Dear AMBER users,
I'm currently trying to carry out MD simulations with sander (MPI parallel version, AMBER11) on a protein system using SHAKE bond constraints for bonds containing hydrogen (sander option ntc=2). These work with up to 14 processors, but with any number of cores higher than that the simulations abort in the first step with the error message:
partition error in shake on processor 0
this processor has atoms 1 through 8265
atom 8265 is within this range
atom 8266 is not within this range !
According to what I read in the AMBER mailing list archive, this error message means that the two atoms of a hydrogen-containing bond to be constrained by SHAKE are split across processors, thus causing a problem for SHAKE. However, it was also mentioned that sander should take care of this properly and should only possibly get issues with nonstandard residues. The two atoms in question above are N and H of a threonine's backbone, i.e. they belong to a standard residue. I thus do not understand why sander is having problems here. If I switch off the SHAKE constraints (ntc=1), the simulation runs without problems with more than 14 cores.
A second thing that puzzles me is the fact that the error message is the same (i.e. same range of atoms for processor 0 and same two atoms causing a problem) no matter how many processors I use (as long as I use more than 14). The total number of atoms in my system is 122321, which, when divided by 8265, yields roughly 14.8. This seems to suggest that sander always places 8265 atoms on each of the first 14 cores as long as there are more than 14 cores, splitting the remaining atoms between all remaining cores. (And in the case of 14 cores or less it would then be forced to use a higher number of atoms per core, as 14*8265<122321. This would explain why the simulation works with 14 cores.) However, this seems absurd and I can't imagine that it really works like that. (But I must say I couldn't find anything about this in the manual.)
Thus I'd like to know how to make sander take care of the SHAKE bonds properly, and, if the atom decomposition is as odd as it appears, how to change this decomposition.
I'm not sure if this problem might be related to the specific platform on which I'm using sander, but it seems to be more general; thus I'm writing to the AMBER mailing list first before turning to the supercomputer's support team.
I hope this information is specific enough, but I'm happy to provide further information as well as whole input or output files if these are needed.
Best regards,
Marian Breuer
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Oct 31 2011 - 11:30:02 PDT
