System.Threading.ThreadPool excluding a core?

Question

I have a multi-core architecture computer that is executing processes using .Net 4.5.2 System.Threading.ThreadPool namespace. The processes are long duration math computations. These processes might execute for days and even weeks. I do not want to create my own Thread Pools. Using the System.Threading.ThreadPool namespace is very nice. However, on my multi-core architecture computer, the Thread Pool Manager is very greedy and load balances across all of the cores. Unfortunately, my processes on each core are also greedy. They want to monopolize the core and execute at 100% until it completes its assignment. I'm fine with this, except that the operating system freezes up. Literally, I can't move the mouse around and interact with the desktop. What I would like to do is reserve one of the Cores for the Operating System, so that the mouse and gui are still responsive. It seems logical to exclude a core (and its available threads) for the OS to operate.

Does anyone know how to accomplish this using System.Threading.ThreadPool?

****ANSWER****

To begin, my question is faulty. This was due to my inexperience with the subject. Second, if your google search brought you to this question, it means that your thinking is also faulty; evidenced by your google search words. But this is good. Now you can learn the proper way. And here it is.

The short answer to this question: System.Threading.ThreadPool cannot solve your issue.

A slightly better answer: The Task Parallel Library (TPL) is a set of public types and APIs in the System.Threading and System.Threading.Tasks namespaces in the .NET Framework 4.0. The TPL scales the degree of concurrency dynamically to efficiently use all the cores that are available. By using TPL, you can maximize the performance of your code while focusing on the work that your program is designed to accomplish.

Good luck and happy coding to you!

Answer 1

@ScottChamberlain: because the math operation relies on extracting a value from an external network resource. That resource may be blocked for reasons I can't control (perhaps another user is accessing it, blocking until they release it). – sapbucket

What you need to do is separate the getting data from the network resource from the processing of that resoruce. Use a BlockingCollection as a buffer to pipeline between your two parts.

BlockingCollection<YourData> _collection = new BlockingCollection<YourData>();

public void ProcessInParallel()
{
    //This starts up your workers, it will create the number of cores you have - 1 workers.
    var tasks = new List<Task>();
    for(int i = 0; i < Environment.NumberOfLogicalCores - 1; i++)
    {
        var task = Task.Factory.StartNew(DataProcessorLoop, TaskCreationOptions.LongRunning);
        tasks.Add(task);
    }

    //This function could be done in parallel too, _collection.Add is fine with multiple threads calling it.
    foreach(YourData data in GetYourDataFromTheNetwork())
    {
        //Put data in to the collection, the first worker available will take it.
        _collection.Add(data);
    }

    //Let the consumers know you will not be adding any more data.
    _collection.CompleteAdding();

    //Wait for all of the worker tasks to drain the collection and finish.
    Task.WaitAll(tasks);
}

private void DataProcessorLoop()
{
    //this will pull data from the collection of work to do, when there is no work to do
    //it will block until more work shows up or CompleteAdding is called.
    foreach(YourData data in _collection.GetConsumingEnumerable())
    {
        CrunchData(data)
    }
}