Sturla Molden <sturla.molden@...> writes:

> 
> Rich Lewis <rl403@...> wrote:
> 
> > I have been running RandomForestRegressor models for a while now on my
> > MacBook using the n_jobs=-1 option, which has worked well in the past.
> 
> The cue here might be "MacBook"...
> 
> Which LAPACK is NumPy and SciPy linked against?
> 
> If multiprocessing is used with Accelerate Framework it will hang. The bug
> is actually in multiprocessing (fixed in Python 3.4, but will not be
> backported to 2.7). If you use multiprocessing you should make sure you use
> Intel MKL, OpenBLAS, or ATLAS instead of Accelerate.
> 
> Sturla
> 

I see the same error using iPython on Linux with custom use of
multiprocessing Pool/Queue (not scikit-learn).  Strangely, the first use of
multiprocessing always works fine -- the error only appears in subsequent
uses after some delay.  So far I haven't found a solution other than
restarting the iPython server.

 Greg


------------------------------------------------------------------------------
Open source business process management suite built on Java and Eclipse
Turn processes into business applications with Bonita BPM Community Edition
Quickly connect people, data, and systems into organized workflows
Winner of BOSSIE, CODIE, OW2 and Gartner awards
http://p.sf.net/sfu/Bonitasoft
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to