[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Freepooma-devel] Hidroforge
From: |
Richard Guenther |
Subject: |
Re: [Freepooma-devel] Hidroforge |
Date: |
Fri, 10 Mar 2006 20:55:14 +0100 (CET) |
On Sat, 4 Mar 2006 address@hidden wrote:
Dear developers of freepooma.
Hello Tenerife ;)
First let me apologize for not answering earlier - I had to think some
time on the suggestions I'd like to make. Now, I definitely applaud
your move in creating a free project around hydrodynamic simulation
technology which will be more in spirit of Open Source projects than
what we have now in the Astrophysics community (I'm thinking of a few
tools available to the general public, but still developed/improved
independently of each other, like zeus, nirvana or flash). I also
see a similar project, though not C++ centric, namely the Cactus code.
Your incentive to use FreePOOMA as a basis for a C++ hydrodynamic
code repository is probably the best you can do (apart from starting
from scratch) - at the time I did search for a basis to do hydro with
C++ there was no better tool/library available that allowed parallel
operation. Still FreePOOMA is not perfect and you will experience its
rough edges, more or less, depending on the specific use. I used
a modified FreePOOMA to implement a basic Nd HD code based on the
zeus numeric scheme, and a collegue of mine has added MHD support on
top of that. I know of two other people exploring the possibility to
add radiative transfer to the supported equations.
I will offer you help as far as possible and am currently trying to
put our FreePOOMA modifications on a branch, so they are publically
visible. As I'm now in the compiler business and no longer actively
doing simulations, my focus has somewhat changed in the last months,
so a lot of FreePOOMA improvement projects/ideas have stalled.
Following links on the page below you might find more information
on FreePOOMA and TraMP, the hydro code developed. Also on
http://www.suse.de/~rguenther/tramp3d/ you can see compiler performance
monitored for a FreePOOMA application - one rough edge of FreePOOMA,
requiring appropriately optimizing compilers (gcc 4.1.0 is a good
choice, the FedoraCode 5 edition includes OpenMP support).
With best regards,
Richard.
--
Richard Guenther <richard dot guenther at uni-tuebingen dot de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/