Performance using parallel solver

rodrigovime · November 11, 2024, 15:25

Hello everyone,
I have a quick question about the performance of parallel solver in ANSYS FLUENT.
My current work station and academic license enables me to run a simulation using 56 logical units. I have 8 FLUENT cases and I want them to run in the same workbench.
What will be faster? To use 56 logical units in every case and actualize the project from Workbench, or assign 7 logical units to each project?
This may be a dumb question, but thank you for your answer anyway.

CFDKareem · November 11, 2024, 16:36

This will all depend on your specific model and how well it runs with parallel cores. For my work, I almost always find it more efficient run 8 models in parallel then doing each in serial with all cores.

MKuhn · November 12, 2024, 03:39

I agree with CFDKareem. One reason is that you need also time for data change between the cores. There seems to be a point that the solver then needs more time to communicate between the cores, as you gained by increasing the cores.

rodrigovime · November 12, 2024, 13:29

Thank you both for your answers. It do it in parallel with less cores then.

November 11, 2024, 15:25	Performance using parallel solver	#1
rodrigovime New Member Rodrigo Villarreal Join Date: Nov 2016 Posts: 29 Rep Power: 10	Hello everyone, I have a quick question about the performance of parallel solver in ANSYS FLUENT. My current work station and academic license enables me to run a simulation using 56 logical units. I have 8 FLUENT cases and I want them to run in the same workbench. What will be faster? To use 56 logical units in every case and actualize the project from Workbench, or assign 7 logical units to each project? This may be a dumb question, but thank you for your answer anyway.

November 11, 2024, 16:36		#2
CFDKareem Senior Member Kareem Join Date: Nov 2022 Location: New York Posts: 125 Rep Power: 5	This will all depend on your specific model and how well it runs with parallel cores. For my work, I almost always find it more efficient run 8 models in parallel then doing each in serial with all cores. MKuhn likes this. __________________ Please like the answer if it helped! Video Tutorials and Tips: https://www.youtube.com/@cfdkareem/featured

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
foam-extend-4.1 release	hjasak	OpenFOAM Announcements from Other Sources	19	July 16, 2021 06:02
[OpenFOAM.org] Custom solver not running in parallel	syavash	OpenFOAM Installation	12	June 6, 2019 16:02
Hybrid discretisation - blend factor	gcoopermax	CFX	5	September 23, 2016 09:05
"Define_Profile" UDF for parallel solver	Antoine	Fluent UDF and Scheme Programming	9	February 29, 2016 07:09
rhoCentralFoam solver with Slip BCs fails in Parallel Only	JLight	OpenFOAM Running, Solving & CFD	2	October 11, 2012 22:08

November 12, 2024, 03:39		#3
MKuhn Senior Member Moritz Kuhn Join Date: Apr 2010 Location: Germany, Dresden Posts: 218 Rep Power: 17	I agree with CFDKareem. One reason is that you need also time for data change between the cores. There seems to be a point that the solver then needs more time to communicate between the cores, as you gained by increasing the cores.

November 12, 2024, 13:29		#4
rodrigovime New Member Rodrigo Villarreal Join Date: Nov 2016 Posts: 29 Rep Power: 10	Thank you both for your answers. It do it in parallel with less cores then.