2008/1/8, Ray Schumacher <[EMAIL PROTECTED]>: > > At 04:27 AM 1/8/2008, you wrote: > > 4. Re: parallel numpy (by Brian Granger) - any info? > > (Matthieu Brucher) > >From: "Matthieu Brucher" <[EMAIL PROTECTED]> > > > >MKL does the multithreading on its own for level 3 BLAS instructions > >(OpenMP). > > There was brief debate yesterday among the Pythonians in the lab as > to whether the MKL operates outside of the GIL, but it was general > understanding that it does.
As Numpy releases the GIL, it does ;) It is still unclear to me whether Python/numpy compiled with MKL > would be freely re-distributable, as the MSVC version is. > > >Then you must use tools like the processing module, MPI, ... > > I have been using shared memory and numpy with good results lately on > Win32 with tagged memory. > > http://projects.scipy.org/pipermail/numpy-discussion/2007-October/029505.html > > On *NIX, arrayfrombuffer(), mmap and MAP_SHARED, or POSH > http://poshmodule.sourceforge.net/ > seem like the ticket. I hope that one day, there will be a Python version that will be able to handle distributed computers, that is distributing one thread per computer (or core). It could be interesting to use if the algorithm was parallelized. Matthieu -- French PhD student Website : http://matthieu-brucher.developpez.com/ Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92 LinkedIn : http://www.linkedin.com/in/matthieubrucher
_______________________________________________ Numpy-discussion mailing list Numpy-discussion@scipy.org http://projects.scipy.org/mailman/listinfo/numpy-discussion
