用64位模式得到的结果是 0 microseconds
用32位模式得到的结果是 1 microsecond