After a lot of messing around with pycuda attempting to solve my "Python:
version mismatch?" error, I figured out that the boost library is calling
the system python that comes installed on macs (the one in
/System/Library/Frameworks instead of the one in /Library/Frameworks). I
changed all of the install settings, and boost says that it is using the
proper python, but it is not. Despite all indications, boost continues to
access the system python. So to fix this, I created a symbolic link in the
system python that redirects to the real python. This is kind of a hack, but
it worked temporarily. Obviously I can't leave that, but I don't know what
else to do.
So now I am at the point where pycuda runs, but still doesn't work. For some
reason, the Cuda.init() call run when importing pycuda cannot find any
devices... I am on one of the unibody MacBook Pros, and I am wondering if
the problem might be the presence of both graphics cards (I don't think the
9400M supports Cuda).
If anyone has any insight, I would really appreciate it!
Thanks!
>>>Chris Heuser
_______________________________________________
PyCUDA mailing list
[email protected]
http://tiker.net/mailman/listinfo/pycuda_tiker.net
