[AMBER] Files work with sander and not pmemd

From: Nicolas Cheron <nicolas.cheron.boulot.gmail.com>
Date: Wed, 7 Nov 2018 12:12:38 +0100

Dear all,

I have files that were used previously to perform EVB simulations with
sander (prmtop and rst). I now want to perform simulation on a pure VB
state (no more EVB) and was thinking of using pmemd, however I can't. For
example, with this md.in file

imin=0, irest=1, ntx=5,
ntpr=500, ntwr=500,
ntwx=500, ntwe=500, ntwv=500,
nstlim=50000, dt=.001,
ntb=1, ntp=0,
ntc=2, ntf=2,

and this command "mpirun sander.MPI -O -i md.in -p VB1.prmtop -c
Equil300.rst -o Prod.out -r Prod.rst -x Prod.mdcrd -e Prod.mden -inf
Prod.mdinfo -v Prod.mdvel", everything runs smoothly. If I take the same
file and replace sander.MPI by pmemd.MPI, it crashes (no Prod.* file
written and error files copied below). I have tried on our local cluster
with amber16 and on national servers with amber14, it never works. On our
cluster, the error file says

Program received signal SIGSEGV: Segmentation fault - invalid memory

Backtrace for this error:
#0 0x7fd7813bf2ef in ???
#1 0x422d44 in ???
#2 0x51ddb2 in ???
#3 0x505082 in ???
#4 0x40590c in ???
#5 0x7fd7813ab444 in ???
#6 0x405e66 in ???
#7 0xffffffffffffffff in ???

and the output file says

mpirun noticed that process rank 0 with PID 0 on node nnode015 exited on
signal 11 (Segmentation fault).

If I use only "pmemd" instead of "mpirun pmemd.MPI", it also crashes. I am
a little bit lost of what could be the problem. Could there be something in
the prmtop file that prevent it from being used with pmemd? Did I miss
something else?

Thank you for your help.

AMBER mailing list
Received on Wed Nov 07 2018 - 03:30:02 PST
Custom Search