Parallel extensions in .Net 4.0
Download as PPTX, PDF2 likes928 views
The document discusses parallel extensions in .NET 4.0, including an overview of parallel loops, tasks, PLINQ, and coordination data structures (CDS). It notes that parallelism relies on increasing CPU core counts rather than frequency. Parallel loops, tasks, and PLINQ allow exploiting multi-core systems. Tasks control work using a task scheduler and thread pool. PLINQ parallelizes LINQ queries. CDS like concurrent collections help synchronize concurrent tasks. Examples demonstrate parallel algorithms and performance analysis tools.
Parallel extensions in .Net 4.0
- 1. Parallel Extensions in .Net 4.0Dmytro Maleev for Lviv.Net User Group
- 2. AgendaКуда катится мир?Hello, Parallel Extensions!Parallel vs MultithreadingParallel LoopsTasksPLINQCDS & Thread EnchantmentsReferencesQ&A
- 3. Куда катится мирForget about CPU frequency increase! There are a lot of issues with this, which will be fixed in future!1998 Intel 8048680Mhz1 Core2007Core 2 Duo 2.13 GHz2 Core2000Celeron 400Mhz1 Core2010Intel i5 3.3GHz2 Core 4 Threads2003Athlon XP 1.6+GHz1 Core?
- 4. Куда катится мирFor now…ENLARGE YOUR…CPU CORE COUNT!PENPEN
- 5. Hello, Parallel Extensions!Parallel Extensions, previously known as the Parallel Framework Extensions or PFX, is a managed concurrency library being developed by a collaboration between Microsoft Research and the CLR team at Microsoft. It is composed of two parts: Parallel LINQ (PLINQ) andTask Parallel Library (TPL).It also consists of a set of coordination data structures (CDS) – sets of data structures used to synchronize and co-ordinate the execution of concurrent tasks.Thelibrary was released as a CTP on November 29, 2007and refreshed again in December 2007 and June 2008. Microsoft has announced that the Parallel Extensions to .NET will release as part of the .NET 4.0 Framework release.©Wikipedia
- 6. Hello, Parallel Extensions!Visual Studio Debugging and profiling supportParallel ExtensionsPLINQTask Parallel LibraryTask SchedulerCDS
- 7. Parallel vs MultithreadingMultithreaded!=parallelization“ If on a single core machine you are using threads and it makes perfect sense for your scenario, then you are not "doing parallelism", you are just doing multithreading””On a single core you can use threads and you can have concurrency, but to achieveparallelism on a multi-core box you have to identify in your code the exploitableconcurrency: the portions of your code that can truly run at the same time.”Daniel Moth
- 9. Dark Side of Paprallelization and MultithreadingRace conditionshttp://en.wikipedia.org/wiki/Race_condition2. Deadlockshttp://en.wikipedia.org/wiki/Deadlockhttp://en.wikipedia.org/wiki/Dining_philosophers_problemThread starvationDifficult to code and debugEnvironmental
- 11. Parallel LoopsParallel.For()Parallel.For(0, 100, i => { Console.WriteLine("This is I-I-I-I-I-I: {0}", i); });Parallel.ForEach()Parallel.ForEach(ThisIsCollection, collectionItem =>{collectionItem.Hello();});Parallel OptionsParallel.Invoke()
- 12. Parallel Loops. Use Force wisely!Parallelization Can Hurt Performancehttp://msdn.microsoft.com/en-us/library/dd560853(VS.100).aspxhttp://en.wikipedia.org/wiki/Context_switch
- 14. Task - Task SchedulerTask is a new class that represents the work you want completed. There are methods to create, schedule, and synchronize tasks in your application.Tasks are controlled by task scheduler.Tasks scheduler works with thread pool.
- 15. Task In DeepTasks can:Task can be createdTask can wait! Task can simply Wait(), WaitAll() or WaitAny().Task knows when it is completed ( IsCompleted property)Task can ContinueWith()Task can return valueTask has Options and Status
- 17. PLINQPLINQ is just parallelized version of LINQNot parallelized: LINQ-to-SQL & LINQ-to-Entity.LINQ:var query = from s in someCollectionlet result = CoolService.CallService(s)select result;PLINQ:varquery = from s in someCollection.AsParallel()let result = CoolService.CallService(s)select result;
- 18. PLINQ. How it works?
- 19. PLINQ. Overview Ordering ResultsAsOrdered()ForAll Operator()AsSequential()WithMergeOptionsParallel Performance Analyzer. Just for rich
- 20. DEMO Baby Names
- 21. CDS & Thread EnchantmentsThread Enchantments:Thread.Yield()Monitor.Enter()Concurrent CollectionsConcurrentStack (LIFO)ConcurrentQueue (FIFO)ConcurrentDictionaryConcurrentBagBlockingCollection
- 22. Synchronization PrimitivesBarrier“Let’s meet near monument and then go to have a beer”Cancellation TokensCountDownEventManualResetEventSlim and SemaphoreSlimSpinLockThreadLocal<T>
- 23. References Parallel Programming with .NEThttp://blogs.msdn.com/b/pfxteam/Wikihttp://en.wikipedia.org/wiki/Parallel_ExtensionsIntroducing .NET 4.0http://www.amazon.com/Introducing-NET-4-0-Visual-Experts/dp/143022455XLviv.Net User Grouphttp://dotnetug-lviv.blogspot.com/
- 24. Q&A?
- 25. If you still have a questions Mail me:diwingless@gmail.comSkype me:hmmidmaTwitter:dimko1