Parallel Xplor?

Kay Diederichs Kay.Diederichs at uni-konstanz.de
Thu Sep 18 09:14:55 EST 1997


Steffen Graether wrote:
> 
> Steffen Graether (steffen at crystal) wrote:
> : I was wondering what the latest status is on a parallel version of xplor.
> : Is there any possibility in volunteering for testing of such a version?
> 
> Arrrgghh. I forgot to also ask: has anyone tried compiling xplor with a
> compiler that 'parallelizes' programmes for you?
> 
> Steffen
> [for a complete address see below]
> 
> --
> *************************************************************************
> *                               *  graethes at jeff-lab.queensu.ca         *
> * Steffen Graether              *  steffen at crystal.biochem.queensu.ca   *
> *                               *                                       *
> *************************************************************************

My experience (about one year ago) with X-PLOR 3.1f on a SGI
PowerChallenge with 10 R8000 processors is that parallelizing is not
worth it, if only the (IRIX 6.0.1 Power Fortran's) compiler's built-in
intelligence is used. The CPU time went down a little if more than 1
processor was used, but the wall clock time was always longer, maybe
because I didn't have the machine for me alone )-:!

If you are willing to insert compiler directives it might be a different
story.

I plan to repeat the experiment with X-PLOR 3.8 on a 2-processor Octane.
-- 
Kay Diederichs                    
email: Kay.Diederichs at uni-konstanz.de     Tel: +49 7531 88 4049  Fax:
3183
Fakultaet fuer Biologie, Universitaet Konstanz 
Postfach 5560 (M656), D-78434 Konstanz, Germany




More information about the Xtal-log mailing list