From help-octave-request at bevo dot che dot wisc dot edu  Mon Feb  7 11:35:52 2000
Subject: Re: NaN problem
From: stevenj at gil-galad dot mit dot edu
To: help-octave at bevo dot che dot wisc dot edu
Date: Mon, 7 Feb 2000 12:37:59 -0500 (EST)

Whoops, I forgot the subject line on my last message (see below).

If you are really worried about this sort of thing, by the way, you can
always modify my patch so that it does --without-fastblas by default.  So,
the user would specifically have to configure --with-fastblas to use
external BLAS and LAPACK libraries, and would presumably know what she is
doing and value the additional performance over any quirks.

Steven

---------- Forwarded message ----------
Date: Mon, 7 Feb 2000 12:30:46 -0500 (EST)
From: stevenj at gil-galad dot mit dot edu
Reply-To: stevenj at alum dot mit dot edu
To: help-octave at bevo dot che dot wisc dot edu

JWE wrote:
> What do the vendor-specific versions of the BLAS do for this code?
> What does ATLAS do?

ATLAS 3.0beta handles it correctly:

octave:1> [0;0]*[ inf nan ]
ans =

           NaN           NaN
           NaN           NaN

octave:2> [ inf; nan ]*[0 0]
ans =

           NaN           NaN
           NaN           NaN

I don't know about the vendor BLAS libraries.

Steven




-----------------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.che.wisc.edu/octave/octave.html
How to fund new projects:  http://www.che.wisc.edu/octave/funding.html
Subscription information:  http://www.che.wisc.edu/octave/archive.html
-----------------------------------------------------------------------