为什么 C++ 只比 VBA 快 4倍？
用C++和VBA分别写了一段用随机数字测试http://zhihu.com/equation?tex=%5Cpi+的值的算法，用500万个随机数，发现VBA需要570毫秒，而C++则需要150毫秒。这也太让人失望了把，C++写起来那么麻烦，起码也要快个几个数量级吧？python就更别说了，居然要7秒钟。求各位大神帮我看看是不是我的代码没优化好？（平台是Visual Studio，已经开了Release模式）
计算方法：生成http://zhihu.com/equation?tex=a,http://zhihu.com/equation?tex=b两个在0 到 http://zhihu.com/equation?tex=r 之间的随机数，数一数这些数字里面有多少个落在了半径为http://zhihu.com/equation?tex=r的1/4圆的扇形里面，用这个数字代表扇形面积，用总随机数数量代表正方形面积。因为扇形的面积是 http://zhihu.com/equation?tex=+%5Cfrac%7B%5Cpi+r%5E2%7D%7B4%7D+,而正方形的面积是http://zhihu.com/equation?tex=r%5E2, 可得知http://zhihu.com/equation?tex=%5Cpi+=扇形面积/正方形面积 * 4
C++代码：
#include "stdafx.h"#include <iostream>#include <time.h>using namespace std;void main(){ double st = clock(); double rand_max = 32767; srand((int)time(0)); unsigned int simulate_total = 2500000; unsigned int inside_count = 0; unsigned int radius = rand_max * rand_max; unsigned int randA; unsigned int randB; unsigned int randA_opp; unsigned int randB_opp; for (unsigned int i = 1; i < simulate_total; i++){ randA = rand(); randB = rand(); if ((randA * randA + randB * randB) < radius){ inside_count++; } randA_opp = rand_max - randA; randB_opp = rand_max - randB; if ((randA_opp * randA_opp + randB_opp * randB_opp) < radius){ inside_count++; } } cout << inside_count / double(simulate_total) * 2 << endl; cout << clock() - st << endl;}
VBA代码:
Sub simulate_pi()Dim area_count As Double: area_count = 0Dim simulate_count As Double: simulate_count = 2500000Dim i As DoubleDim randA As Double, randB As DoubleDim randA_opp As Double, randB_opp As DoubleFor i = 1 To simulate_count randA = Math.Rnd() randB = Math.Rnd() randA_opp = 1 - randA randB_opp = 1 - randB If randA * randA + randB * randB < 1 Then area_count = area_count + 1 End If If randA_opp * randA_opp + randB_opp * randB_opp < 1 Then area_count = area_count + 1 End IfNext iDebug.Print "Estimate: ", area_count / simulate_count * 2End Sub'VBA内测时间的方法：'新建一个module，把以下代码复制进去，然后运行test.Option ExplicitPrivate Declare Function getFrequency Lib "kernel32" _Alias "QueryPerformanceFrequency" (cyFrequency As Currency) As LongPrivate Declare Function getTickCount Lib "kernel32" _Alias "QueryPerformanceCounter" (cyTickCount As Currency) As Long'Function MicroTimer() As Double'' Returns seconds.' Dim cyTicks1 As Currency Static cyFrequency As Currency ' MicroTimer = 0' Get frequency. If cyFrequency = 0 Then getFrequency cyFrequency' Get ticks. getTickCount cyTicks1' Seconds If cyFrequency Then MicroTimer = cyTicks1 / cyFrequencyEnd FunctionSub test() Dim st: st = MicroTimer Call simulate_pi Debug.Print (MicroTimer - st) * 1000End Sub
Python代码:#!/usr/bin/python# Filename : pi_simulate.pyimport randomimport timetime_start = time.perf_counter()inside_count = 0simulate_total = 2500000randA = 0randB = 0for i in range(1, simulate_total): randA = random.uniform(0,1) randB = random.uniform(0,1) if ((randA * randA + randB * randB) < 1): inside_count = inside_count + 1 randA_opp = 1 - randA randB_opp = 1 - randB if ((randA_opp * randA_opp + randB_opp * randB_opp) < 1): inside_count = inside_count + 1print (inside_count/simulate_total*2)print ("Time spent", time.perf_counter() - time_start)
页:
[1]