并行性:cpu处理指令是并行的,前一条指令处理时,紧接着前一条的指令也会开始处理,如果后一条指令和前一条指令无关,那么程序性能会得到提升,我们常常会把一个操作分解成几条指令并行来提升性能。
实例代码:
long long getSystemTime() {
struct timeb t;
ftime(&t);
return 1000 * t.time + t.millitm;
}
int main()
{
int * b_test_data = (int *)malloc(80000000*sizeof(int));
int b_loop;
long long b_begin, b_end, b_ttl1 = 1, b_ttl2 = 1;
for(b_loop = 0; b_loop < 80000000; b_loop++)
{
b_test_data[b_loop] = 1;
}
b_begin = getSystemTime();
for(b_loop = 0; b_loop < 80000000 - 1; b_loop += 2)
{
b_ttl1 = b_ttl1*b_test_data[b_loop]*b_test_data[b_loop + 1];
}
for(; b_loop < 80000000; b_loop++)
{
b_ttl1 = b_ttl1*b_test_data[b_loop];
}
b_end = getSystemTime();
printf("time = %lu, b_ttl = %lu\n", b_end - b_begin, b_ttl1);
return 0;
}
运行结果:
并行:
int main()
{
int * b_test_data = (int *)malloc(80000000*sizeof(int));
int b_loop;
long long b_begin, b_end, b_ttl1 = 1, b_ttl2 = 1;
for(b_loop = 0; b_loop < 80000000; b_loop++)
{
b_test_data[b_loop] = 1;
}
b_begin = getSystemTime();
for(b_loop = 0; b_loop < 80000000 - 1; b_loop += 2)
{
b_ttl1 = b_ttl1*b_test_data[b_loop];
b_ttl2 = b_ttl2*b_test_data[b_loop + 1];
}
b_ttl1 *= b_ttl2;
for(; b_loop < 80000000; b_loop++)
{
b_ttl1 = b_ttl1*b_test_data[b_loop];
}
b_end = getSystemTime();
printf("time = %lu, b_ttl = %lu\n", b_end - b_begin, b_ttl1);
return 0;
}
运行结果: