Friday, 6 November 2009

Low-lock algorithms

The F#.NET Journal just published an article about concurrent programming:

"Locks are the defacto-standard approach to concurrent programming. With locks, shared mutable resources are protected using mutual exclusion to ensure consistency of the shared state from the point of view of different threads. However, locks are slow and, consequently, many situations can benefit from the avoidance of locks in favor of lower-level solutions. Much research has gone into non-blocking (obstruction-free, lock-free and wait-free) algorithms. This article introduces the concepts that underpin the implementation of non-blocking algorithms, including the .NET memory model, and explains how this can be leveraged to avoid locks with the simple example of lazy evaluation..."

To read this article and more, subscribe to The F#.NET Journal today!

No comments: