.Net 4比.Net 3.5慢吗? [英] Is .Net 4 slower than .Net 3.5?

问题描述

大家好，

我正在使用VS 2008和.Net 3.5测试某些计算的运行时.我以某种方式对结果感到不满意，我以为我启动了VS 2010 Express并尝试使用Parallel Extensions运气.我要做的第一件事是将代码复制到新的VS 2010项目中，然后重新运行测试.令我惊讶的是，它比.Net 3.5(66秒，38秒)要慢得多.我使用调试器外部的发布版本运行了这两个测试.这真让我感到困惑，因为我不敢相信性能下降会如此之大.

这是我的测试代码:

Hi all,

I was testing the runtime of some calculations with VS 2008 and .Net 3.5. Somehow unsatisfied with the result I thought I startup VS 2010 Express and try my luck with the Parallel Extensions. The first thing I''ve done was to copy the code to a new VS 2010 project and rerun my test. To my surprise it was by far slower than in .Net 3.5 (66 seconds, 38 seconds). I ran both tests with release builds outside the debugger. This really puzzles me as I somehow can''t believe the performance drop is so great.

Here is my test code:

using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.Linq;
using System.Text;

namespace SaTest
{
    class Program
    {
        struct Structure
        {
            public float X;
            public float Y;
        }


        static void Main(string[] args)
        {
            var l = new Structure[2000000];
            var r = new Random();
            for (int i = 0; i < l.Length; i++)
            {
                l[i].X = (float)(r.NextDouble() * 1000000);
                l[i].Y = (float)(r.NextDouble() * 1000000);
            }

            int counter = 0;
            double dist = 2000;
            double distSquare = dist * dist;

            var sw = new Stopwatch();
            sw.Start();
            for (int i = 0; i < l.Length; i++)
            {
                for (int j = 0; j < 2000; j++)
                {
                    double tx = l[i].X - l[j].X;
                    double ty = l[i].Y - l[j].Y;
                    if (tx * tx + ty * ty <= distSquare)
                        //if (Math.Sqrt(tx * tx + ty * ty) <= dist)
                        counter++;
                }
            }
            sw.Stop();
            Console.WriteLine(counter);
            Console.WriteLine(sw.Elapsed);
            Console.ReadLine();
        }
    }
}

有没有人有过类似的经历?还是我的代码有些奇怪?如果有人可以测试一下并告诉我我的结果是否可重现，我将不胜感激.

Has anyone made similar experiences? Or is something weird wrong with my code? I would appreciate if anyone could test this and tell me if my results are reproducable.

推荐答案

在发布模式下使用.Net 2.0、3.5或4.0进行编译时，经过优化和没有溢出检查，我得到的计时是相同的.在所有情况下都为20.76-20.77秒.

XP SP3，Intel Core 2 Duo T7250、3GB RAM.

艾伦.

P.S.我删除了vars，以获取要为.NET 2编译的代码，并将修改后的源代码也用于3.5和4.

Hi,
I get identical timings when compiled with .Net 2.0, 3.5 or 4.0 in release mode with optimisations and no overflow checking. 20.76 - 20.77 seconds in all cases.

XP SP3, Intel Core 2 Duo T7250, 3GB RAM.

Alan.

P.S. I removed the vars to get the code to compile for .NET 2 and used the modified source for 3.5 and 4 as well.

这篇关于.Net 4比.Net 3.5慢吗?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！