2012/10/18 Gael Varoquaux <[email protected]>: > On Thu, Oct 18, 2012 at 07:02:41AM +0100, Andreas Mueller wrote: >> In principle I think it should be possible to run with the standard numpy as >> well. > > One thing to keep in mind is that accelerate leads to crashes after doing > a fork. Thus if you want to do multiprocessing/joblib-based parallel > computing, it should not be used. > > In general, I would recommend to rely rather on something like atlas when > doing scientific computing with Python. > > Still, we should be able to build correctly :$
Even though it's not officially supported by Apple, the "bug" seems to have been fixed in 10.8. The fork bug is still there in 10.7. In 10.6, grand central dispatch is either not shipped by default or at least not used by numpy hence the problem does not happen. David is working on making it easier to build numpy / scipy against OpenBlas (which should be simpler to build than atlas and sometimes faster) however I could not manage to build / install a recent enough llvm on my box (so far). http://cournape.wordpress.com/2012/10/10/notes-on-building-numpyscipy-with-openblas/ I have not tried that hard though. -- Olivier http://twitter.com/ogrisel - http://github.com/ogrisel ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_sfd2d_oct _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
