CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Siemens

Parallel simulation on STAR-CCM+ 11.06.010-R8

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   June 8, 2017, 04:29
Default Parallel simulation on STAR-CCM+ 11.06.010-R8
  #1
New Member
 
Hossein M.Ghasemi
Join Date: Jun 2017
Posts: 1
Rep Power: 0
Hosseinmgh_chemeng is on a distinguished road
Dear colleagues and friends;
I want to simulate 90,000 particles with 10mm and 15 mm diameter in a container.
I decide to run this simulation on a PC with 64 core i7 Intel Xeon 1.6 Ghz. on windows server 2016 in parallel mode. I set the simulation "parallel on localhost". But when I adjust the simulation on parallel mode, it is too slow
to run relative to the series mode.
However, I tested this simulation on a quad Core i7 3.4 Ghz. cores in parallel mode and the simulation became much faster than previous one.

- Could any one help me to make the simulation faster on the 64 cores PC ?

BR
Hosseinmgh_chemeng is offline   Reply With Quote

Old   August 28, 2017, 05:54
Default
  #2
New Member
 
Danilo Ciliberti
Join Date: Aug 2017
Location: Naples, Italy
Posts: 1
Rep Power: 0
engdancili is on a distinguished road
Hello! Probably you can't get a faster simulation, and here I explain why. Usually it is a matter of mesh size: the STAR-CCM+ manual advices to use a CPU every 250000 cells. Further increase in CPUs number may lead to bottlenecks and the simulation will slow down, because the time needed to exchange information among CPUs or cores is longer than the time used to solve the equations.

Second: recently, Siemens has issued on the Steve Portal an article where it is stated that hyper threading technology may increase the computational time, since mesh generation and flow solving almost use the 100% of a CPU thread, so that the task switching on a physical core due to hyper threading could slow down the simulation, although hyper threading may speed up other phases like post-processing. Are these 64 cores physical? Or are you on a machine with 2 CPUs with 16 physical core each and 32 logical cores each?

Third and perhaps most important: the Intel Xeon you are using is just 1.6 GHz per core, a computational power far below the laptop where I am writing... To solve your issue try to use less cores according to your mesh size. However, since the CPU clock frequency is slow, do not expect much improvement, because the parallelization will help you on big size. You also did not mention about the RAM of the workstation.
engdancili is offline   Reply With Quote

Old   August 30, 2017, 08:21
Default
  #3
Senior Member
 
Ping
Join Date: Mar 2009
Posts: 290
Rep Power: 12
ping is on a distinguished road
lagrangian solving is very different to eulerian flow solving with one of the main issues being the partitioning required by the particle solver changes with time so you might find that dynamic partitioning helps which is far more important for unstready cases
i would also search on the steve support portal for other speed up tips
ping is offline   Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Engine Simulation in parallel Peter_600 OpenFOAM 0 July 26, 2012 06:02
Compressor Simulation in Star CCM+ chili023 STAR-CCM+ 13 August 11, 2011 02:22
Fluent Vs Star CCM firda Main CFD Forum 3 February 26, 2011 03:51
A genearl question on STAR CCM nstar STAR-CCM+ 5 June 24, 2009 09:39
Trimmed cell and embedded refinement mesh conversion issues michele OpenFOAM Other Meshers: ICEM, Star, Ansys, Pointwise, GridPro, Ansa, ... 2 July 15, 2005 04:15


All times are GMT -4. The time now is 00:07.