home *** CD-ROM | disk | FTP | other *** search
- Path: hubcap.clemson.edu!hubcap!mjs
- From: mjs@hubcap.clemson.edu (M. J. Saltzman)
- Newsgroups: comp.lang.c
- Subject: Re: Matrix Multiplication
- Date: 14 Feb 96 15:25:45 GMT
- Organization: Clemson University
- Message-ID: <mjs.824311545@hubcap>
- References: <DLyvL5.9uH@cwi.nl> <4fsfrj$pbs@mserv1.dl.ac.uk>
- NNTP-Posting-Host: hubcap.clemson.edu
-
- I.J.Bush@dl.ac.uk (I.J. Bush) writes:
-
- >Well, just a slightly different perspective from a FORTRAN coder
- >in the high performance computing field. What FORTRAN people do,
- >at least in my area, for a matrix mult is not to write their own
- >routine to do it, but to use the vendor supplied BLAS library
- >[...]
-
- >So basically what I am saying is that if you want to do a lot
- >matrix mults you might consider trying to use the BLAS library,
- >but then again you know C, and I believe I know FORTRAN :)
-
- Of course, on most systems, nothing stops a C programmer from calling
- BLAS library routines either. The only inconvenience is dealing with
- the row-major vs. column-major layout in C and FORTRAN. Dik's point
- (in essence) is that one could conceivably write C-style (row-major)
- BLAS routines that run competitively with the FORTRAN ones.
-
- --
- Matthew Saltzman
- Clemson University Math Sciences
- mjs@clemson.edu
-