AMBER Archive (2008)

Subject: RE: AMBER: nmode: number of atoms limitation

From: Ross Walker (ross_at_rosswalker.co.uk)
Date: Wed May 07 2008 - 16:39:44 CDT

Hi Qi,

 

You probably don’t have enough memory in your machine to run the simulation.
Thus it is probably swapping like crazy. You either need to buy more memory
or find a machine with more memory. You might also consider getting a
development allocation of 30,000 SUs on the NSF’s SGI Altix at NCSA
(Cobalt). This sort of machine is ideal for such simulations.

 

See: https://pops-submit.teragrid.org/

 

All the best

Ross

 

From: owner-amber_at_scripps.edu [mailto:owner-amber_at_scripps.edu] On Behalf Of
Qi Yan
Sent: Wednesday, May 07, 2008 11:26 AM
To: amber_at_scripps.edu
Subject: AMBER: nmode: number of atoms limitation

 

Hi,all:

I'm using amber9 to do nmode analysis. The atoms in my system are 8454 which
are more than maxatom=8000 written in sizes2.h.

Therefore I did some modification as below
parameter (memdrv=10000000)
parameter (maxatom=9000)
parameter (maxint=500)
parameter (maxvec=500)
parameter (maxdih = 55000)
parameter (maxdia = 35000)
parameter (maxinb = 90000)
parameter (maxbon = 26000)
parameter (maxbnh = 20000)
parameter (maxang = 30000)
parameter (maxanh = 30000)

and then I recompiled nmode. Actually, nmode can work after this
modification, but when I check the process of CPU, I found that %CPU is ~4
which is too small and %MEM is ~90 which is very high. The nmode with 8454
atoms has been running for a whole day and I still can not get result. When
I use a system with 3244 atoms to test this recompiled system, it worked
very well and I can get the result within 20 minutes.

Does anybody know what is the reason?

Thanks in advance,

Qi

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu