[cegan09]

I didn't want to clutter the results thread, but I cannot get the benchmark to run. Before I waste more time I just wanted to check if I'm doing anything dumb.


It looks like the benchmark in this post:
OpenFOAM benchmarks on various hardware
might be intended to be used with 2112 and not openFOAM 10?

So far I've managed to catch and fix all the errors through meshing while using 10, but I'm still getting errors during the actual runs. I just wanted to see if I was being dumb and not seeing a version of the pre-made scripts that were meant for 10.


alternatively, is it bad to install 2112 alongside 10? I know there are some differences. I have no good reason for having picked 10 other than it was the first one I managed to install when I started trying to learn openFOAM.

[wkernkamp]

You can definitely install in parallel. On linux, the set-up happens when you "source OpenFOAM-v2112/etc/bashrc. A similar setup happens for OpenFOAM 10. Only after this set-up you can run the programs. Sometimes, the run files are defined permanently by sourcing an OpenFOAM install when you login.


If you get a version of the benchmark to run on OpenFOAM 10, please post it to the thread for others to use. Also, you should post here any difficulties you experience with benchmarking OpenFOAM 10 (so that we can help).

[cegan09]

Thank you. I'll try to do both, meaning get 2112 installed and the benchmark running, along with post what I've found with getting the benchmark to run in 10.

[cegan09]

I got 2112 installed, and the benchmark runs out of the box. That's good. I posted results in the benchmark thread. I've not yet proved that I haven't broken my other simulation work, I have to check that those still source to openFoam 10 until I prove what I need to change to run them on either.


I've not yet gotten it to work fully on 10, but I'm closer? Here are my notes on what I've had to do so far.
In the Allmesh files (serial and parallel), surfaceFeatureExtract needs to be changed to surfaceFeatures
I had to add the surfaceFeaturesDict file to the system folder to go with the function change
In the meshQualityDict file I had to change the include to be "caseDicts/mesh/generation/meshQualityDict" in order for it to source correctly



Doing the above the benchmark will mesh all the cases successfully.


simpleFoam still doesn't run. It's mad about streamlines and I've not yet sorted out exactly what it wants different.
From log.simplefoam

[8] 
[8] 
[8] --> FOAM FATAL ERROR: 
[8] Unknown function type streamLine

Valid functions are : 

60
(

streamFunction
streamlines


)
[8] 
[8] 
[8]     From function static Foam::autoPtr<Foam::functionObject> Foam::functionObject::New(const Foam::word&, const Foam::Time&, const Foam::dictionary&)
[8]     in file db/functionObjects/functionObject/functionObject.C at line 101.
[8]

I snipped all but two of the functions from that list for the sake of keeping the post from becoming a wall of text.

I've tried editing the streamLines file in system to have the type streamlines, and I get a new error. This is where I've gotten to, haven't fully digested this one to find the solution yet. I'll keep poking at it in my spare time.

[0] 
[0] 
[0] --> FOAM FATAL IO ERROR: 
[0] keyword direction is undefined in dictionary "/home/user/benchmarks/dev/3/run_32/system/controlDict/functions/streamLines"
[0] 
[0] file: /home/user/benchmarks/dev/3/run_32/system/controlDict/functions/streamLines from line 12 to line 52.
[0] 
[0]     From function const Foam::entry& Foam::dictionary::lookupEntry(const Foam::word&, bool, bool) const
[0]     in file db/dictionary/dictionary.C at line 831.
[0]

[wkernkamp]

I was thinking about the ongoing problem of them (the OpenFOAM wizards) changing the rules on us. Maybe download the entire motorBike case from the tutorial and modifying that working case to end up with the benchmark grid size.