Re: reference implementation of BLAS

[Warren Turkal <wt at atmos dot colostate dot edu>]

On Sunday 24 September 2006 18:29, Steve Kargl wrote:
> On Fri, Sep 22, 2006 at 12:37:46PM -0600, Warren Turkal wrote:
> > On Thursday 21 September 2006 15:54, Jerry DeLisle wrote:
> > > I have been building LAPACK which includes BLAS fairly regularly and
> > > most recently, last week.  All built fine.  Can you send me an example
> > > of the failures you are seeing?  I usually do:  grep fail *.out in the
> > > TESTING directory to get a list of failures.
> >
> > These tests are CBLAS tests. The build log is at [1] for you to check
> > out.
> >
> > [1] webdav://scratch.biocycle.atmos.colostate.edu/BLAS-gfortran/ouput.txt
>
> Are you using -ffast-math?  DON'T!

To my knowledge, I am not using -ffast-math. In fact, I shouldn't be using any 
optimization options on the compile. I just looked at the ouput.txt I 
published, and it doesn't have the strings "fast" and "-O" in it anywhere.

wt
-- 
Warren Turkal, Research Associate III/Systems Administrator
Colorado State University, Dept. of Atmospheric Science