CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Installation on Windows, Mac and other Unsupported Platforms (http://www.cfd-online.com/Forums/openfoam-installation-windows-mac/)
-   -   Patches for OpenFOAM 1.7 on MacOS X (http://www.cfd-online.com/Forums/openfoam-installation-windows-mac/77570-patches-openfoam-1-7-macos-x.html)

gschaider June 27, 2010 18:52

Patches for OpenFOAM 1.7 on MacOS X
 
2 Attachment(s)
Hi!

Attached there are two patches that patch the OpenFOAM and the ThirdParty-directory of the new 1.7-release.

For installation and requirements (case sensitive file system) see for instance http://openfoamwiki.net/index.php/Ho...enFOAM_v16_Mac (if somebody feels like rewriting/adapting that: go ahead)

Some remarks on the patch:
- only the darwinIntel64-architecture has been tested
- the patch assumes that you use gcc 4.5 from MacPorts. If you don't: replace all instances of the form gcc-mp-4.5 with the compiler of your choice
- tecplotWriter does not compile
- as soon as paraview is compiled the PV3FoamReader will compile. It will load as a plugin into paraview. Nevertheless: it will crash paraview as it needs a different version of the libstdc++.6.dylib than the one paraview uses (paraview fails to compile with the same compiler as OF). Only solution: remove the libPV3FoamReader_SM.dylib and use the OF-Reader built into paraview. Alternative would be to compile OF with the v4.2-compiler that comes with MacOS X

Bernhard

podallaire June 27, 2010 22:22

You rock!

Thanks

gschaider June 28, 2010 08:22

Quote:

Originally Posted by podallaire (Post 264742)
You rock!

Thanks

Actually it was quite easy as I had maintained patches of 1.6.x anyway and there were only two changes required

Bernhard

niklas June 28, 2010 08:40

Have you guys tried this?

http://www.parallels.com/uk/products/desktop/

I recently switched to this instead of dealing with the mac-issues.

Since I dont do any serious work on my mac, but only write code and run simple tests I find this
to be an incredible piece of software.

You can download a free trial version for 2 weeks.

podallaire June 28, 2010 09:12

Quote:

Originally Posted by gschaider (Post 264803)
Actually it was quite easy as I had maintained patches of 1.6.x anyway and there were only two changes required

Bernhard

Cool / I'm currently compiling with gcc-4.3, so far so good !


Quote:

Originally Posted by niklas (Post 264805)
Have you guys tried this?

http://www.parallels.com/uk/products/desktop/

I recently switched to this instead of dealing with the mac-issues.

Since I dont do any serious work on my mac, but only write code and run simple tests I find this
to be an incredible piece of software.

You can download a free trial version for 2 weeks.

Yes, I know this software / I use VMware when Windows is needed ... It's good to have a native version of OpenFOAM on the mac, easier to prepare cases before sending to the cluster.

Thanks !

PO

chrisb June 29, 2010 09:19

Working with finkproject compilers
 
Hi,
Just thought I would let everyone know that the patches work as advertised with fink project compilers (gcc45 package).

To do modify the patch file (OpenFOAM-1.7-Mac.patch):
Replace -mp- with -fsf-
Replace 4.3 with 4.5


Chris Butler

leosding July 4, 2010 06:55

Bernhard,
thanks a lot.
Wonderful! The OF1.7 mac port is now ok to run on snow leopard 10.6.3 with gcc4.5 installed by macports.
Because of the error of 'time', the foamToTecplot360 has not been compiled.
The Paraview 3.8 for Mac is now availabe for post-processing. :)

Leo

chrisb July 5, 2010 00:44

Problems OpenFOAM 1.7 mac
 
Hi all,

I have run into some problems with OpenFOAM-1.7 on mac os 10.6.4. The tutorial for nonNewtonianIcoFoam appears to fail (due to a solution singularity).

This at first glance appears to be due to blockMesh failing to produce an adequate mesh (see the attached file). The tutorial correctly executed with the debian 1.7 install (and 1.6 source install on CentOS).

Has anyone else had any problems? I have created meshes which do not contain the arc object without any trouble.

Any feedback would be much appreciated.

Cheers,
Chris Butler

P.S. If this post is in the wrong forum my apologies.

gschaider July 5, 2010 08:47

Quote:

Originally Posted by chrisb (Post 265676)
Hi all,

I have run into some problems with OpenFOAM-1.7 on mac os 10.6.4. The tutorial for nonNewtonianIcoFoam appears to fail (due to a solution singularity).

This at first glance appears to be due to blockMesh failing to produce an adequate mesh (see the attached file). The tutorial correctly executed with the debian 1.7 install (and 1.6 source install on CentOS).

Has anyone else had any problems? I have created meshes which do not contain the arc object without any trouble.

Any feedback would be much appreciated.

Cheers,
Chris Butler

P.S. If this post is in the wrong forum my apologies.

I can reproduce that. Which compiler are you using (I use gcc-4.5. My 1.6/gcc45-installation is broken, so I can't check. I'm suspecting the compiler because it is my impression that he handles FP-operations somewhat differently). If I do blockMesh with 1.6/gcc45 then checkMesh and the solver have no problem, so the problem seems to be located with blockMesh

Bernhard

gschaider July 5, 2010 08:59

Quote:

Originally Posted by gschaider (Post 265750)
I can reproduce that. Which compiler are you using (I use gcc-4.5. My 1.6/gcc45-installation is broken, so I can't check. I'm suspecting the compiler because it is my impression that he handles FP-operations somewhat differently). If I do blockMesh with 1.6/gcc45 then checkMesh and the solver have no problem, so the problem seems to be located with blockMesh

Bernhard

The weirdness continues: if I do the blockMesh with with 1.7/gcc45-debug(!) I have no problem. So it seems that something is going wrong during optimization (in gcc45)

BTW: have a look at the mesh in paraview. No wonder checkMesh doesn't like it.

podallaire July 5, 2010 09:00

I just tested (I have compiled OF-1.7.x with gcc-4.3.4) / no problem here with this tutorial.

gschaider July 5, 2010 13:51

Quote:

Originally Posted by gschaider (Post 265750)
I can reproduce that. Which compiler are you using (I use gcc-4.5. My 1.6/gcc45-installation is broken, so I can't check. I'm suspecting the compiler because it is my impression that he handles FP-operations somewhat differently). If I do blockMesh with 1.6/gcc45 then checkMesh and the solver have no problem, so the problem seems to be located with blockMesh

Bernhard

Update: I repaired my 1.6/gcc-4.5 installation. Now the following happens:

1.6/gcc-4.5-Opt: broken Mesh
1.6/gcc-4.5-Debug: mesh OK
1.6/gcc-4.3-Opt: mesh OK
1.7/gcc-4.5-Opt: mesh broken

So the problem seems to be gcc-4.5 with optimization. The remaining question: does this only happen on a Mac or elsewhere, too?

Bernhard

chrisb July 5, 2010 19:51

gcc 4.4
 
Hi,
I am building OF-1.7 with gcc-4.4 on both OSX and linux, however, I do not have a linux box with gcc above 4.4.4. So if anyone could test this tutorial on linux with gcc-4.5 it would be appreciated.


CB

chrisb July 5, 2010 23:16

Quote:

Originally Posted by chrisb (Post 265836)
Hi,
I am building OF-1.7 with gcc-4.4 on both OSX and linux, however, I do not have a linux box with gcc above 4.4.4. So if anyone could test this tutorial on linux with gcc-4.5 it would be appreciated.


CB

blockMesh (OF-1.7) runs correctly on the (nonNewtonianIcoFoam) tutorial under both OSX and Linux when using gcc-4.4

CB

leosding July 6, 2010 03:29

Quote:

Originally Posted by gschaider (Post 265750)
I can reproduce that. Which compiler are you using (I use gcc-4.5. My 1.6/gcc45-installation is broken, so I can't check. I'm suspecting the compiler because it is my impression that he handles FP-operations somewhat differently). If I do blockMesh with 1.6/gcc45 then checkMesh and the solver have no problem, so the problem seems to be located with blockMesh

Bernhard

Bernhard,
I run the case of coalChemistryFoam/simplifiedSiwek, at time=0.252 it broken, the log reads:
=======================================
Courant Number mean: 0.1533929535 max: 0.6093346791
deltaT = 0.0005
Time = 0.252

Cloud: coalCloud1
Total number of parcels added = 27
Total mass introduced = 0.0001
Current number of parcels = 27
Current mass in system = 7.634816181e-05
Mass transfer phase change = 2.487806921e-06
Mass transfer devolatilisation = 1.096268892e-05
Mass transfer surface reaction = 1.020134234e-05

Cloud: limestoneCloud1
Total number of parcels added = 18
Total mass introduced = 0.0001
Current number of parcels = 18
Current mass in system = 0.0001

Solving chemistry
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
DILUPBiCG: Solving for Ux, Initial residual = 0.06843280288, Final residual = 5.521803873e-06, No Iterations 3
DILUPBiCG: Solving for Uy, Initial residual = 0.08299776218, Final residual = 6.610758425e-07, No Iterations 4
DILUPBiCG: Solving for O2, Initial residual = 0.02991408207, Final residual = 2.316092842e-07, No Iterations 4
DILUPBiCG: Solving for CH4, Initial residual = 0.02443785456, Final residual = 2.649269485e-07, No Iterations 4
DILUPBiCG: Solving for H2, Initial residual = 0.0315238668, Final residual = 3.822091915e-07, No Iterations 4
DILUPBiCG: Solving for CO2, Initial residual = 0.02938438982, Final residual = 4.003208991e-07, No Iterations 4
DILUPBiCG: Solving for H2O, Initial residual = 0.03523143241, Final residual = 1.795963946e-08, No Iterations 5
DILUPBiCG: Solving for hs, Initial residual = 0.04197246509, Final residual = 5.114508673e-07, No Iterations 4


--> FOAM FATAL ERROR:
attempt to use janafThermo<equationOfState> out of temperature range 0 -> 5000; T = -461.698

From function janafThermo<equationOfState>::checkT(const scalar T) const
in file /Users/dingsudong/OpenFOAM/OpenFOAM-1.7.0/src/thermophysicalModels/specie/lnInclude/janafThermoI.H at line 63.

FOAM aborting
===========================================
but on my linux workstation, it's ok.

the configuration as below,
Mac:
Mac OS X 10.6.3, gcc-mp-4.5, OpenFoam 1.7
Linux:
OpenSuse 11.1, gcc-4.3, Openfoam 1.7

would you reproduce it?

Leo

chrisb July 6, 2010 07:48

Quote:

Originally Posted by leosding (Post 265869)
Bernhard,
I run the case of coalChemistryFoam/simplifiedSiwek, at time=0.252 it broken, the log reads:

I get a similar error for gcc 4.4.4 for both linux and mac OS (OF-1.7). Unfortunately I don't know enough about the code (or the chemistry) to help you much further, however, it seems like there may be some stability problems with the equations you are solving. I noticed 2-3 time-steps before the temperature dramatically increased tmax = O(2500K). I am completely unfamiliar with the problem at hand, however, this did not seem consistent with the previous time-steps.

Cb

gschaider July 7, 2010 08:06

Quote:

Originally Posted by leosding (Post 265869)
Bernhard,
I run the case of coalChemistryFoam/simplifiedSiwek, at time=0.252 it broken, the log reads:
=======================================
Courant Number mean: 0.1533929535 max: 0.6093346791
deltaT = 0.0005
Time = 0.252



--> FOAM FATAL ERROR:
attempt to use janafThermo<equationOfState> out of temperature range 0 -> 5000; T = -461.698

From function janafThermo<equationOfState>::checkT(const scalar T) const
in file /Users/dingsudong/OpenFOAM/OpenFOAM-1.7.0/src/thermophysicalModels/specie/lnInclude/janafThermoI.H at line 63.

FOAM aborting
===========================================
but on my linux workstation, it's ok.

On my linux-workstation (gcc 4.4.0) I get the same error (with a slightly different time and a different temperature. Funny thing is that when I run it in Debug-mode (same sources, same machine, same compiler) it is slightly different again

So it is not a Mac-problem but a problem with the case (and possibly the compiler)

Bernhard

leosding July 7, 2010 20:00

Quote:

Originally Posted by leosding (Post 265869)
Bernhard,
I run the case of coalChemistryFoam/simplifiedSiwek, at time=0.252 it broken, the log reads:
=======================================
Courant Number mean: 0.1533929535 max: 0.6093346791
deltaT = 0.0005
Time = 0.252

--> FOAM FATAL ERROR:
attempt to use janafThermo<equationOfState> out of temperature range 0 -> 5000; T = -461.698

From function janafThermo<equationOfState>::checkT(const scalar T) const
in file /Users/dingsudong/OpenFOAM/OpenFOAM-1.7.0/src/thermophysicalModels/specie/lnInclude/janafThermoI.H at line 63.

FOAM aborting
===========================================
but on my linux workstation, it's ok.

the configuration as below,
Mac:
Mac OS X 10.6.3, gcc-mp-4.5, OpenFoam 1.7
Linux:
OpenSuse 11.1, gcc-4.3, Openfoam 1.7

would you reproduce it?

Leo

Bernhard,
I just check the log file for the calculation on Linux port again, you are right. it's same error as you said that with a slightly different time and a different temperature.

It should be reported to developer as a bug.

regards,
Leo

gschaider July 8, 2010 08:22

Quote:

Originally Posted by leosding (Post 266254)
I just check the log file for the calculation on Linux port again, you are right. it's same error as you said that with a slightly different time and a different temperature.

It should be reported to developer as a bug.

You're right. Go ahead: report

gschaider July 8, 2010 08:41

Quote:

Originally Posted by gschaider (Post 266364)
You're right. Go ahead: report

Sorry. Just noticed: you already did


All times are GMT -4. The time now is 04:51.