The TPL (Task Parallel Library) of NET 4 contains a plethora of datatypes for implicit / explicit data and task parallelism + several synchronization primatives. Of course, when you learn something, its never the end of the story…
http://code.msdn.microsoft.com/ParExtSamples contains a class library which leverages and complements the functionality available in the .NET Framework 4 TPL. Please find below my spelunk into this class library:
Parallel General Extensions
- AggregateExceptionExtensions
- BlockingCollectionExtensions – the precursor to C# 5 Asynchrony TAP pattern Join construct ?
- CompletedTask – provides access to an already completed task – useful for using ContinueWith overloads that arnt StartNew equivelents
- ConcurrentDictionaryExtensions
- IProducerConsumerCollectionExtensions
- LazyExtensions
- LinqToTasks – provides LINQ support for Tasks
- ParallelLinqOptions – provides a grouping for PLINQ options
- PlinqExtensions
- TaskCompletionSourceExtensions
- TaskExtensions
Coordination Data Structures
- ActionCountdownEvent – runs an Action when a CountdownEvent reaches zero
- AsynchCache – Caches asynchronously retrieved data – eg streamed data
- ConcurrentPriorityQueue – a thread safe priority queue structure
- ConcurrentRandomNumberGenerator – a thread safe psuedo random number generator
- ObjectPool – a thread safe random number generator
- ObservableConcurrentCollection – a thread safe concurrent collection for use with data binding
- ObservableConcurrentDictionary – a thread safe concurrent dictionary for use with data binding
- Pipeline – provides support for pipeline data processing
- ProducerConsumerCollectionBase – Provides a base implementation for producer-consumer collections that wrap other producer-consumer collections.
- ReductionVariable – Provides a reduction variable for aggregating data across multiple threads involved in a computation
- SpinLockClass – provides a simple referrence type wrapper for SpinLock struct
APM
- FileAsync – provides asych counterparts to members of the File class
- StreamExtensions – extension methods for asynchronously working with streams
- WebRequestExtensions - extension methods for asynchronously working with web requests
EAP
- EAPCommon
- PingExtensions – extension methods for asynchronously working with Ping
- SmtpClientExtensions – extension methods for asynchronously working with SMTP client
- WebClientExtensions – extension methods for asynchronously working with Web Client
TaskFactoryExtensions
- TaskFactoryExtensions
- SelfReplicatingTask – provides a task capable of replicating itself
Parallel Algorithms
- ParallelAlgorithms
Partitioners
-
ChunkPartitioner – partitions an enumerable into chunks based on user supplied criteria
-
RangePartitioner – partitions a range into sub-ranges. For very small loop bodies with a large number of iterations that require equal amounts of processing time, Parallel.ForEach with a RangePartitioner may be more efficient than a Parallel.For.
Task Schedulers
-
IOTaskScheduler - Provides a task scheduler that targets the I/O ThreadPool
-
LimitedConcurrencyLevelTaskScheduler - Provides a task scheduler that ensures a maximum concurrency level while running on top of the ThreadPool.
-
OrderedTaskScheduler - Provides a task scheduler that ensures only one task is executing at a time, and that tasks execute in the order that they were queued.
-
PrioritizingTaskScheduler - Provides a task scheduler that supports reprioritizing previously queued tasks
-
RoundRobinTaskScheduler - Enables the creation of a group of schedulers that support round-robin scheduling for fairness.
-
StaTaskScheduler - Provides a scheduler that uses STA threads
-
SynchronizationContextTaskScheduler - Provides a task scheduler that targets a specific SynchronizationContext
-
ThreadPerTaskScheduler - Provides a task scheduler that dedicates a thread per task
-
WorkStealingTaskScheduler -
