[Beowulf] Beowulf, Gentoo and Navier Stokes solvers

Jim Lux James.P.Lux at jpl.nasa.gov
Mon Feb 6 14:53:13 PST 2006


At 09:00 AM 2/6/2006, Alan Jones wrote:
>Hi All,
>
>I'll start by introducing myself being my first post on the list. As the 
>header suggests my name is Alan - I'm head of R&D at a production (visual 
>effects) house in London.
>
>I'm currently looking into developing a cluster to perform fluid 
>simulation and google, while providing plenty of information, hasn't shown 
>what I'm after.

Hmm.. a lot of the CFD codes I've seen tend to be very application 
specific.  With all finite element codes, specifying the model and the 
simulation environment (grid style, step sizes, time steps, etc.) is most 
of the battle, since the actual grunt work of running the model is actually 
pretty straight forward from a coding standpoint (although it consumes the 
vast majority of the processor cycles).

But, then, this is no different than CGI.. it's specifying the model that's 
the hard part, not firing off the rendering.

So, you might want to poke at all those CFD codes out there and see which 
ones are suited to what YOU are trying to model.  Supersonic flow in a gas 
turbine isn't likely to be handled by the same code as modeling the flow of 
cooling air in a cluster server room or, for that matter, tornado simulator 
dynamics (which was MY foray into CFD).

While this list isn't necessarily a great place to ask about the details of 
CFD codes (maybe it is, though), I'll bet that the folks on the list, if 
they don't know, they do know someone who does, and will get you started in 
the right direction.



James Lux, P.E.
Spacecraft Radio Frequency Subsystems Group
Flight Communications Systems Section
Jet Propulsion Laboratory, Mail Stop 161-213
4800 Oak Grove Drive
Pasadena CA 91109
tel: (818)354-2075
fax: (818)393-6875





More information about the Beowulf mailing list