Combining parallel result output files

Post processing utility for Elmer
Post Reply
Ric.E
Posts: 8
Joined: 01 Dec 2011, 00:55
Antispam: Yes

Combining parallel result output files

Post by Ric.E »

Hi,

I have managed to run my simulation in parallel and recombined the case.ep.i files and viewed the result in ElmerPost, but I can't figure out how to combine the results files. I added in the solver: Output File = case.result and it has produce case.result.i files but I can't see what to do with them. I want to use them to import the results into Garfield++. Previously I have been doing this with single processing and it has worked fine. Can anyone suggest a way to do this?

Many Thanks
Richard Eggleston
raback
Site Admin
Posts: 4826
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Combining parallel result output files

Post by raback »

Hi Richard

Googling around it seems that Garfield++ includes some interfacing to Elmer.
http://garfieldpp.web.cern.ch/garfieldpp/about/

The documentation of Garfield++ is sparse and I could not find any information regarding the Elmer interface. However, it seems reasonable to assume that the interface is then made for serial runs only. The .result file of Elmer is mainly intended for restarts with the same mesh and no unification feature currently exists. How to optimally do this is depends at least as much on Garfield++ than it does on Elmer. Is Garfield++ in charge of the mesh generation and hence would know the initial global numbering?

-Peter
Ric.E
Posts: 8
Joined: 01 Dec 2011, 00:55
Antispam: Yes

Re: Combining parallel result output files

Post by Ric.E »

Hi Peter,

Thanks for your reply. I have already used the serial output from Elmer in Garfield without problem - it takes the mesh.* files and the case.result file as input. I was hoping to speed up the process by using the parallel results file(s) as my mesh is quite complicated and I'm trying to make it quite fine (not coarse). It therefore takes a long time in the solver running serial. But if this isn't implemented then I'll just have to be patient with the serial solver.

Thanks again
Richard
Post Reply