|
[Sponsors] |
November 15, 2005, 12:48 |
CFD Lab
|
#1 |
Guest
Posts: n/a
|
Hi all:
I am ecuadorian, and my university would like to set a little CFD lab, my question is: What kind of computer and compiler should they adquire? Thanks in advance, Ruben Paredes |
|
November 15, 2005, 13:40 |
Re: CFD Lab
|
#2 |
Guest
Posts: n/a
|
It depends on how much you intend to spend and what kind of application you'd like do develop.
If you don't have much money to spend, go through a Beowulf cluster built with pieces easily found at a normal computer market and develop your CFD applications based on message passing parallelism. With a bit more money you can try to get a supercluster like SGI Altix or Cray XD1. The prices of these machines can be scaled up according to the number of CPUs, memory, etc...(for a SGI Altix, I guess, the prince begin in 90 mil US$ up to infinity ;o) For example, on my lab we have a very modest SGI Altix 350 with 16 cpu's and 28 Gb of memory, with this machine we have been able to perform simulations with millions of equations in a reasonable time. Cheers Renato. note: About the compiler you'll need Fortran, C/C++ and MPI. |
|
November 16, 2005, 10:32 |
Re: CFD Lab
|
#3 |
Guest
Posts: n/a
|
Hi Ruben,
for smaller student projects one or more PCs with 2-4GB RAM and fast CPUs (3GHz) should be enough. Parallel processing licenses might be good. Don´t let you dicourage by Renato. Greetings Tilman |
|
November 16, 2005, 16:48 |
Re: CFD Lab
|
#4 |
Guest
Posts: n/a
|
Thanks Tilman and Renato
Don´t worry about the Renato message, he only wanna help. I am pretty interested in CFD, and in my University is something new. The basic idea is to get a postgraduate degree and to come back to Ecuador with some expirience in CFD modelling. At the same time, they(ESPOL University) adquiere some equipment and conform a work team. Best Regards, Ruben |
|
November 17, 2005, 14:05 |
Re: CFD Lab
|
#5 |
Guest
Posts: n/a
|
There are a lot of academic CFD labs around the world. If you are serious about this, I would look around on the internet (search University websites) and see what other guys have set up. Then, if you find something close to your own interest, contact those people directly. I am sure most of them will be glad to help you out with some information on their systems, cost, performance, best compilers, and so on... This way you'll get a much better overview and more reliable expertise than you'll find in any forum... including this one.
|
|
November 18, 2005, 06:53 |
Re: CFD Lab
|
#6 |
Guest
Posts: n/a
|
In that case, it would be easier - and cheaper - to begin with a small beowulf cluster with linux and MPICH, which is a free version of mpi. As you start getting some results you can expand your cluster by just adding new PCs to it. I'm brazilian, and in my university we started a cluster with 3 PCs in 2002, and now we have 10 pcs (PIV - 3 GHz, 1.5 Gb RAM each). Of course it is less powerful than commercial clusters, but it is much cheaper to start with and to upgrade.
-Márcio |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Setting up a new CFD lab | asitkm | Hardware | 1 | January 8, 2011 14:55 |
ASME CFD Symposium, Atlanta, July 2001 | Chris R. Kleijn | Main CFD Forum | 0 | September 13, 2000 04:48 |
ASME CFD Symposium, Atlanta, July 2001 | Chris R. Kleijn | Main CFD Forum | 0 | August 21, 2000 04:49 |
Since Last June | John C. Chien | Main CFD Forum | 3 | July 12, 1999 09:38 |
Which is better to develop in-house CFD code or to buy a available CFD package. | Tareq Al-shaalan | Main CFD Forum | 10 | June 12, 1999 23:27 |