Brennan Spies writes about various problems that current concurrency paradigm experiences and mentions a few potential approaches for solving the problem.
“We’ve all heard the same story: the “free lunch” for developers is over. The exponential rocket ride of single-processor performance is no longer feasible, and in order to sustain Moore’s Law, scaling out with more processor cores is now the convention. Developers eager to give their software a performance boost can no longer simply wait for the next generation of hardware; instead they must write more concurrent code in order to maximize utilization of the available CPUs.
But concurrent programming (at least as it is practiced now) is complicated, and many developers cringe at the thought of race conditions, deadlocks, contention, and other problems that await them in the murky waters of multi-threading…In many ways they are in the same situation that they were with memory management back in the 1990’s (before garbage collection became mainstream). Is there a “silver bullet” for concurrency waiting out there as well?
In this article I will cover some of the current and potentially near-future approaches to handling concurrency, and evaluate some of the potential “saviors-in-waiting”.”
Related Links
The Kilim framework
“Beautiful Concurrency“, a draft chapter by Simon Peyton Jones for the O’Reilly book
Thoughts on Concurrency - Part 1 Part 2
