#parallel-programming

[lm3m]

[stupidjavascript-blog]

This is some really cool stuff.

[jzinedine]

On Threading

I'm reading parts of CLR via C# 3rd Edition by Jeffrey Richter, mostly threading part.

I think it's one of the most insightful books on threading in Windows, he has an intense knowledge of threading implementation details, patterns and anti-patterns on Windows and CLR. 

I recommend it to all who wants to have a profound practical knowledge of threading in .NET .

[federalbottocks]

Guy L. Steele Jr. believes that it should not be the programmer’s job to think about parallelism, but languages should provide ways to transparently run tasks in parallel. This requires a new approach in building languages supporting algorithms built on independence and build-and-conquer principles rather than on linear decomposition of problems. 

Interesting lecture, with a perhaps too long intro, but it makes its point. Probably  every person who dealt with parallelism paradigms today, probably thought about this at least once.

The tools I learn in Parallel And Distributed Programming course (MPI, OpenMP and such) are not only old, but feel like a reworking of old linear concepts. You work with the same sequential mindset, only with multiple slaves, where you are the master.

Are stateless semi-functional patterns the future? It seems to me that they are, because that's how complex systems (ie biology) work. Getting parallel results from semi-autonomous agents. Now that's interesting. Another take-out for me is the need to analyze a problem mathematically, before setting out to design an algorithm. And to keep in mind Algebra concepts (associativity, commutativity and so forth) when thinking about dividing problem domains.