Reputation: 116910
I need to do some extensive searching and string comparisons and for this I figure that a compiled program is much better than an interpreted ones especially after seeing some comparison studies. I came across ParallelPython which was beautiful. It has autodiscovery for clusters and can pretty much do all the load balancing for me as well.
My first question is, is it a good idea to just go ahead with Python on a cluster having 20 nodes or do I switch to C++? If I need to switch then is there a good alternative to ParallelPython for C++ that provides features like load balancing and autodiscovery for node?
Upvotes: 1
Views: 139
Reputation: 564471
You can always use ParallelPython for your high level work, and call into C++ code for the "hard-core" processing, as needed.
That being said, there are options in the C++ world. The most common cluster-based technology is MPI. Some implementations provide load balancing and auto-discovery, though it's not in the core spec.
Upvotes: 1