CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Software User Forums > OpenFOAM > OpenFOAM Bugs

Pow in lib64tlslibmso6 SigFpe when running coodles with SpalartAllmaras

Register Blogs Community New Posts Updated Threads Search

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   December 6, 2007, 03:03
Default Hi forum and Henry! I've ru
  #1
Member
 
lillberg's Avatar
 
Eric Lillberg
Join Date: Mar 2009
Location: Stockholm
Posts: 80
Rep Power: 17
lillberg is on a distinguished road
Send a message via Skype™ to lillberg
Hi forum and Henry!

I've run into this problem when running coodles in parallel using the Spalart-Allmaras DES model.

[3] #0 Foam::error::printStack(Foam:stream&) in "/local/users/exterli/OpenFOAM/OpenFOAM-1.4.1/lib/linux64GccDPOpt/libOpe
nFOAM.so"
[3] #1 Foam::sigFpe::sigFpeHandler(int) in "/local/users/exterli/OpenFOAM/OpenFOAM-1.4.1/lib/linux64GccDPOpt/libOpenFOAM.s
o"
[3] #2 ?? in "/lib64/tls/libc.so.6"
[3] #3 ?? in "/lib64/tls/libm.so.6"
[3] #4 ?? in "/lib64/tls/libm.so.6"
[3] #5 pow in "/lib64/tls/libm.so.6"
[3] #6 Foam::pow(Foam::Field<double>&, Foam::UList<double> const&, double const&) in "/local/users/exterli/OpenFOAM/OpenFO
AM-1.4.1/lib/linux64GccDPOpt/libOpenFOAM.so"
[3] #7 void Foam::pow<foam::fvpatchfield,>(Foam::GeometricFiel d<double,>&,
Foam::GeometricField<double,> const&, Foam::dimensioned<double> const&) in "/local/users
/exterli/OpenFOAM/OpenFOAM-1.4.1/lib/linux64GccDPOpt/libcompressibleLESmodels.so "
[3] #8 Foam::tmp<foam::geometricfield<double,> > Foam::pow<Foam::fvPatchField, Foam::vol
Mesh>(Foam::GeometricField<double,> const&, Foam::dimensioned<double> const&) in "/local/
users/exterli/OpenFOAM/OpenFOAM-1.4.1/lib/linux64GccDPOpt/libcompressibleLESmode ls.so"
[3] #9 Foam::tmp<foam::geometricfield<double,> > Foam::pow<Foam::fvPatchField, Foam::vol
Mesh>(Foam::GeometricField<double,> const&, double const&) in "/local/users/exterli/OpenF
OAM/OpenFOAM-1.4.1/lib/linux64GccDPOpt/libcompressibleLESmodels.so"
[3] #10 Foam::compressible::LESmodels::SpalartAllmaras::fw (Foam::GeometricField<double,>
const&) const in "/local/users/exterli/OpenFOAM/OpenFOAM-1.4.1/lib/linux64GccDPOpt/libcompressibl eLESmodels.so"
[3] #11 Foam::compressible::LESmodels::SpalartAllmaras::co rrect(Foam::tmp<foam::geometri cfield<foam::tensor<double>, Foam:
:fvPatchField, Foam::volMesh> > const&) in "/local/users/exterli/OpenFOAM/OpenFOAM-1.4.1/lib/linux64GccDPOpt/libcompressibl
eLESmodels.so"
[3] #12 Foam::compressible::LESmodel::correct() in "/local/users/exterli/OpenFOAM/OpenFOAM-1.4.1/lib/linux64GccDPOpt/libco
mpressibleLESmodels.so"
[3] #13 main in "/local/users/exterli/OpenFOAM/OpenFOAM-1.4.1/applications/bin/linux64GccDPOpt/c oodles"
[3] #14 __libc_start_main in "/lib64/tls/libc.so.6"
[3] #15 Foam::lduMatrix::lduMatrix(Foam::lduMatrix&, bool) in "/local/users/exterli/OpenFOAM/OpenFOAM-1.4.1/applications/b
in/linux64GccDPOpt/coodles"
srun: error: n107: task3: Floating point exception (core dumped)

Running the same mesh with the same code using any other LES model works just fine. Tried setting floatTransfer to 0 but it didn't change anything. Printed the max/min values of what goes into fw but couldn't see anything unusual. The simulation runs for some 10.000 timesteps and the solution look Ok.

The cluster has an older kernel and glibc and the problem does not occur on my laptop which is running openSuSE 10.3. Hence, this is likely a system specific problem. Anyhow, I would like to hear if someone has any idea how to overcome this. (except upgrading the cluster nodes, 400 cpu's and I'm not admin :-( )

Best Regards

//Eric
lillberg is offline   Reply With Quote

Old   December 6, 2007, 04:06
Default Hi Eric, Try with this htt
  #2
Senior Member
 
Join Date: Mar 2009
Posts: 854
Rep Power: 22
henry is on a distinguished road
Hi Eric,

Try with this SpalartAllmaras.C

Henry
henry is offline   Reply With Quote

Old   December 7, 2007, 02:50
Default Works like a charm, thanks a l
  #3
Member
 
lillberg's Avatar
 
Eric Lillberg
Join Date: Mar 2009
Location: Stockholm
Posts: 80
Rep Power: 17
lillberg is on a distinguished road
Send a message via Skype™ to lillberg
Works like a charm, thanks a lot Henry!

Just a question: Why bounding r by 10?

//Eric
lillberg is offline   Reply With Quote

Old   December 7, 2007, 08:11
Default r must be clipped otherwise th
  #4
Senior Member
 
Join Date: Mar 2009
Posts: 854
Rep Power: 22
henry is on a distinguished road
r must be clipped otherwise the (r^6)^6 causes a floating point exception. Clipping r with 10 ensures that (r^6)^6 does not cause over-flow in either single or double precision.

Henry
henry is offline   Reply With Quote

Old   December 7, 2007, 09:17
Default Thanks again Henry! //Eric
  #5
Member
 
lillberg's Avatar
 
Eric Lillberg
Join Date: Mar 2009
Location: Stockholm
Posts: 80
Rep Power: 17
lillberg is on a distinguished road
Send a message via Skype™ to lillberg
Thanks again Henry!

//Eric
lillberg is offline   Reply With Quote

Reply


Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Bug in SpalartAllmaras seb62 OpenFOAM Bugs 39 May 30, 2012 15:25
SpalartAllmaras question egp OpenFOAM Running, Solving & CFD 45 October 28, 2010 04:30
SpalartAllmaras DES question ivan_cozza OpenFOAM Running, Solving & CFD 0 December 15, 2008 07:34
YPlus for SpalartAllmaras ddigrask OpenFOAM Running, Solving & CFD 1 December 12, 2008 15:29
meshing error in Gambit: SIGFPE Ralf Schmidt FLUENT 0 November 2, 2005 16:28


All times are GMT -4. The time now is 08:54.