I want to congratulate parallel python developers, because PP works in a very perfomance-predictable way.
And multiprocessing package - does not (in python2.6)-
http://bugs.python.org/issue5000
How it can be that multiprocessing package is so buggy, i.e. parallel version of iterations runs about 5 times !!! slower than normal sequential iterations.
And your PP package behaves in predictable way (speeds-up process by a factor=number of cores).
Cool!.