[overture] Re: Failed some CG regression tests

  • From: Darin <dpdinca@xxxxxxxxxxx>
  • To: overture@xxxxxxxxxxxxx
  • Date: Tue, 12 May 2009 22:45:47 +0000 (UTC)

Thanks Bill, 

The seg fault in particular had me worried that I was linking to 
a bad library, so I'm happy to see there's another likely explanation. 

Darin 



----- Original Message ----- 
From: "Bill Henshaw" <henshaw@xxxxxxxx> 
To: overture@xxxxxxxxxxxxx 
Sent: Tuesday, May 12, 2009 2:54:34 PM GMT -08:00 US/Canada Pacific 
Subject: [overture] Re: Failed some CG regression tests 

Hi Darin, 

I think you are fine. That one test for cgcns does seg fault on 
some machines but we have fixed that for the next release. The 
other "errors" seem to be due to slightly different answers that 
result from different compilers. 

Regards, 
Bill 

Darin wrote: 
> Hello, 
> 
> I've built Overture.v23 and cg.v23 on a 64 bit ubuntu machine, and 
> while I passed all of the Overture regression tests, I failed three 
> tests for cg, all associated with the compressible Navier Stokes 
> solver. One failure was a seg fault. 
> 
> I'm predominantly interested in the incompressible Navier Stokes 
> solver, so I'm hoping I'm ok. I'd like to verify this, though. 
> 
> FYI, here is some potentially relevant information: 
> 
> used gcc, g++, gfortran, version 4.2.3 
> 
> configured hdf5 as follows: 
> computation@uE1:~/libraries/hdf5-1.6.8$ export CC=gcc; export CXX=g++; 
> export F77=gfortran 
> computation@uE1:~/libraries/hdf5-1.6.8$ ./configure 
> --prefix=/home/computation/hdf5 
> 
> configured A++ as follows: 
> export CC=cc; export CXX=g++; export F77=gfortran 
> ./configure --prefix="/home/computation/A++P++" --enable-SHARED_LIBS 
> 
> configured petsc as follows: 
> export PETSC_DIR=/home/computation/libraries/petsc-2.3.2-p10 
> export PETSC_ARCH=linux-gnu-opt 
> export PETSC_LIB=$PETSC_DIR/lib/$PETSC_ARCH 
> ./config/configure.py --prefix=/home/computation/petscOverture 
> -PETSC_ARCH=linux-gnu-opt --with-cc=gcc -with-debugging=0 
> -with-fortran=0 -with-matlab=0 -with-mpi=0 -with-shared=1 
> -with-dynamic=1 --with-blas-lapack-dir=/usr/lib/atlas --doCleanup=0 2>&1 
> | tee configureLog 
> 
> configured Overture as follows: 
> tcsh 
> source defenv 
> ./configure linux CC=g++ cc=gcc FC=gfortran useHDF5 multigrid 
> 
> Below is the regression testing ouput. 
> 
> Thanks, 
> Darin 
> 
> uE1:~/libraries/cg.v23/cns/check> ./check.p 
> 
> ================================================================================
>  
> This perl script will run regression tests 
> (The lists of tests is in the file named 'tests') 
> Usage: 
> check.p [testName] [replace][check=<checkFileDirectory>] 
> [-solver=name] [-verbose=<num>] 
> [testName] = name of a particular test to run (by default do all) 
> replace : replace check files with the new ones computed 
> verbose : verbose=0 : print very litte, verbose=1 print more 
> ==============================================================================
>  
> 
> program=CGBUILDPREFIX/cns/bin/cgcns 
> checkFile=cns.check 
> Using program = /home/computation/libraries/cg.v23/cns/bin/cgcns 
> Using the executable: /home/computation/libraries/cg.v23/cns/bin/cgcns 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./../cmd/tz.cmd -g=square10 -tf=.1 -tp=.1 -go=go >! tzSquare.out 
> checkCheckFiles: files cns.check and tzSquare.check agree at 2 times , 
> maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "tzSquare" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./../cmd/tz.cmd -g=cic -tf=.1 -tp=.1 -go=go >! tzCic.out 
> checkCheckFiles: files cns.check and tzCic.check agree at 2 times , 
> maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "tzCic" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./../cmd/tz.cmd -g=box10 -tf=.05 -tp=.05 -go=go >! tzBox.out 
> Segmentation fault 
> *** There was an error running 
> /home/computation/libraries/cg.v23/cns/bin/cgcns and generating tzBox **** 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./square.cns >! square.cns.out 
> checkCheckFiles: files cns.check and square.cns.check agree at 3 times , 
> maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "square.cns" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./cicSubsonic >! cicSubsonic.out 
> checkCheckFiles: files cns.check and cicSubsonic.check agree at 3 times 
> , maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "cicSubsonic" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./shockTubeg >! shockTubeg.out 
> checkCheckFiles: files cns.check and shockTubeg.check agree at 3 times , 
> maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "shockTubeg" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./cic.godunov >! cic.godunov.dp.out 
> checkCheckFiles: files cns.check and cic.godunov.dp.check agree at 2 
> times , maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "cic.godunov.dp" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./detonationOneStepIdeal >! detonationOneStepIdeal.out 
> checkCheckFiles: files cns.check and detonationOneStepIdeal.check agree 
> at 3 times , maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "detonationOneStepIdeal" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./cicShockMove >! cicShockMove.out 
> checkCheckFiles: files cns.check and cicShockMove.check agree at 5 times 
> , maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "cicShockMove" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./detonationMieGruneisen >! detonationMieGruneisen.out 
> checkCheckFiles: files cns.check and detonationMieGruneisen.check agree 
> at 2 times , maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "detonationMieGruneisen" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./multi >! multi.out 
> checkCheckFiles: files cns.check and multi.check agree at 2 times , 
> maxDiff=3.00e-07, (tol=1.00e-05) 
> ... "multi" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./ignitionAndGrowth >! ignitionAndGrowth.out 
> checkCheckFiles: files cns.check and ignitionAndGrowth.check agree at 1 
> times , maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "ignitionAndGrowth" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./eulerAxisymmetric >! eulerAxisymmetric.out 
> checkCheckFiles: files cns.check and eulerAxisymmetric.check agree at 3 
> times , maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "eulerAxisymmetric" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./sqStrat_implicit >! sqStrat_implicit.out 
> checkCheckFiles: files cns.check and sqStrat_implicit.check agree at 3 
> times , maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "sqStrat_implicit" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./sqStrat_newton >! sqStrat_newton.out 
> checkCheckFiles: files cns.check and sqStrat_newton.check agree at 6 
> times , maxDiff=0.00e+00, (tol=1.00e-05) 
> ... "sqStrat_newton" appears to be correct 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./square.mp >! square.mp.out 
> checkCheckFiles: files do not agree, maxDiff=6.570000e-01, 
> file1=cns.check, file2=square.mp.check 
> Component 0: values: 9.6000e-01 (file1) and 9.0000e-01 (file2) differ 
> by 6.00e-02. 
> Component 0: norms: 1.0300e+00 (file1) and 1.0470e+00 (file2) differ 
> by 1.70e-02. 
> Component 1: norms: 5.4170e-02 (file1) and 1.3720e-01 (file2) differ 
> by 8.30e-02. 
> Component 3: values: 9.8000e-01 (file1) and 9.7000e-01 (file2) differ 
> by 1.00e-02. 
> Component 3: norms: 1.0130e+00 (file1) and 1.0280e+00 (file2) differ 
> by 1.50e-02. 
> Component 4: values: 1.9000e-01 (file1) and 1.8000e-01 (file2) differ 
> by 1.00e-02. 
> Component 5: values: -7.6000e-01 (file1) and -7.9000e-01 (file2) differ 
> by 3.00e-02. 
> Component 5: norms: 0.0000e+00 (file1) and 6.5700e-01 (file2) differ 
> by 6.57e-01. 
> Component 7: norms: 1.6200e+00 (file1) and 1.6720e+00 (file2) differ 
> by 5.20e-02. 
> ++++ The check files do not agree +++++ 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./mpTestV >! mpTestV.out 
> checkCheckFiles: files do not agree, maxDiff=1.000000e-05, 
> file1=cns.check, file2=mpTestV.check 
> Component 7: norms: 1.3560e-02 (file1) and 1.3570e-02 (file2) differ 
> by 1.00e-05. 
> ++++ The check files do not agree +++++ 
> running: /home/computation/libraries/cg.v23/cns/bin/cgcns noplot 
> ./fourComp >! fourComp.out 
> checkCheckFiles: files cns.check and fourComp.check agree at 5 times , 
> maxDiff=1.00e-14, (tol=1.00e-05) 
> ... "fourComp" appears to be correct 
> ******************************************************** 
> ************ : There were 3 ERRORS **************** 
> ******************************************************** 
> 

Other related posts: