Experience in building OpenFOAM on LinuxIA64

jens_klostermann · November 30, 2006, 15:21

David

@1) We are now using lam for $WM_MPLIB but it also worked with openmpi for $WM_MPLIB

@2) can't remember, sorry

@3) yes, but if you set the number of cpus for compilation higer e.g. $WM_NCOMPPROCS=8 you will get some speedup (I think to the order of 2h or so)

Jens

dhebert · December 1, 2006, 07:57

Many thanks to Jens for help getting me started on the SGI/Altix. The use of WM_NCOMPPROCS was a BIG help! I am still having problems I am hoping to get help with...

The OS on the machine is SGI ProPack 4.0.3, with gcc version 3.3.3 and icc version 9.0. (Note it is a supercomputer resource). When compiling, openfoam uses icpc, and I set the WM_MPLIB flag to LAM. Here are some problems I hope someone can help with...

1) I get warning #1357: optimization disabled due to excessive resource requirements. Is there a way to get around this?

2) PatchTool gives lots of Java errors. I certainly don't need anything graphical to run on the SGI/Altix machine, so is there a way to turn off anything that uses Java?

3) decomposePar gives the following error:

undefined reference to `Foam::Cloud<foam::indexedparticle>::readFields()'

4) several mico errors:
ld: /u/ac/dhebert/OpenFOAM/OpenFOAM-1.3/src/mico-2.3.11/platforms/linuxIA64I64DPOpt/ lib/libmicocoss2.3.11.a(CosNaming.o): @gprel relocation against \dynamic symbol _marshaller__seq_CosNaming_Binding

Any help is appreicated, I anticipate having large simulations I would like to run parallel.

Thanks again,

David

dhebert · December 2, 2006, 09:24

I just wanted to write and say it seems all of the compilation warnings/problems were fixed when I installed a local version of gcc-4.1.1. Highly recommend doing this instead of using local compilers.

I am still having a problem with decomposePar, namely

undefined reference to `Foam::Cloud<foam::indexedparticle>::readFields()'

If anyone has any idea how to fix this, I would be interested.

Thanks again,

David

rameshbvb · July 25, 2010, 09:09

Hallo Henry,

I am investigating turbulent lean premixed high pressure flames at stationary gas turbine conditions using your combustion model (unstrained laminar flame speed; solving for Xi transport) in RANS scenario. The combustion temperature is going beyond the adiabatic flame temperature to as high as 6000 K.
It would be helpful if you shed some light on this.

Best Regards,
Ramesh

November 30, 2006, 15:21	David @1) We are now using	#41
jens_klostermann Senior Member Jens Klostermann Join Date: Mar 2009 Posts: 117 Rep Power: 17	David @1) We are now using lam for $WM_MPLIB but it also worked with openmpi for $WM_MPLIB @2) can't remember, sorry @3) yes, but if you set the number of cpus for compilation higer e.g. $WM_NCOMPPROCS=8 you will get some speedup (I think to the order of 2h or so) Jens

December 1, 2006, 07:57	Many thanks to Jens for help g	#42
dhebert Member David Hebert Join Date: Mar 2009 Posts: 31 Rep Power: 17	Many thanks to Jens for help getting me started on the SGI/Altix. The use of WM_NCOMPPROCS was a BIG help! I am still having problems I am hoping to get help with... The OS on the machine is SGI ProPack 4.0.3, with gcc version 3.3.3 and icc version 9.0. (Note it is a supercomputer resource). When compiling, openfoam uses icpc, and I set the WM_MPLIB flag to LAM. Here are some problems I hope someone can help with... 1) I get warning #1357: optimization disabled due to excessive resource requirements. Is there a way to get around this? 2) PatchTool gives lots of Java errors. I certainly don't need anything graphical to run on the SGI/Altix machine, so is there a way to turn off anything that uses Java? 3) decomposePar gives the following error: undefined reference to `Foam::Cloud<foam::indexedparticle>::readFields()' 4) several mico errors: ld: /u/ac/dhebert/OpenFOAM/OpenFOAM-1.3/src/mico-2.3.11/platforms/linuxIA64I64DPOpt/ lib/libmicocoss2.3.11.a(CosNaming.o): @gprel relocation against \dynamic symbol _marshaller__seq_CosNaming_Binding Any help is appreicated, I anticipate having large simulations I would like to run parallel. Thanks again, David

December 2, 2006, 09:24	I just wanted to write and say	#43
dhebert Member David Hebert Join Date: Mar 2009 Posts: 31 Rep Power: 17	I just wanted to write and say it seems all of the compilation warnings/problems were fixed when I installed a local version of gcc-4.1.1. Highly recommend doing this instead of using local compilers. I am still having a problem with decomposePar, namely undefined reference to `Foam::Cloud<foam::indexedparticle>::readFields()' If anyone has any idea how to fix this, I would be interested. Thanks again, David

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
New building machine for OpenFoam	gerbervdgraaf	OpenFOAM Installation	23	December 9, 2009 02:39
Building OpenFOAM on IRIX	lakeat	OpenFOAM Installation	7	July 16, 2008 07:27
Building OpenFOAM 13 on solaris10Sun Opteron AMD64	phil	OpenFOAM Installation	0	August 28, 2006 01:12
Building OpenFoAm on SGI Altix 64bits	anne	OpenFOAM Installation	8	June 15, 2006 09:27
Problem with LinuxIA64 installation of OpenFOAM13	antares	OpenFOAM Installation	0	May 25, 2006 12:19

July 25, 2010, 09:09		#44
rameshbvb New Member Ramesh Join Date: Aug 2009 Posts: 4 Rep Power: 16	Hallo Henry, I am investigating turbulent lean premixed high pressure flames at stationary gas turbine conditions using your combustion model (unstrained laminar flame speed; solving for Xi transport) in RANS scenario. The combustion temperature is going beyond the adiabatic flame temperature to as high as 6000 K. It would be helpful if you shed some light on this. Best Regards, Ramesh