Hi! I just wonder if there are any plans for building any support using NIVIDAS CUDO plattform into python? For example when implementing Monte Carlo simulation performance always is an issue so extended numpy/numeric or a math library using the CUSO plattform could be a "cheap" grid approach.
I myself implemented a simple "batch grid" for distribute calculations to be executed in parallell supporting a typical pattern as: for x in collect_data(): post(calc(x)) syncronize() generate_report(get_calculations()) So one idea supporting this pattern would be to define some kind of new form of generators which allows independ calculations be executed in parallell and depending on form returns the calculation either in synchronized way (i.e. returned in sequence order) or in a asyncronized order. One other thing: for a long time I worked with a financial system, SunGard FRONT ARENA, which has integrated python into the system in a clever way according to Fredrik Lundh. I suggest that you extend the application list with this system. /Fredrik Kant www.kantconsulting.se -- http://mail.python.org/mailman/listinfo/python-list