When it rains, it pours --- articles on parallel processing that is

Wow, my reading pile just got a lot bigger today.  First, the latest (Oct 2008) issue of MSDN contains 4 articles related to parallel processing: 

1. "Design considerations for parallel programming"
2. "Improved Support for Parallelism in the next version of Visual Studio"
3. "Solving 11 Likely Problems in your Multithreaded Code"
4. "Build Concurrent Apps From Simple F# Expressions"

I'm currently reading #4, since I'm a believer that functional programming may be one of the ways we make parallel processing safer, scalable, and generally more approachable.  I don't see much in F# yet, it seems right now to be just another way to gain access to PFx and MPI.NET.  But the potential is definitely there. 

If that isn't enough, the latest issue (Oct 2008) of ACM's Queue magazine ("Architecting Tomorrow's Computing") presents a broader view of the concurrency problem, with articles on

1. "Real-world Concurrency"
2. "Erlang for Concurrent Programming"
3. "Parallel Programming with Transactional Memory"

Erlang is another functional language, but with a proven process-based approach to concurrency and parallel processing.  I'm just learning about Erlang now, I really really need to implement some apps and collect some performance numbers.  Erlang's process-oriented concurrency model was originally designed for system building and robustness, so the question is how it fares for HPC.  Transactional memory is an interesting approach --- in short, you don't worry about accessing shared memory, you just go for it and check when you're done to see if a conflict has occurred.  If not, you commit, otherwise you rollback and try again.

Anyway, happy reading!