[gmx-developers] segfault goes away when restarting from cpt

[Michael Brunsteiner]

 hi,
I post this here as it might be developers rather than a user issue ...I ran an NPT sim with simulated annealing of an amorphous solid sample withsome organic molecules,as in:
gmx grompp -f md-1bar-353-253.mdp -p sis3-7-simp.top -c up-nr2-3.gro -o do-nr2-3.tpr
nohup gmx mdrun -v -deffnm do-nr2-3 > er 2>&1 &

 after around 30 nano secs the simulation stops without further notice.neither in the log-file nor in stdout or stderr there are any indicators of what happenedbut when i look into the relevant syslog file i find:

May  5 03:24:38 rcpe-sbd-node03 kernel: [82541302.295784] gmx[2218]: 
segfault at ffffffff9d3ebea0 ip 00007f5b708be3a1 sp 00007f5b657f9dc0 error 7 in libgromacs.so.2.3.0[7f5b706d9000+1d1e000]

when i restart the sim on the same computer and from the last cpt file, as in: 

nohup gmx mdrun -v -deffnm do-nr2-3 -cpi do-nr2-3.cpt -noappend > er 2>&1 &

the sim happily continues beyond the point where it previously seg-faulted without any further issues ... 

tpr file is too large to attach (if anybody's interested i can upload it somewhere)below i put the last 30 or so lines of both stderr+stdout and the log-fileI believe the warning at the end of stderr is harmless, but even if it actually is the reasonfor the segfault this still does not explain why nothing is written to stderr when it happensand why the sim works when restarted from the cpt file ... can it be that this is a hardware issue??
regards,Michael

 stderr+stdout:[..]    Brand:  Intel(R) Core(TM) i7-4930K CPU @ 3.40GHz
    SIMD instructions most likely to fit this hardware: AVX_256
    SIMD instructions selected at GROMACS compile time: AVX_256

  Hardware topology: Full, with devices
  GPU info:
    Number of GPUs detected: 1
    #0: NVIDIA GeForce GTX 780, compute cap.: 3.5, ECC:  no, stat: compatible

Reading file do-nr2-3.tpr, VERSION 2016.3 (single precision)
Changing nstlist from 20 to 40, rlist from 1.2 to 1.2

Using 1 MPI thread
Using 12 OpenMP threads 

1 compatible GPU is present, with ID 0
1 GPU auto-selected for this run.
Mapping of GPU ID to the 1 PP rank in this node: 0

starting mdrun 'system'
110000000 steps, 110000.0 ps.
step   80: timed with pme grid 40 40 24, coulomb cutoff 1.200: 81.2 M-cycles
step   80: the box size limits the PME load balancing to a coulomb cut-off of 1.368
step  160: timed with pme grid 32 36 24, coulomb cutoff 1.368: 72.9 M-cycles
step  240: timed with pme grid 36 36 24, coulomb cutoff 1.264: 75.7 M-cycles
step  320: timed with pme grid 36 40 24, coulomb cutoff 1.216: 78.6 M-cycles
step  400: timed with pme grid 40 40 24, coulomb cutoff 1.200: 81.2 M-cycles
              optimal pme grid 32 36 24, coulomb cutoff 1.368
step 31031000, will finish Fri May  5 14:27:08 2017
Step 31031061  Warning: pressure scaling more than 1%, mu: 0.999153 0.982333 0.997814



log-file:[..]
           Step           Time
       31030000    31030.00000

Current ref_t for group System:    327.9
   Energies (kJ/mol)
           Bond          Angle    Proper Dih.  Improper Dih.          LJ-14
    8.65826e+03    1.38650e+04    1.08781e+04    4.26101e+02    6.01339e+03
     Coulomb-14        LJ (SR)   Coulomb (SR)   Coul. recip.      Potential
   -2.98635e+04   -1.12638e+04    1.63267e+04    2.11350e+02    1.52514e+04
    Kinetic En.   Total Energy    Temperature Pressure (bar)
    2.22541e+04    3.75055e+04    3.28069e+02    6.14802e+02

           Step           Time
       31031000    31031.00000

Current ref_t for group System:    327.8
   Energies (kJ/mol)
           Bond          Angle    Proper Dih.  Improper Dih.          LJ-14
    8.41290e+03    1.38660e+04    1.08950e+04    3.51583e+02    5.79937e+03
     Coulomb-14        LJ (SR)   Coulomb (SR)   Coul. recip.      Potential
   -2.99386e+04   -1.15255e+04    1.64549e+04    2.30994e+02    1.45468e+04
    Kinetic En.   Total Energy    Temperature Pressure (bar)
    2.27307e+04    3.72775e+04    3.35095e+02   -1.54620e+00








     -- 
Gromacs Developers mailing list

* Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List before posting!

* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists

* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers or send a mail to gmx-developers-request at gromacs.org.

   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20170505/5927754d/attachment.html>