parallel processers Scaling for CFD

rishik686 · December 9, 2022, 05:15

Hi there,
I have a query regarding the scaling of parallel processors Simulation for CFD.
what is the scaling law for parallel processing simulation for CFD based on domain decomposition?

I have performed scaling in Openfoam, but need some validation support for it.

LuckyTran · December 9, 2022, 09:12

In many practical cases, it goes as bad as sqrt(N). If it starts to go even worse than I stopped adding additional threads and stick with sqrt(N).

flotus1 · December 9, 2022, 10:40

It depends...
General recommendations for CFD hardware [WIP]

You can get pretty much anything, depending on how you do your analysis. Linear scaling, superlinear scaling, scaling that could be fitted with Amdahl's law, no scaling at all. negative scaling...

Probably a good idea to start with terminology. Are you interested in strong or weak scaling? Intra-node or inter-node?

rishik686 · December 10, 2022, 06:37

Quote:

Originally Posted by LuckyTran

In many practical cases, it goes as bad as sqrt(N). If it starts to go even worse than I stopped adding additional threads and stick with sqrt(N).

Thanks lucky for the response.
i will apply this

rishik686 · December 10, 2022, 06:39

Quote:

Originally Posted by flotus1

It depends...
General recommendations for CFD hardware [WIP]

You can get pretty much anything, depending on how you do your analysis. Linear scaling, superlinear scaling, scaling that could be fitted with Amdahl's law, no scaling at all. negative scaling...

Probably a good idea to start with terminology. Are you interested in strong or weak scaling? Intra-node or inter-node?

I will try linear scaling analysis.
Thanks for the response

December 9, 2022, 05:15	parallel processers Scaling for CFD	#1
rishik686 New Member Rishi Kumar Join Date: Jun 2022 Posts: 20 Rep Power: 4	Hi there, I have a query regarding the scaling of parallel processors Simulation for CFD. what is the scaling law for parallel processing simulation for CFD based on domain decomposition? I have performed scaling in Openfoam, but need some validation support for it.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
parallel run in foam extend	CRI_CFD	OpenFOAM CC Toolkits for Fluid-Structure Interaction	5	March 28, 2024 09:36
simpleFoam parallel	AndrewMortimer	OpenFOAM Running, Solving & CFD	12	August 7, 2015 19:45
simpleFoam in parallel issue	plucas	OpenFOAM Running, Solving & CFD	3	July 17, 2013 12:30
parallel scaling with sliding interfaces	laf	FLUENT	1	March 23, 2007 12:50
Parallel Computing Classes at San Diego Supercomputer Center Jan. 20-22	Amitava Majumdar	Main CFD Forum	0	January 5, 1999 13:00

December 9, 2022, 09:12		#2
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,761 Rep Power: 66	In many practical cases, it goes as bad as sqrt(N). If it starts to go even worse than I stopped adding additional threads and stick with sqrt(N).

December 9, 2022, 10:40		#3
flotus1 Super Moderator Alex Join Date: Jun 2012 Location: Germany Posts: 3,427 Rep Power: 49	It depends... General recommendations for CFD hardware [WIP] You can get pretty much anything, depending on how you do your analysis. Linear scaling, superlinear scaling, scaling that could be fitted with Amdahl's law, no scaling at all. negative scaling... Probably a good idea to start with terminology. Are you interested in strong or weak scaling? Intra-node or inter-node?