开方运算的消耗

最新推荐文章于 2023-08-11 16:17:17 发布

xiaolomg

最新推荐文章于 2023-08-11 16:17:17 发布

阅读量1.6k

点赞数

分类专栏： Work_Note

本文链接：https://blog.csdn.net/boyxiaolong/article/details/24293221

版权

Work_Note 专栏收录该内容

58 篇文章 0 订阅

订阅专栏

看到服务器做碰撞检测居然用了开放运算，很显然开方是牛顿迭代法是很耗CPU的，不过我来测试下：

#include <iostream>
#include <string>
#include <locale>
#include <random>
#include <set>
#include <stdio.h>
#include <windows.h>
#include <time.h> //time_t time()  clock_t clock()
#include <Mmsystem.h>             //timeGetTime()
#pragma comment(lib, "Winmm.lib")   //timeGetTime()
using namespace std;

static std::default_random_engine  m_Generator;

void SqrtTest(int n)
{
	printf("%s\n", __FUNCTION__);
	std::uniform_int_distribution<float> distribution(0, n);
	
	DWORD  dwGTCBegin = GetTickCount();

	for (int i = 0; i < n; ++i)
	{
		float a = distribution(m_Generator);
		float b = distribution(m_Generator);
		float c = distribution(m_Generator);
		float d = distribution(m_Generator);

		float e = sqrt(a*a + b*b+ c*c);
		if (e > d){
			//printf(" ");
		}		
	}

	printf("%d\n", GetTickCount() - dwGTCBegin);  
}

void SqureTest(int n)
{
	printf("%s\n", __FUNCTION__);
	std::uniform_int_distribution<float> distribution(0, n);
	
	DWORD  dwGTCBegin = GetTickCount();

	for (int i = 0; i < n; ++i)
	{
		float a = distribution(m_Generator);
		float b = distribution(m_Generator);
		float c = distribution(m_Generator);
		float d = distribution(m_Generator);

		float e = a*a + b*b+ c*c;
		if (e > d * d){
			//printf(" ");
		}		
	}

	printf("%d\n", GetTickCount() - dwGTCBegin);  
}

int main(void) {
	int n = 10000000;

	SqrtTest(n);
	SqureTest(n);
}

千万次的计算相差也不是很大，但毕竟这不是实际情况：