Newsgroups: comp.parallel.mpi
From: dongarra@dasher.cs.utk.edu (Jack Dongarra)
Subject: ScaLAPACK software update
Organization: Computer Science Dept, University of Tennessee, Knoxville
Date: 18 May 1996 20:12:53 GMT
Message-ID: <4nlb05INN79k@CS.UTK.EDU>
Subject: ScaLAPACK software update
The ScaLAPACK project is a collaborative effort between:
Oak Ridge National Laboratory Rice University
University of Tennessee University of California, Berkeley
University of California, Los Angeles University of Illinois
The ScaLAPACK project is made up of 4 components:
dense matrix software (ScaLAPACK)
large sparse eigenvalue software (PARPACK)
sparse direct systems software (CAPSS)
preconditioners for large sparse iterative solvers (PARPRE)
ScaLAPACK, version 1.2, includes routines for the solution of linear
systems of equations, symmetric positive definite banded linear systems
of equations, condition estimation and iterative refinement, for LU and
Cholesky factorization, matrix inversion, full-rank linear least
squares problems, orthogonal and generalized orthogonal factorizations,
orthogonal transformation routines, reductions to upper Hessenberg,
bidiagonal and tridiagonal form, reduction of a symmetric-definite
generalized eigenproblem to standard form, the symmetric, generalized
symmetric and the nonsymmetric eigenproblem.
Software is available in single precision real, double precision real,
single precision complex, and double precision complex. The software has
been written to be portable across a wide range of distributed-memory
environments such as the Cray T3, IBM SP, Intel series, TM CM-5,
clusters of workstations, and any system for which PVM or MPI is available.
A draft ScaLAPACK Users' Guide and a comprehensive Installation Guide is
provided, as well as test suites for all ScaLAPACK, PBLAS, and BLACS routines.
Future releases of ScaLAPACK will include routines for the solution of
general banded linear systems, general and symmetric positive definite
tridiagonal systems, rank-deficient linear least squares problems,
generalized linear least squares problems, and the singular value
decomposition. Also available will be the full-index PBLAS, which will
have the majority of alignment restrictions removed, as well as the
ability to operate on partial first blocks. The next release of
ScaLAPACK is slated for Fall, 1996.
PARPACK (Parallel ARPACK) is an extension of the ARPACK software package
used for solving large scale eigenvalue problems on distributed memory
parallel architectures. The message passing layers currently supported
are BLACS and MPI. Serial ARPACK must be retrieved and installed prior to
installing PARPACK. All core ARPACK routines are available in single
precision real, double precision real, single precision complex, and double
precision complex. An extensive set of driver routines are available
for ARPACK and a subset of these are available for parallel computation with
PARPACK. These may be used as templates that are easily modified to construct
a problem specific parallel interface to PARPACK.
CAPSS is a fully parallel package to solve a sparse linear system of the form
Ax=b on a message passing multiprocessor; the matrix A is assumed to be
symmetric positive definite and associated with a mesh in two or three
dimensions. This version has been tested on the Intel Paragon and makes
possible efficient parallel solution for several right hand side vectors.
PARPRE is a package of parallel preconditioners for general sparse
matrices. It includes classical point/block relaxation methods,
generalized block SSOR preconditioners (this includes ILU), and domain
decomposition methods (additive and multiplicative Schwarz, Schur
complement). The communication protocol is MPI, and low level
routines from the Petsc library are used, but installing the complete
Petsc library is not necessary.
Funding for this effort comes in part from DARPA, DOE, NSF, and CRPC.
For more information on the availability of each of these packages and
their documentation, consult the scalapack index on netlib.
The URL is:
http://www.netlib.org/scalapack/index.html
or type:
echo "send index from scalapack" | mail netlib@www.netlib.org
Comments/suggestions may be sent to scalapack@cs.utk.edu.
Regards,
Jack Dongarra
--
**************************************************************
Jack Dongarra dongarra@cs.utk.edu 104 Ayres Hall
423-974-8295 fax: 423-974-8296 Knoxville TN, 37996
http://www.netlib.org/utk/people/JackDongarra.html