Regarding using MPI, I've made a proof of concept FFI for OpenMPI. I believe that Grant Rettke was working on one too, but I didn't see any evidence of it online.
It is available at: http://github.com/jeapostrophe/openmpi I'd like to get it ready for wider use. It's not ready for PLaneT yet, but you can see an example at: http://github.com/jeapostrophe/openmpi/blob/master/examples/master_server.rkt Here's an example output: ~github/openmpi/examples % mpirun -np 8 racket -t master_server.rkt [0/8 jay-mccarthys-macbook.local]: I am the master [0/8 jay-mccarthys-macbook.local]: Received: Processor 1 reporting! [0/8 jay-mccarthys-macbook.local]: Received: Processor 2 reporting! [0/8 jay-mccarthys-macbook.local]: Received: Processor 3 reporting! [1/8 jay-mccarthys-macbook.local]: I am the servant [1/8 jay-mccarthys-macbook.local]: Received: Hello 1... [2/8 jay-mccarthys-macbook.local]: I am the servant [2/8 jay-mccarthys-macbook.local]: Received: Hello 2... [3/8 jay-mccarthys-macbook.local]: I am the servant [3/8 jay-mccarthys-macbook.local]: Received: Hello 3... [4/8 jay-mccarthys-macbook.local]: I am the servant [4/8 jay-mccarthys-macbook.local]: Received: Hello 4... [5/8 jay-mccarthys-macbook.local]: I am the servant [5/8 jay-mccarthys-macbook.local]: Received: Hello 5... [6/8 jay-mccarthys-macbook.local]: I am the servant [6/8 jay-mccarthys-macbook.local]: Received: Hello 6... [7/8 jay-mccarthys-macbook.local]: I am the servant [7/8 jay-mccarthys-macbook.local]: Received: Hello 7... [0/8 jay-mccarthys-macbook.local]: Received: Processor 4 reporting! [0/8 jay-mccarthys-macbook.local]: Received: Processor 5 reporting! [0/8 jay-mccarthys-macbook.local]: Received: Processor 6 reporting! [0/8 jay-mccarthys-macbook.local]: Received: Processor 7 reporting! Jay On Mon, Oct 4, 2010 at 9:14 AM, chris lanz <[email protected]> wrote: > Thanks to Jay and Robby for their replies. > > I honestly don't care what form of message-passing I use, I just don't want to > > write my own. My app will eventually run different very large programs > (100,000 > > lines of C-like procedural code) on clustered machines controlled by a head > node. > > I have to rewrite all my original code in SOMETHING (it's in PASCAL because > > that's what I'm virtuosic in and that's what I've been using for 25 years) > and I > was preparing to rewrite in C and use MPI, but I've decided it's better to > > switch all at once. > > If I learn Gambit Scheme (that apparently includes an Erlang link), > how far from > Racket is that? > > Or is there a better solution? For instance, should I use Racket (which is my > > preference because we might start teaching it here) and assume that by the > time > > I need it (a few years from now) there will be some intrinsic clustering > facilities? > > THANKS > > CHRIS LANZ > > [email protected] > > > > > > > _________________________________________________ > For list-related administrative tasks: > http://lists.racket-lang.org/listinfo/users > -- Jay McCarthy <[email protected]> Assistant Professor / Brigham Young University http://teammccarthy.org/jay "The glory of God is Intelligence" - D&C 93 _________________________________________________ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/users
