set 4 servers to calculate together (in parallel)

losiola · January 23, 2020, 05:45

Hello guys ,
i have a questions and i hope that you can help me with it .
in our lab we have 4 servers poweredge runing windows server2012 and with valid ansys 19 licence for all of them , we usually use each server to run it's independant simulation (in parallel using server cores) but i want to know if it is possible that i connect all 4 of them to make sort of mini cluster and run 1 simulation that gonna be calculated on all 4 of them together

please if anyone knows how can i do that please explain it to me ?

Many thanks

evcelica · January 23, 2020, 18:14

Yes, you can do this.
Do you have enough HPC licenses to distribute to enough cores to make this worth while?

You would want some sort of fast low latency interconnect between the machines.

Infiniband is the standard. Maybe 10gigE, but I have no experience with that, as I use infiniband.

You would have to install MPI on the machines, and set them up for distributed computing. Instructions for how to do this are in the ANSYS documentation.

The version of MPI (Intel, IBM, or MS-MPI) will depend on which network cards and drivers you use. Most newer Infiniband drivers for windows only support MS-MPI.

This will most likely be a fair amount of work to get this all set up and working properly if you are not familiar with such things, so just don't expect it to be plug and play.

January 23, 2020, 05:45	set 4 servers to calculate together (in parallel)	#1
losiola Member Os Join Date: Jun 2017 Posts: 80 Rep Power: 9	Hello guys , i have a questions and i hope that you can help me with it . in our lab we have 4 servers poweredge runing windows server2012 and with valid ansys 19 licence for all of them , we usually use each server to run it's independant simulation (in parallel using server cores) but i want to know if it is possible that i connect all 4 of them to make sort of mini cluster and run 1 simulation that gonna be calculated on all 4 of them together please if anyone knows how can i do that please explain it to me ? Many thanks

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Error running simpleFoam in parallel	Yuby	OpenFOAM Running, Solving & CFD	14	October 7, 2021 04:38
The problem when i use parallel computation for mesh deforming.	Hiroaki Sumikawa	OpenFOAM Running, Solving & CFD	0	November 20, 2018 02:58
does Hyperthreading affect the application of UDF?	SJSW	Fluent UDF and Scheme Programming	11	October 10, 2018 22:28
Using #codeStream to set initial field for parallel run	Hisham	OpenFOAM Running, Solving & CFD	7	June 25, 2012 09:48
Help with GNUPlot	Renato.	Main CFD Forum	6	June 6, 2007 19:51

January 23, 2020, 18:14		#2
evcelica Senior Member Erik Join Date: Feb 2011 Location: Earth (Land portion) Posts: 1,184 Rep Power: 23	Yes, you can do this. Do you have enough HPC licenses to distribute to enough cores to make this worth while? You would want some sort of fast low latency interconnect between the machines. Infiniband is the standard. Maybe 10gigE, but I have no experience with that, as I use infiniband. You would have to install MPI on the machines, and set them up for distributed computing. Instructions for how to do this are in the ANSYS documentation. The version of MPI (Intel, IBM, or MS-MPI) will depend on which network cards and drivers you use. Most newer Infiniband drivers for windows only support MS-MPI. This will most likely be a fair amount of work to get this all set up and working properly if you are not familiar with such things, so just don't expect it to be plug and play.