【C# Parallel】开端
使用条件
1、必须熟练掌握锁、死锁、task的知识,他是建立这两个的基础上的。task建立在线程和线程池上的。
2、并不是所有代码都适合并行化。 例如,如果某个循环在每次迭代时只执行少量工作,或它在很多次迭代时都不运行,那么并行化的开销可能导致代码运行更慢。
3、 Parallel.For()和Paraller.ForEach()方法在每次迭代中调用相同的代码,都有返回值ParallelLoopResult
。而Parallel.Invoke()方法允许同时调用不同的方法,返回值viod。
4、 ForEach的格式【Item、否是带有并行选项、【Item 、否是有循环状态、否是有索引 、否是有线程本地变量 】】
For的格式【起始索(引包含),结束索引(不包含),是否有并行选项【起始索引,释放有循环状态,否是有线程本地变量】】
总结
1、C#通过Parallel
类提供并行任务支持,可以很简单的使用线程池。
2、要注意的是,Parallel
类不保障执行顺序。
相关的类:
Parallel
ParallelLoopResult
ParallelLoopState
ParallelOptions
For的格式【起始索(引包含),结束索引(不包含),释放有并行选项【起始索引,释放有循环状态,是否线程本地变量】】
用法一、For(Int32, Int32, Action<Int32>)
int[] count = Enumerable.Range(1, 100).ToArray(); Parallel.For(1, count.Length, indexer => { Console.WriteLine(indexer); });
用法二、For(Int32, Int32, Action<Int32,ParallelLoopState>)
ParallelOptions op = new() { CancellationToken = CancellationToken.None }; Parallel.For(1, 2, (indexer, op) => { Console.WriteLine(indexer); });
用法三、For(Int32, Int32, ParallelOptions, Action<Int32,ParallelLoopState>)
ParallelOptions op = new() { CancellationToken = CancellationToken.None }; op.TaskScheduler = TaskScheduler.Default; op.MaxDegreeOfParallelism = 1; Parallel.For(1, 2, op, indexer => { Console.WriteLine(indexer); });
用法四、
编写具有线程局部变量的 Parallel.For 循环
public static ParallelLoopResult For<TLocal>
( int fromInclusive, //范围下线 包含 int toExclusive, //范围上线 不包含 Func<TLocal> localInit,//TLocal表示线程本地变量的类型 Func<int, ParallelLoopState, TLocal, TLocal> body,//TLocal表示线程本地变量的类型 Action<TLocal> localFinally )
案例:
//TLocal 线程本地 变量的类型 int totalSum = 0; Parallel.For<string>(1, 10, ()=> "开始打印",// "开始打印" 被保存到线程本地变量中 (indexer,loopState, LocalValue) => { Console.WriteLine(LocalValue +" Thread "+ indexer ); return LocalValue += indexer; // LocalValue += indexer被保存到线程本地变量中 }, LocalValue => Console.WriteLine("结束打印") ); /* 输出 开始打印 Thread 1 开始打印 Thread 2 开始打印2 Thread 8 return LocalValue += indexer;被存储到本地线程变量中LocalValue, 开始打印28 Thread 9 所以打印出 开始打印28 开始打印 Thread 6 开始打印 Thread 5 结束打印 开始打印1 Thread 7 结束打印 开始打印 Thread 3 结束打印 结束打印 开始打印 Thread 4 结束打印 结束打印*/
ForEach的格式【Item、否是带有并行选项、【Item 、否是有循环状态、否是有索引 、否是有线程本地变量 】】
Enumerable<TSource>和 Partitioner<TSource> 都是以下格式
ForEach<TSource>(IEnumerable<TSource>, ParallelOptions, Action<TSource>)
ForEach<TSource>(IEnumerable<TSource>, ParallelOptions, Action<TSource,ParallelLoopState>)
ForEach<TSource>(IEnumerable<TSource>, Action<TSource>)
ForEach<TSource>(IEnumerable<TSource>, Action<TSource,ParallelLoopState>)
ForEach<TSource>(IEnumerable<TSource>, Action<TSource,ParallelLoopState,Int64>)
ForEach<TSource>(IEnumerable<TSource>, ParallelOptions, Action<TSource,ParallelLoopState,Int64>)
ForEach<TSource,TLocal>(IEnumerable<TSource>, Func<TLocal>, Func<TSource,ParallelLoopState,TLocal,TLocal>, Action<TLocal>)
ForEach<TSource,TLocal>(IEnumerable<TSource>,
Func<TLocal>,
Func<TSource,ParallelLoopState,Int64,TLocal,TLocal>,
Action<TLocal>)
ForEach<TSource,TLocal>(IEnumerable<TSource>,
ParallelOptions, Func<TLocal>,
Func<TSource,ParallelLoopState,TLocal,TLocal>,
Action<TLocal>)
ForEach<TSource,TLocal>(IEnumerable<TSource>,
ParallelOptions, Func<TLocal>,
Func<TSource,ParallelLoopState,Int64,TLocal,TLocal>,
Action<TLocal>)
用法一:Item
ForEach<TSource>(IEnumerable<TSource>, Action<TSource>)案例如下:
string[] data = { "zero", "one", "two", "three", "four", "five", "six", "seven", "eight", "nine", "ten" }; Parallel.ForEach(data, item => { Console.WriteLine(item); });
用法二:Item 、循环状态、带索引
ForEach<TSource>(IEnumerable<TSource>, Action<TSource,ParallelLoopState,Int64(索引)>)案例如下:
string[] data = { "zero", "one", "two", "three", "four", "five", "six", "seven", "eight", "nine", "ten" }; Parallel.ForEach(data, (item,loopstate,indexer )=> { if(item == "three") { loopstate.Break(); Console.WriteLine($"item:{item} index:{indexer} ShouldExitCurrentIteration:{loopstate.ShouldExitCurrentIteration} LowestBreakIteration(索引):{loopstate.LowestBreakIteration}"); } Console.WriteLine($"item:{item} index:{indexer}" ); });
用法三:Item、带有并行选项、循环状态、索引
ForEach<TSource>(IEnumerable<TSource>, ParallelOptions, Action<TSource,ParallelLoopState,Int64>)案例如下:
string[] data = { "zero", "one", "two", "three", "four", "five", "six", "seven", "eight", "nine", "ten" }; ParallelOptions op = new(); op.TaskScheduler = TaskScheduler.Default;//默认的任务调度器 op.MaxDegreeOfParallelism = 5;//并行度 op.CancellationToken=CancellationToken.None;//禁止取消 Parallel.ForEach(data, op, (item,loopstate,indexer )=> { if(item == "zero") { loopstate.Break(); Console.WriteLine($"item:{item} index:{indexer} ShouldExitCurrentIteration:{loopstate.ShouldExitCurrentIteration} LowestBreakIteration:{loopstate.LowestBreakIteration}"); } Console.WriteLine($"item:{item} index:{indexer}" ); });
用法四:Item、带有并行选项、循环状态、 线程本地变量
ForEach<TSource,TLocal>(IEnumerable<TSource>, ParallelOptions, Func<TLocal>, Func<TSource,ParallelLoopState,TLocal,TLocal>, Action<TLocal>)案例如下:
string[] data = { "zero", "one", "two", "three", "four", "five", "six", "seven", "eight", "nine", "ten" }; ParallelOptions op = new(); op.TaskScheduler = TaskScheduler.Default;//默认的任务调度器 op.MaxDegreeOfParallelism = 5;//并行度 op.CancellationToken=CancellationToken.None;//禁止取消 //自行推断TSource,TLocal Parallel.ForEach(data, op, ()=>"开始输出:",//初始化线程本地变量 (item,loopstate, TLocal) => { if(item == "zero") { loopstate.Break(); Console.WriteLine($"item:{item} index:{TLocal} ShouldExitCurrentIteration:{loopstate.ShouldExitCurrentIteration} LowestBreakIteration:{loopstate.LowestBreakIteration}"); return TLocal+ "=>Break"; } Console.WriteLine($"item:{item} index:{TLocal}" ); return TLocal + "=>complete"; }, local => Console.WriteLine(local) //输出线程本地变量 );
用法五:Item、带有并行选项、循环状态、索引、 线程本地变量
ForEach<TSource,TLocal>(IEnumerable<TSource>, ParallelOptions, Func<TLocal>, Func<TSource,ParallelLoopState,Int64,TLocal,TLocal>, Action<TLocal>)
string[] data = { "zero", "one", "two", "three", "four", "five", "six", "seven", "eight", "nine", "ten" }; ParallelOptions op = new(); op.TaskScheduler = TaskScheduler.Default;//默认的任务调度器 op.MaxDegreeOfParallelism = 5;//并行度 op.CancellationToken=CancellationToken.None;//禁止取消 //自行推断TSource,TLocal Parallel.ForEach(data, op, ()=>"开始输出:",//初始化线程本地变量 (item,loopstate, index, TLocal) => { if(item == "zero") { loopstate.Break(); Console.WriteLine($"item:{item} index:{TLocal} ShouldExitCurrentIteration:{loopstate.ShouldExitCurrentIteration} LowestBreakIteration:{loopstate.LowestBreakIteration}"); return TLocal+ "=>Break"; } Console.WriteLine($"item:{item} index:{TLocal}" ); return TLocal + "=>complete"; }, local => Console.WriteLine(local) //输出线程本地变量 );
用法六:Item 、循环状态、 线程本地变量
ForEach<TSource,TLocal>(IEnumerable<TSource>, Func<TLocal>, Func<TSource,ParallelLoopState,TLocal,TLocal>, Action<TLocal>)案例如下:
string[] data = { "zero", "one", "two", "three", "four", "five", "six", "seven", "eight", "nine", "ten" }; //自行推断TSource,TLocal Parallel.ForEach(data, ()=>"开始输出:",//初始化线程本地变量 (item,loopstate, TLocal) => { if(item == "zero") { loopstate.Break(); Console.WriteLine($"item:{item} index:{TLocal} ShouldExitCurrentIteration:{loopstate.ShouldExitCurrentIteration} LowestBreakIteration:{loopstate.LowestBreakIteration}"); return TLocal+ "=>Break"; } Console.WriteLine($"item:{item} index:{TLocal}" ); return TLocal + "=>complete"; }, local => Console.WriteLine(local) //输出线程本地变量 );
用法七:Item 、循环状态、索引、 线程本地变量
ForEach<TSource,TLocal>(IEnumerable<TSource>, Func<TLocal>, Func<TSource,ParallelLoopState,Int64,TLocal,TLocal>, Action<TLocal>)
string[] data = { "zero", "one", "two", "three", "four", "five", "six", "seven", "eight", "nine", "ten" }; //自行推断TSource,TLocal Parallel.ForEach(data, ()=>"开始输出:",//初始化线程本地变量 (item,loopstate, TLocal,index) => { if(item == "zero") { loopstate.Break(); Console.WriteLine($"item:{item} index:{TLocal} ShouldExitCurrentIteration:{loopstate.ShouldExitCurrentIteration} LowestBreakIteration:{loopstate.LowestBreakIteration}"); return TLocal+ "=>Break"; } Console.WriteLine($"item:{item} index:{TLocal}" ); return TLocal + "=>complete"; }, local => Console.WriteLine(local) //输出线程本地变量 );
Invoke
Invoke(Action[]) :尽可能并行执行提供的每个操作。 返回值Viod
Parallel.Invoke( () => Console.WriteLine("sfsfsdfsdf1"), () => Console.WriteLine("sfsfsdfsdf2"), () => Console.WriteLine("sfsfsdfsdf3") );
Invoke(ParallelOptions, Action[]) :执行所提供的每个操作,而且尽可能并行运行,除非用户取消了操作。返回值Viod
ParallelOptions op = new(); op.TaskScheduler = TaskScheduler.Default;//默认的任务调度器 op.MaxDegreeOfParallelism = 5;//并行度 op.CancellationToken = CancellationToken.None;//禁止取消 Parallel.Invoke( () => Console.WriteLine("sfsfsdfsdf1"), () => Console.WriteLine("sfsfsdfsdf2"), () => Console.WriteLine("sfsfsdfsdf3") );