并行LINQ
.NET4在System.Linq命名空间中包含一个新类ParallelEnumerable,可以分解查询的工作使其分布在多个线程上.尽管Enumerable类给IEnumerable<T>接口定义了扩展方法,但ParallelEnumerable类的大多数扩展方法是ParallelQuery<TSource>类的扩展.一个重要的例外是AsParallel()方法,它扩展了IEnumerable<TSource>接口,返回ParallelQuery<TSource>类,所以正常的集合类可以以平行方式查询.
LINQ比较强大的是还提供了可并行处理的查询,这使得我们可以借助它来完成一些查询处理或处理并行操作.
先来说一下并行集合,并行计算使用的多个线程同时进行计算,所以要控制每个线程对资源的访问,我们先来看一下常用的List<T>集合,在并行计算下的表示:
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
namespace 并行集合和LINQ
{
class Program
{
static void Main(string[] args)
{
List<int> list = new List<int>();
Parallel.For(0, 10000, item =>
{
list.Add(item);
}
);
Console.WriteLine("list的长度为 : {0}",list.Count());
//多测试几次,就是说多运行几次
/*
* 从结构可以看出,我们的结构不是10000.这是为什么呢,这是因为List是费线程安全的
* ,也就是说,任何线程都可以修改他的值
*/
Console.ReadKey();
}
}
}
接下来我们看一下并行集合----线程安全集合,在System.Collections.Concurrent命名空间中,首先看一下ConcurrentBag<T>泛型集合,其用法和List<T>类似:
ConcurrentBag<int> list =new ConcurrentBag<int>();
Parallel.For(0, 10000, item =>
{
list.Add(item);
}
);
Console.WriteLine("ConcurrentBag的长度为 : {0}", list.Count());
//不管运行几次,结果都是10000
分析一下,因为ConcurrentBag是线程安全的,所以每次结果都是正确的.
下面我们来修改代码看看ConcurrentBag里面的数据到底是怎样存放的,修改代码如下:
using System;
using System.Collections.Concurrent;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
namespace 并行集合和LINQ
{
class Program
{
static void Main(string[] args)
{
ConcurrentBag<int> list =new ConcurrentBag<int>();
Parallel.For(0, 10000, item =>
{
list.Add(item);
}
);
Console.WriteLine("ConcurrentBag的长度为 : {0}", list.Count());
//不管运行几次,结果都是10000
int n = 0;
foreach (var item in list)
{
if (n>10)
{
break;
}
n++;
Console.WriteLine("Item[{0}] = {1}",n,item);//,多运行几次,观察这里的输出
}
Console.WriteLine("ConcurrentBag的最大长度为 : {0}",list.Max());
Console.ReadKey();
}
}
}
分析:可以看到ConcurrentBag中的数据并不是按照顺序排列的,顺序是乱的,随机的.我们平时使用的Max,First,Last等LINQ方法都还有.其十分类似Enumerable的用法.
关于线程安全的集合还有很多，和我们平时用的集合都差不多，比如类似Dictionary的ConcurrentDictionary，还有ConcurrentStack，ConcurrentQueue等。
并行LINQ的用法和性能
1.AsParallel
案例:
using System;
using System.Collections.Concurrent;
using System.Collections.Generic;
using System.Diagnostics;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
namespace 并行集合和LINQ
{
class Program
{
static void Main(string[] args)
{
Stopwatch sw = new Stopwatch();//需要添加命名空间 System.Threading.Tasks;
List<Custom> customs = new List<Custom>();
for (int i = 0; i < 2000000; i++)
{
customs.Add(new Custom() { Name = "Jack", Age = 21, Address = "NewYork" });
customs.Add(new Custom() { Name = "Jime", Age = 26, Address = "China" });
customs.Add(new Custom() { Name = "Tina", Age = 29, Address = "ShangHai" });
customs.Add(new Custom() { Name = "Luo", Age = 30, Address = "Beijing" });
customs.Add(new Custom() { Name = "Wang", Age = 60, Address = "Guangdong" });
customs.Add(new Custom() { Name = "Feng", Age = 25, Address = "YunNan" });
}
sw.Start();
var result = customs.Where<Custom>(c => c.Age > 26).ToList();
sw.Stop();
Console.WriteLine("Linq time is {0}.", sw.ElapsedMilliseconds);
sw.Restart();
sw.Start();
var result2 = customs.AsParallel().Where<Custom>(c => c.Age > 26).ToList();
sw.Stop();
Console.WriteLine("Parallel Linq time is {0}.", sw.ElapsedMilliseconds);
Console.WriteLine("运行完毕");//这句话可能需要过一会才能出现
Console.ReadKey();
}
}
public class Custom
{
public string Name { get; set; }
public int Age { get; set; }
public string Address { get; set; }
}
}
分析:多运行即便可以发现,添加了AsParallel()方法的速度差不多快了一倍.其实，AsParallel()这个方法可以应用与任何集合，包括List<T>集合，从而提高查询速度和系统性能。
2.GroupBy方法
在项目中,我们经常要对数据做处理,比如分组统计,我们知道在LINQ中也可以实现,今天来学习以下新的ToLookUp方法,写一个测试方法:其他代码相似,只是测试代码不同
Stopwatch stopWatch = new Stopwatch();
List<Custom> customs = new List<Custom>();
for (int i = 0; i < 2000000; i++)
{
customs.Add(new Custom() { Name = "Jack", Age = 21, Address = "NewYork" });
customs.Add(new Custom() { Name = "Jime", Age = 26, Address = "China" });
customs.Add(new Custom() { Name = "Tina", Age = 29, Address = "ShangHai" });
customs.Add(new Custom() { Name = "Luo", Age = 30, Address = "Beijing" });
customs.Add(new Custom() { Name = "Wang", Age = 60, Address = "Guangdong" });
customs.Add(new Custom() { Name = "Feng", Age = 25, Address = "YunNan" });
}
stopWatch.Restart();
var groupByAge = customs.GroupBy(item => item.Age).ToList();
foreach (var item in groupByAge)
{
Console.WriteLine("Age={0},count = {1}", item.Key, item.Count());
}
stopWatch.Stop();
Console.WriteLine("Linq group by time is: " + stopWatch.ElapsedMilliseconds);
stopWatch.Restart();
var lookupList = customs.ToLookup(i => i.Age);
foreach (var item in lookupList)
{
Console.WriteLine("LookUP:Age={0},count = {1}", item.Key, item.Count());
}
stopWatch.Stop();
Console.WriteLine("LookUp group by time is: " + stopWatch.ElapsedMilliseconds);
Console.WriteLine("运行完毕");//这句话可能需要过一会才能出现
Console.ReadKey();
ToLookup方法是将集合转换成一个只读集合，所以在大数据量分组时性能优于List.