[Dune] Clang++ for Blue Gene/P and Blue Gene/Q.

Dr. Olaf Ippisch olaf.ippisch at iwr.uni-heidelberg.de
Mon Nov 4 15:36:40 CET 2013


Dear Christoph,

I am a bit surprised about one of the statements you cited. I have two
programs solving Richards' equation and transient solute transport,
which are not completely DUNE based, but make use of DUNE-ISTL. I
recently did comparisons on JUQUEEN (the BlueGene/Q in Jülich) comparing
the runtime compiling them with xlc and with gcc. In my experience, the
runtime compiling with gcc was shorter than the runtime with xlc
compiled binaries (even using the rather outdated gcc 4.4). Thus I do
not understand, why there should be a factor of two to gain.

Best regards,
Olaf

Am 30.10.13 15:28, schrieb Christoph Grüninger:
> Hi Dune users on super computers,
> recently I talked to LLVM developer Hal Finkel. He told me about his
> Clang port to Blue Gene/P and Blue Gene/Q. He claims to be usually two
> to three times faster compared to IBM's gcc port and be as fast as XLC.
> As XLC does afaik not compile Dune this might be handy. There are RPMs
> for the download but I did not test them myself.
> 
> More informations:
> https://wiki.alcf.anl.gov/parts/index.php/LLVM
> https://trac.alcf.anl.gov/projects/llvm-bgq/
> 
> Dreaming about factor 3,
> Christoph
> 

-- 
Dr. Olaf Ippisch
Universität Heidelberg
Interdisziplinäres Zentrum für Wissenschaftliches Rechnen
Im Neuenheimer Feld 368, Raum 4.24
Tel: 06221/548252   Fax: 06221/548884
Mail: Im Neuenheimer Feld 368, 69120 Heidelberg
e-mail: <olaf.ippisch at iwr.uni-heidelberg.de>




More information about the Dune mailing list