|
[Sponsors] |
January 18, 2002, 00:34 |
current status of parallel implementation?
|
#1 |
Guest
Posts: n/a
|
Checking in with PHOENICS to learn:
1) Which version of Linux are you currently compiling for, especially the distributed memory parallel version (x86 compatible hardware)? 2) Which version of MPI, LAM-MPI or MPICH are you compiling for (binaries not backward compatible it seems)? 3) Is a Linux F90 compiler supported (note that the new, fast Intel F90 & 95, and C compiler is FREE)? 4) What changes have been made to the parallel version to increase the ability of the user to 'transparently' insert ground coding? By this I mean without regard to the details of how the parallelization is being done. Is there more complete documentation available for this? Thank you! |
|
January 21, 2002, 11:10 |
Free Linux F90 compiler?
|
#2 |
Guest
Posts: n/a
|
Hi!
Sorry I cannot answer any of your questions, but I'd love to know about the "FREE" Fortran compiler for Linux that you mention. Who makes it and where can I find it? I checked Intel's page on their compiler for Linux version 5.0, but it didn't look free to me... Thanks, Olof |
|
January 22, 2002, 02:07 |
Re: Free Linux F90 compiler?
|
#3 |
Guest
Posts: n/a
|
Olof:
There is a software distribution option called, "noncommercial, unsupported" that is not for commercial use (under the 'evaluation copy' link). As most of my applications are for research use, this is not an issue for me. The unsupported part simply means that one can't use their fast response tech support, but the man pages are obviously there. I guess I overlooked the fact that many participants of CFD-Online are in business....and so for them is not free. I found out about it by following the comp.lang.fortran news group. There have been a number of discussions about the merits of all the available compilers, compiler flags, hardware issues, benchmarks, etc. We installed it on an AMD box and it works great so far. And the available benchmarks (which I won't discuss or respond to here) make it very attractive compared to the other commonly used compilers. I hope this helps, gb |
|
January 22, 2002, 05:41 |
Re: current status of parallel implementation?
|
#4 |
Guest
Posts: n/a
|
Dear George
In response to your questions (1) Red Hat 6.2 (Kernel 2.2.12-20smp). (2) MPICH 1.2 (3) I don't know, I use FORTRAN 77 (some subs in C). (4) We have Par-PHOENICS ver 3.2 and we only have the TR316/PAR as a "reference manual" (14 single face pages!!) from August 1999. I don't know if CHAM has released any new TR for parallel version. We have done some work with Par-GROUND and we can share experiences if you want. Regards Kike |
|
January 22, 2002, 06:33 |
Re: current status of parallel implementation?
|
#5 |
Guest
Posts: n/a
|
The new Parallel-PHOENICS provides more functionalities, including: 1-D, 2-D and 3-D domain decompositions, in any x,y,z directions, and xy,xz,yz planes respectively. The domain decomposition is performed automatically, using an algorithm, for achieving the best load balancing. It is available on Win2000/NT SMP and clusters, as well as LINUX clusters. The MPI used for LINUX is LAM-6.5.4, and for NT is MPICH.NT.1.2.1. The version of LINUX used for the latest port is RedHat7.0. The compilers used and supported are the GNU gcc, and g77, supplied with the RH installation. No changes have been made to the parallel version to increase the ability of the user to 'transparently' insert ground coding as yet ! Documentation is also to be updated.
|
|
January 22, 2002, 17:13 |
thanks- very helpful
|
#6 |
Guest
Posts: n/a
|
Kike & Nikos:
Thank you both for the prompt response. My beowulf cluster is running RH 5.1 and a VERY old version of LAM-MPI (6.1). I desperetly need to upgrade my whole system- wipe the slate clean so to speak- but wanted to see where PHOENICS was at if I did. Glad to hear that CHAM is staying current. I've always had good, clear advice in my past experiences with Nikos. Thank you both again, this kind of interchange is what makes CFD-Online great! |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Parallel implementation of k-e model | Niels | Main CFD Forum | 6 | June 27, 2008 15:31 |
diff. beetwen current time and current time step | Tomik | FLUENT | 0 | March 3, 2008 08:00 |
OpenFOAM Debian packaging current status problems and TODOs | oseen | OpenFOAM Installation | 9 | August 26, 2007 14:50 |
VOF status | Paul | Main CFD Forum | 6 | June 30, 2003 21:26 |
UDFs Parallel Implementation Rules | Greg Perkins | FLUENT | 0 | February 4, 2001 06:59 |