I wrote a while back asking about some problems I had with SANDER runs
ending with:
COORDINATE RESETTING (SHAKE) WAS NOT ACCOMPLISHED
WITHIN 3000 ITERATIONS
The issues with temp0 == tempi vs temp0 != tempi were brought up and I
thought the problem had been solved when I reduced the chance of
point-heating errors. I have since, however, found this not to be the
case. In fact, the error seems to be less set-up related and more an
issue with my sytem (software or hardware). I have observed multiple
sander runs crash within minutes of each other, all giving the same Shake
error. The runs used similar input files, however they had been running
for very different amounts of time. The failures also appear to be more
likely when logged into the console under my account. Has anyone seen
anything similar to this? Any suggestions as to what I should try?
My set-up is as follows
Dual PIII 650 running RedHat 6.1 (RedHat default SMP kernel), AMBER
compiled under gcc v. 2.91.66 using the default gcc MACHINE file (no
attempts at making it SMP capable were made as multiple simultaneous runs
are commonly used).
The following is a representative sander run performed on a 20 residue
peptide in a water box.
&cntrl
ntx=5
irest=1
imin=0,
tempi=498, temp0=498, ntt=1, tautp=0.2,
ntp=1, taup=0.2,
ntb=2, ntc=2, ntf=2,
nstlim=1300000, dt=.002
ntwe=100, ntwx=100, ntpr=50,
&end
Please e-mail me any questions you feel appropriate and I will respond to
the newsgroup with the answers.
Matt
_________________________
R. M. Fesinmeyer
rmf_at_u.washington.edu
Department of Chemistry
University of Washington
Received on Mon Jun 26 2000 - 15:11:29 PDT
