[Chicago] MPI in Python?

Massimo Di Pierro mdipierro at cs.depaul.edu
Wed Feb 27 18:48:23 CET 2008

In fecmiqcd you can define:

int L[3]={100,100,100}
mdp_lattice cube(3,L);
mdp_field<float> psi(cube)
mdp_site x(cube);

and then

forallsites(x) psi(x)=1.0/6.0*(psi(x-0)+psi(x+0)+psi(x-1)+psi(x+1)+psi 

The latter line solves the Laplace equation in 3D in parallel. MPI is  
taken care of for you.


On Feb 27, 2008, at 11:32 AM, skip at pobox.com wrote:

>     Dan> What are you attempting to do with MPI? I write threaded C  
> at work
>     Dan> and its no big chore to use a few globals and locks to get  
> things
>     Dan> done.
> Oh, sure, but I'm interested in using the 50 or 60 cpu cores I have
> available when doing compute-intensive work, just what MPI should  
> be good
> for.  I like the threading API (or I should say I'm comfortable  
> with it), so
> extending it to a multiprocessing API would make for a small  
> semantic leap
> in my feeble brain.  If that sort of thing can be built on top of  
> MPI, so
> much the better.
> Skip
> _______________________________________________
> Chicago mailing list
> Chicago at python.org
> http://mail.python.org/mailman/listinfo/chicago

More information about the Chicago mailing list