Newsgroups: comp.parallel
From: rvdg@cs.utexas.edu (Robert van de Geijn)
Subject: General Sparse Matrix Distribution
Organization: CS Dept, University of Texas at Austin
Date: Thu, 7 Dec 1995 19:29:40 GMT
Message-ID: <4a7fb4$13oi@daffy.cs.utexas.edu>

In a recent article, Jeff Taylor writes:
    
    >
    >Hi
    >
    >I am looking for references about sparse matrix multiplication on
    >parallel computers. Especially, I am interested in data distribution
    >and layout patterns for general sparse matrices.
    >
    
            I am also interested in these types of problems.  Please post
    replies here and/or send them to me as well.  
    
    Thanks!
    
    Jeff Taylor
    University of Pennsylvania
    
    Jeff@aries.wistar.upenn.edu

I will assume that by general sparse, you mean matrices
with an unknown sparsity structure, or one that cannot
be easily exploited.  This situation is similar to the
one posed in the NAS parallel CG benchmark.  We wrote
two papers concerning this, 

J. G. Lewis, D. G. Payne, and R. A. van de Geijn, ``Matrix-Vector
Multiplication and Conjugate Gradient Algorithms on Distributed Memory
Computers,'' Scalable High Performance Computing Conference 1994

J. G. Lewis and R. A. van de Geijn, ``Implementing Matrix-Vector
Multiplication and Conjugate Gradient Algorithms on Distributed Memory
Multicomputers,'' Supercomputing~'93.

Details can be found in 

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

click on "conference publications"

I believe the postscript for the SC93 paper is messed up.
The SHPCC94 paper gives more details.

Notice that the two dimensional data distribution advocated in these
papers is not necessarily the best in the case of a matrix with known
sparsity structure.

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 

-- 



