Newsgroups: comp.parallel,comp.parallel.pvm
From: rvdg@cs.utexas.edu (Robert van de Geijn)
Subject: Matrix Multiplication
Organization: CS Dept, University of Texas at Austin
Date: Thu, 9 Nov 1995 10:00:56 GMT
Message-ID: <47sjgo$1588@daffy.cs.utexas.edu>

In comp.parallel (moderated) #13178 Pick Daniel writes:

> 
> 
> In article <475gre$spf@diable.upc.es> Antonio Scotti 
> <scotti@centauro.upc.es> writes:
>
> >Hi,
> >does anybody know how many and which algorithms for parallel matrix
> >multiplication are there around?
> >Does anybody have references to some relevant literature?
> >
> >Thanks
> >-- Antonio
> >
> >
>      I think at the moment you write your own

In our paper (van de Geijn and Watts) "SUMMA: Scalable Universal
Matrix Multiplication Algorithm" we give general approaches to
C = A B, C = A^T B, C = A B^T and C = A^T B^T.  We even include
MPI code in the paper.  In a followup paper (Chtchelkanova, Gunnels, Morrow, 
Overfelt, and van de Geijn), we give parallel implementations of all 
"level 3 Basic Linear Algebra Subprograms", and the associated code
is available via anonymous ftp under gnu-license.  Again, the implementation
is done using vanilla MPI.  All algorithms attain in practice 
within 20 percent of theoretical peak.  Most attain within 5-10% of
peak.

For information see

http://www.cs.utexas.edu/users/rvdg

and click on "Reports" or "Major Software Projects"

Enjoy
Robert

=========================================================================

Robert A. van de Geijn                  rvdg@cs.utexas.edu  
Associate Professor                     http://www.cs.utexas.edu/users/rvdg
Department of Computer Sciences         (Work)  (512) 471-9720
The University of Texas                 (Home)  (512) 251-8301 
Austin, TX 78712                        (FAX)   (512) 471-8885 



