Windows (x64) Binaries for OpenFOAM-2.1.x using OpenMPI
 

Hi
I have uploaded my binaries for OpenFOAM-2.1.x here.

This was compiled using the instructions outlined by Symscape and provided here.

The only difference between Symscape's is this is compiled against OpenMPI and does not use MSMPI.

I have not tested this extensively and hope to get some feedback on this.

Hope to document the process so enable recreating this with ease.

I will push the source soon (I have a crawling connection).

Cheers!
JLight

Very cool! I am going to try your binary over the weekend. Also, I'm interested in this documentation. Was this compiled on linux with MinGW?

Yes, it was compiled on linux with MinGW-w64.

Dude you are awesome, I was just going to try and tackle this binary build using the instructions from symscape... and then I found this... great work!

You said you were looking for some feedback. I have the first bit. It seems like the sourceforge binaries might not include everything needed to run. I receive this attached error when trying to run blockMesh...

Attachment 17385

I may have missed a step, I am most definitely not a linux expert, but it appears some of the MinGW-w64 libraries may not have been included (or couldn't be) as per symscape's instructions.

If you can't update the install on sourceforge, do you know where I can get the required dll files? Do I have to build MinGW myself? Thanks! And great work!

Hi thanks for this. I think I had zlib installed on my Windows from some previous work.

The easiest workaround is for me to upload the dll (Tha I had cross compiled) to sourceforge. I have updated the Readme. Please try it and tell me if it works.

Glad to know you like this.

Cheers!

NJ

Hi All
Just a quick update.

I have pushed the source of OpenFOAM with the Symscape patch and brought it up to commit d957081e1400ad2fd88fefcf7db3599dd66b0ddf (2 Dec 2012).

The recompiled binaries is upload to SourceForge this very moment.

I am in the process of writing out the steps needed to do it yourselves and is located here.

Cheers!

NJ

Awesome, thank you for taking the time to do this. I have tested blockMesh with the included zlib.dll file and it now runs without a hitch. I'll let you know when I have tested it more extensively or find anything amiss. Thanks again!

This is really nice. I just downloaded and tested the install...so far its working great. I'm interested in the notes on how to compile...since this is what I would like to do over the holiday with some custom solvers. Thanks again and excellent work.

I second Dan's motion, I have some custom solvers as well that would be nice to cross compile. If and when you have the time to finish the wiki page. I will gladly give it a run through to see if everything is clear and concise and to make sure I'm able to successfully compile. Thanks again!

Update
 

Quick Update

I just updated and pushed the source that bring it up to commit c79762488bb97ec8cb786ecff81560e3409762f0 (02-Jan-2013).

The work on the wiki is going a bit slow, I have done about 30%. Hoping to complete it all by the end of the week.

Cheers!

NJ

The wiki is looking good, keep up the good work. Be sure to post back when you're finished, I'd be happy to take it for a test drive. :cool:

parallel problem
 

Hi,
I installed Openmpi and unarchive openfoam in c:\openfoam

I excuted dosmode.bat and try to run tutorial

so
blockmesh ==> work
icofoam ==> work

but when I try to calculate tutorial as parallel. this error is appeard.

can you help me this problem

Try use application name wit extension :

I got It!

thank you!!!

Kuo
 

Hi Everyone,

I used to use the windows version openfoam v2.1 on my desktop.

I have a problem about merge the mesh from a parallel job. I try to use the command "reconstructPar" to merge the results of those processors' data including the meshes. I think I should use "reconstructParMesh", but I can not find it in the bin folder.

could you do me a favor ? thinks.

Hi Joseph,

Great job done, its really working fine. Just a small query, once the calculation is finished.I used foamtoVTK utility in the dos prompt to get the data readable in Paraview. Because normally in linux one uses the internal utility paraFoam to open the calculated case in the current working case directory.

Is there alternative approach to directly post-process the calculation in Paraview.

Thanks.

incomplete set of binaries
 

There is no "parafoam.exe" among your binaries.
Moreover, "ParaView" package directories do not exist at all.
Hence, all 3 corresponding environmental variables in setvar.bat to run DOS_MODE.bat remain undefined.

With your binaries a user can run CFD but for post-processing he needs other piece of software. Which is not the case of the Linux based distribution of OF-2.1.x or available Win64 binaries of OF-1.5.x and OF-1.6.x and OF-1.7.x -- all they are freeware.
So, my private opinion is that without "parafoam" any OF distribution does not look complete.
Cheers.

Lucky for us, Paraview has a native Windows install available from their website which we can use to open OF jobs! This isn't the case with OpenFOAM, and is why people like JLight are awesome. :cool:

Cheers

reconstructParMesh
 

Hi Nishit,

I could not find a reconstructParMesh option in your Openfoam 2.1x Win 64 version.
Do you plan to update it with reconstructParMesh feature?
Basically, I could not run any parallel simplefoam simulations after running snappyhexmesh using mpirun.

Do you have any advice to fix this issue?

Thanks,
Zsolt