home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.sgi
- Path: sparky!uunet!usc!rpi!batcomputer!ghost.dsi.unimi.it!itnsg1.cineca.it!root
- From: root@itnsg1.cineca.it (Valter Cavecchia)
- Subject: Re: BLAS, scientific subroutines etc.
- Message-ID: <1992Aug21.082920.15515@itnsg1.cineca.it>
- Organization: Laboratorio di Fisica Computazionale, INFM. Trento Italia
- References: <1992Aug19.130918.28150@itnsg1.cineca.it>
- Date: Fri, 21 Aug 1992 08:29:20 GMT
- Lines: 23
-
- root@itnsg1.cineca.it (Valter Cavecchia) writes:
-
- >I noticed that SGI gave use a version of the BLAS routines.
- >However BLAS itself is very low level, what I'm looking for is something
- >like Linpack (a set of Fortran callable subroutines that are able to do
- >matrix diagonalization, eigenvalues and so on).
- >Is there anything around (BLAS based and already ported to SGI)?
-
- > Thanks a lot in advance,
- > valter
-
- ok, thanks a lot to everyone that answer my question.
- Everyone pointed me to LAPACK (which is the evolution of LINPACK+EISPACK),
- it seems that this set of routines is the state of the art of the
- matrix manipulation software (and is public domain, or something like that).
- The good news is that SGI people already ported LAPACK to the SGI machines,
- the software may be downloaded from sgi.com (beware of the files size!).
- LAPACK is based on BLAS which is already available for SGI computers
- (parallelized and optimized), therefore the performance should be very good.
-
- Thanks a lot again,
- valter
-
-