Hello!
Thanks for the fast answer.
I did the FEEDS and it says all errors under 0.24% and all tests passed.
They all have suse Linux 10. So, there can be little difference between the systems?
And it depends on the math system?
Is there a side or something, where I can read something about?
thx for your help
RALF
-----Ursprüngliche Nachricht-----
Von: FSL - FMRIB's Software Library im Auftrag von Jens Sommer
Gesendet: Fr 06.03.2009 13:29
An: [log in to unmask]
Betreff: Re: [FSL] 32bit/64bit systems - difference at bedpostx results at the same dataset
It is - depending on the precision of your math libraries in the base system
(I suppose it's both linux). Did you run FEEDS (FSL Evaluation and Example
Data Suite) on both systems? What does it say?
Bye,
Jens Sommer
"Lützkendorf, Ralf" schrieb am 2009-03-06:
> Hello.
> is there the possibility that there are difference between 32bit and
> 64 core systems if Im using bedpostx?
> I used the newest version of FSL and installed it on two different
> systems. I used bedpostx to estimate the FAs and the dyadics on the
> same data sets.
> At the end, the results were different. Vectors are nearly the some
> but some are a several degrees different, also at the main fibers.
> I hope somebody can help or explain it.
> Ralf
> +++++++++++++++++++++++++++++++++++++++++++++++
> Ralf Lützkendorf
> Institute for Biometry and Medical Informatics
> University of Magdeburg, Germany
> phone: +49 391 67 13546
> fax: +49 391 67 13536
> IMPORTANT LINKS:
> Newsworthy:
> http://whales.greenpeace.org/global
> Permanent:
> http://www.sharkproject.org/
> http://www.bergwaldprojekt.de/
> http://www.greenpeace.de/
> http://www.seashepherd.org/
> http://www.wdcs-de.org/
> http://www.we-feed-the-world.at/
|