Re: Octave QR factorization

[Ross Vandegrift]

On Wed, Dec 17, 2003 at 02:40:34PM -0600, John W. Eaton wrote:
> If Matlab gets it "right" on x86 hardware, then perhaps it does not
> use the internal 80-bit register operations.  Does anyone know of a
> way to do that with gcc without using -ffloat-store?  (I'd rather not
> use -ffloat-store by default due to the potential performance
> penalty.)

Someone should check on an Intel machine first.  I recently built Octave
for my system, and all the dependencies, and there's a huge warning in
the ATLAS library against using 3DNow!  The README claims that the
floating point in AMD processors isn't 100% IEEE and that enabling
3DNow! in ATLAS can cause numerical problems.

I wonder if this is the cause?

The script does fail on my AMD Thunderbird, but I ignored the ATLAS
warnings and build with 3DNow - burning speed was more important to me
::-)


-- 
Ross Vandegrift
address@hidden

A Pope has a Water Cannon.                               It is a Water Cannon.
He fires Holy-Water from it.                        It is a Holy-Water Cannon.
He Blesses it.                                 It is a Holy Holy-Water Cannon.
He Blesses the Hell out of it.          It is a Wholly Holy Holy-Water Cannon.
He has it pierced.                It is a Holey Wholly Holy Holy-Water Cannon.
He makes it official.       It is a Canon Holey Wholly Holy Holy-Water Cannon.
Batman and Robin arrive.                                       He shoots them.



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

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