Dear AMBER users,
when I run mm-pbsa over a series of coordinate snapshots
(generated by the mm-pbsa script in a previous run using
the GC=1 option) to calculate the binding free energy
(options MM, PB and MS =1, GB=0) the calculation fails with
this error message in the last line of mmpbsa.log:
p0_21449: p4_error: OOPS: shmat failed : 25
this occurs after ~110-150 processed snapshots (it
processes all _com snapshots and then fails between the
10th to 50th _rec snapshot at different times in different
runs) The output generated up to then looks reasonable.
I am using mpirun and have modified mmpbsa to use parallel
sander as described in the archives, by changing (diff
mm_pbsa.old mm_pbsa):
1468c1468
< my $command = $r_pro->{"SANDER"} . " -O" .
---
> my $command = "/usr/local/mpich-1.2.4/bin/mpirun " .
$r_pro->{"SANDER"} . " -O" .
mm-pbsa ran well in a test with the same setup using only
20 snapshots.
I suppose this is a problem with running sander, not with
delphi or my snapshot files but I don't understand the
error message.
Has someone encountered this error before?
Any ideas what could have gone wrong?
Kind regards,
Thomas Steinbrecher
Received on Tue Jan 28 2003 - 07:02:04 PST
