An alternative that doesn't seem to have been mentioned, which I am
very pleased with and have had lots of success with, is Sage. It's a
large mathematical computing environment like Matlab but mainly
developed by mathematicians (so lots of symbolic and abstract stuff),
written in Python and leveraging a lot of other open-source efforts
(NumPy, GAP, Singular, etc.), and comes in one big tarball that you
can put anywhere and works with almost any distro. It comes with its
own Python installation built-in ("sage -python" or "sage -ipython"),
so I just installed Boost and PyCUDA against that (no dicking with
distro-specific garbage) and use PyCUDA without any problems. However,
all this is useful after you've figured out the driver/SDK issues :(.Sage also has Cython, which involves adding a few type declarations to your Sage/Python code to get C speed. (Of course things never work as simply as that, and like everything there are idiosyncrasies, but probably nothing insurmountable.) I haven't yet personally used this (will very soon, to do stuff on non-CUDA cluster nodes), but it might hit the sweet spot for machine efficiency and programmer productivity for you. This overview presentation should be a good point to start (it's astonishing how much stuff they've put into Sage): http://wstein.org/talks/20090310-msri-sd14/msri-talk.pdf On Fri, May 8, 2009 at 1:46 PM, Andrew Wagner <[email protected]> wrote: > Hello- > > Is there a recipe out there for getting PyCUDA (including > dependencies!) up and running that is known to just work? > > I'm hopeful that if I can get PyCUDA installed properly, it will > become my primary research tool. I used MATLAB exclusively for years > and loved it. Once that became impossible for performance reasons I > switched to C++ and CUDA. Now my code runs a lot faster, but I'm only > about 1/20th as productive as I was with MATLAB. Since I'm hoping > this will be my primary research tool, I'm willing to do a clean > install of any OS (though I prefer anything but windows, and have a > slight preference for mac or debian). > > Even if there is no well-tested recipe, I'd appreciate guidance on > what platforms, package managers, library versions, etc... are best > supported. I already spent a couple days tinkering with this, and I > plan to devote up to a week more, starting in about one week. I've > been using cuda on OS X for about a semester, but I'm pretty new to > python. > > Thanks a lot! > Drew > > _______________________________________________ > PyCuda mailing list > [email protected] > http://tiker.net/mailman/listinfo/pycuda_tiker.net > _______________________________________________ PyCuda mailing list [email protected] http://tiker.net/mailman/listinfo/pycuda_tiker.net
