[gmx-developers] freeenergy/simtemp and freeenergy/vdwalone fail on rhel6/sandy bridge
Mark Dixon
m.c.dixon at leeds.ac.uk
Mon Oct 21 13:45:10 CEST 2013
Hi,
Possibly related to Friday's "freeenergy.relative unstable?" thread...
A source install of gromacs 4.6.3 passes the regression tests on one of
our clusters, but the serial double-precision and MPI double-precision
builds fail for a couple of freeenergy tests (simtemp and vdwalone) on a
second cluster.
We used the same recent compiler (intel/13.1.3.192), the same method
(below) and the same MPI (openmpi/1.6.5) on both machines; however, one
machine is running RHEL5 and is based on Intel Nehalem processors, whereas
the other is running CentOS6 and is based on Intel Sandy Bridge
processors.
I wondered if anyone here had noticed this?
The failed numbers are only right to ~2 sig figs. Should I be concerned
with the build on the sandy bridge machine?
Thanks,
Mark
--
-----------------------------------------------------------------
Mark Dixon Email : m.c.dixon at leeds.ac.uk
HPC/Grid Systems Support Tel (int): 35429
Information Systems Services Tel (ext): +44(0)113 343 5429
University of Leeds, LS2 9JT, UK
-----------------------------------------------------------------
# Serial, double-precision build method
cmake -DGMX_DOUBLE=ON -DREGRESSIONTEST_DOWNLOAD=ON -DGMX_BUILD_OWN_FFTW=ON -DGMX_X11=ON -DCMAKE_INSTALL_PREFIX=$prefix ../gromacs-4.6.3
make
make check
# simtemp checkpot.out on centos6/sandy bridge machine
comparing energy file reference_d.edr and ener.edr
There are 33 terms in the energy files
There are 6 terms to compare in the energy files
LJ (SR) step 20: 5620.37, step 20: 5609.04
LJ (SR) step 40: 5650.96, step 40: 5628.03
Angle step 60: 8.55224, step 60: 8.65452
LJ (SR) step 60: 5674.57, step 60: 5643.39
Angle step 80: 7.04247, step 80: 6.98253
LJ (SR) step 80: 5697.85, step 80: 5658.71
LJ (SR) step 100: 5789.57, step 100: 5748.48
Coulomb (SR) step 100: -31075.7, step 100: -31043.8
Files read successfully
# vdwalone checkpot.out on centos6/sandy bridge machine
comparing energy file reference_d.edr and ener.edr
There are 40 terms in the energy files
There are 11 terms to compare in the energy files
LJ (SR) step 20: 4318.48, step 20: 4307.94
LJ (SR) step 30: 4320.84, step 30: 4302.5
LJ (SR) step 40: 4340.35, step 40: 4315.17
LJ (SR) step 50: 4371.42, step 50: 4339.86
LJ (SR) step 60: 4409.49, step 60: 4370.77
Coulomb (SR) step 60: -28193.8, step 60: -28158.5
Files read successfully
# MPI, double-precision build method
cmake -DGMX_DOUBLE=ON -DGMX_MPI=ON -DGMX_OPENMP=OFF -DREGRESSIONTEST_DOWNLOAD=ON -DGMX_BUILD_OWN_FFTW=ON -DGMX_X11=ON -DCMAKE_INSTALL_PREFIX=$prefix ../gromacs-4.6.3
make
make check
# simtemp checkpot.out on centos6/sandy bridge machine
comparing energy file reference_d.edr and ener.edr
There are 33 terms in the energy files
There are 6 terms to compare in the energy files
LJ (SR) step 20: 5620.37, step 20: 5607.86
LJ (SR) step 40: 5650.96, step 40: 5625.5
LJ (SR) step 60: 5674.57, step 60: 5643.92
LJ (SR) step 80: 5697.85, step 80: 5659.46
Coulomb (SR) step 80: -30834, step 80: -30801.7
LJ (SR) step 100: 5789.57, step 100: 5747.16
Coulomb (SR) step 100: -31075.7, step 100: -31038.9
Files read successfully
# vdwalone checkpot.out on centos6/sandy bridge machine
comparing energy file reference_d.edr and ener.edr
There are 40 terms in the energy files
There are 11 terms to compare in the energy files
LJ (SR) step 20: 4318.48, step 20: 4307.03
LJ (SR) step 30: 4320.84, step 30: 4301.78
LJ (SR) step 40: 4340.35, step 40: 4313.95
LJ (SR) step 50: 4371.42, step 50: 4338.84
LJ (SR) step 60: 4409.49, step 60: 4371.53
Coulomb (SR) step 60: -28193.8, step 60: -28158.6
Files read successfully
More information about the gromacs.org_gmx-developers
mailing list