Book Image

Mastering C# Concurrency

Book Image

Mastering C# Concurrency

Overview of this book

Starting with the traditional approach to concurrency, you will learn how to write multithreaded concurrent programs and compose ways that won't require locking. You will explore the concepts of parallelism granularity, and fine-grained and coarse-grained parallel tasks by choosing a concurrent program structure and parallelizing the workload optimally. You will also learn how to use task parallel library, cancellations, timeouts, and how to handle errors. You will know how to choose the appropriate data structure for a specific parallel algorithm to achieve scalability and performance. Further, you'll learn about server scalability, asynchronous I/O, and thread pools, and write responsive traditional Windows and Windows Store applications. By the end of the book, you will be able to diagnose and resolve typical problems that could happen in multithreaded applications.
Table of Contents (17 chapters)
Mastering C# Concurrency
Credits
About the Authors
About the Reviewers
www.PacktPub.com
Preface
Index

Latency and the coarse-grained approach with TPL


Raw performance, or the number of calculations per second that our program is able to perform, is not always a most important goal to achieve. Sometimes it is even more important to stay responsive and interact with the user as fast as possible. Unfortunately, it is not easy to achieve both these advantages at the same time; there are situations when we need to choose our primary goal.

To simulate such a situation, let's create a combination of coarse-grained computational tasks that takes a long time to complete and runs in the background, and a number of short-lived tasks representing user interaction. We would like these short tasks to run as fast as possible with low latency. Now we write a code to test how these long-running tasks can affect latency:

for (var longThreadCount = 0; longThreadCount < 24; longThreadCount++)
{
  // Create coarse grained tasks
  var longThreads = new List<Task>();
  for (var i = 0; i < longThreadCount...