Ole Nielsen wrote: > Cheers and thanks Now that it is possible to purchase servers from tier-1 manufacturers with dual quad-core processes (8 CPUs) for under $10k, complete with discs and lots of RAM - that is, no tedious-to-set-up, space-occupying, power-hungry and high-maintenance Beowulf-style cluster of commodity boxes necessary - we are very keen to use PyPar to re-visit parallelisation of our NetEpi epidemiological analysis code, but we are aware that both our code and PyPar need to be updated to use NumPy as opposed to the now deprecated Numeric Python. We still regard PyPar is still the easiest-to-use wrapper for MPI for Python.
Tim C -- http://mail.python.org/mailman/listinfo/python-list
