|
[Sponsors] |
April 9, 2013, 02:16 |
Machine for FLUENT multiphase
|
#1 |
New Member
Join Date: Mar 2013
Posts: 13
Rep Power: 13 |
Hello
As far as I known ones of the major factors are memory frequency and ratio of memory channels to number of cores. Ansys HPC license variants of interest for me are 8, 32 and 128 cores. 2 kinds of problems: 1) - 3d double precision - 5 mln cells - transient - kw-sst - Euler multiphase - timestep 5*10e-4 - flow time up to a minute - Calculation time not more than a week 2) - 3ddp - 30 mln cells - transient - kw-sst - VOF geo-reconstruct - timestep 10e-4 - flow time up to a minute - Calc time not more than 2 weeks Which hardware for each problem would you recommend? What results in faster solution and what is more expensive cluster with infiniband or XEONs on one board? Could you recommend some distributors of such machines? |
|
April 9, 2013, 03:42 |
|
#2 |
Senior Member
Charles
Join Date: Apr 2009
Posts: 185
Rep Power: 18 |
The first question is how budget sensitive you are? If you can afford a 128 core HPC license and the 16 x 8-core Xeon CPUs plus the infiniband network, then that is going to be by far your best system, but it's going to set you back a bit. The small problem size won't benefit from such a big system, but the larger one will. The smaller problem will fit nicely with the 32 core HPC license, and then you would probably do best with two dual CPU Xeon's networked together directly.
There are other factors that come into play, such as where the hardware will sit. You can't have a blade or rackmount server in the same office space as people, due to the noise, for example. But really, you have to try & estimate the performance and then build up a spreadsheet to figure out what will be best value for money. elnexus.com is a handy site for estimating costs. |
|
April 9, 2013, 05:20 |
|
#3 |
New Member
Join Date: Mar 2013
Posts: 13
Rep Power: 13 |
Thank you a lot, CapSizer! I'll definitely try this site.
|
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Difference of multicomponent and multiphase homogenous flows | Luk_Fiz | CFX | 11 | April 4, 2013 05:29 |
Use data from multiphase into DO radiation | yashmash | FLUENT | 0 | October 7, 2011 03:57 |
multiphase multicomponent physics | ckleanth | CFX | 3 | June 4, 2009 20:15 |
Distributed Parallel on dual core remote machine | Justin | CFX | 1 | February 3, 2008 17:23 |
PC vs. Workstation | Tim Franke | Main CFD Forum | 5 | September 29, 1999 15:01 |