OpenFOAM: advice on parallel computing
Hi all,
I'm planning to use OpenFOAM to analyze the following physics: flow in porous zone, probably with rotating mesh (fans), conjugate heat transfer, radiation and combustion. I will deal with a computational domain with 10-20 millions of cells, approximately. As workstation, my intention is to have a total amount of number 2 intel XEON processor with 128GbRAM. What I would like to know is: is it better to have two processor with less total number of cores(8), but higher frequency(3.5GHz)? Or to have a lower frequency rate (2.2Ghz) but higher number of cores (20)? I know that it sounds like:"bird in the hand might be worth two in the bush", but if some user can share his/her experienxe, it will be of help. Thanks. Regards. Michele |
The high core count Xeon processors are pretty much a waste of money for CFD workloads. 20 CPU cores will be severely bottle-necked by only 4 memory channels.
The most important thing is to maximize memory throughput first, so choose a Xeon processor that supports DDR4-2400. The rest depends on your budget. The best Xeon processor for CFD right now is the Xeon E5-2687w v4. I had to cheap out a bit on our last workstation so it only has 2 E5-2650v4. But despite the low budget and clock speed it is still considerably faster than our previously fastest workstation. |
Thank you very much for the note.
Regards. Michele Sent from my ASUS_X008D using CFD Online Forum mobile app |
All times are GMT -4. The time now is 05:35. |