Re: Large MD simulations and atom number limits with NAMD

From: Peter Freddolino (beakerbreaker_at_gmail.com)
Date: Mon Jun 30 2008 - 14:43:08 CDT

Hi Andrew,
this is currently being worked on internally here, and we have run
significantly larger test systems. I assume this will probably be
publicly released with namd 2.7 sometime this fall, but can't guarantee
this just yet.
Best,
Peter

Andrew Thomas wrote:
> Dear all,
>
> I am interested in running very large MD simulations (>4 million
> atoms). I have read a previous post suggesting that NAMD is limited to
> ~4 million atoms due to the use of 22-bit atom IDs. I was wondering if
> there was a workaround for this limitation. For testing I've set up
> different boxes of waters all setup using the same scripts:
> 0) small systems 100k to 3.9 million atoms work fine, no problems,
> 1) a system with 4.29 millions atoms gives the error:
> ################
> Info: Entering startup phase 0 with 1886724 kB of memory in use.
> Info: Entering startup phase 1 with 1886724 kB of memory in use.
> Info: Entering startup phase 2 with 2893392 kB of memory in use.
> Info: Entering startup phase 3 with 2926956 kB of memory in use.
> Info: PATCH GRID IS 20 (PERIODIC) BY 20 (PERIODIC) BY 20 (PERIODIC)
> Info: REMOVING COM VELOCITY 0.00477832 -0.00342197 -0.000182436
> FATAL ERROR: Mother atom with hydrogenGroupSize of 0!
> FATAL ERROR: See http://www.ks.uiuc.edu/Research/namd/bugreport.html
> ------------- Processor 0 Exiting: Called CmiAbort ------------
> ################
> 2) a system with 12.5 million atoms gives a Segmentation fault:
> ################
> Info: COORDINATE PDB test.pdb
> Info: STRUCTURE FILE final.2.psf
> Info: PARAMETER file: CHARMM format!
> Info: PARAMETERS /home/athomas/toppar/par_all27_prot_lipid.prm
> Info: USING ARITHMETIC MEAN TO COMBINE L-J SIGMA PARAMETERS
> Info: SUMMARY OF PARAMETERS:
> Info: 180 BONDS
> Info: 447 ANGLES
> Info: 566 DIHEDRAL
> Info: 46 IMPROPER
> Info: 6 CROSSTERM
> Info: 114 VDW
> Info: 0 VDW_PAIRS
> Segmentation fault
> ################
>
> Memory does not seem to be the problem; I'm working on a 24 GB system
> and memory only goes to ~ 10-20% before crashing.
>
> Any help would be most appreciated, I've searched the archives as well
> as net searches and I've not seen a solution. I presume there is some
> fix for this as Klaus Schulten is looking towards simulations well
> over the 4million atom limit.
>
> Andy Thomas, Ph.D.
> Dept. Biochemistry
> University of Iowa
>
>
>
> ------------------------------------------------------------------------
> The other season of giving begins 6/24/08. Check out the iím
> Talkathon. Check it out!
> <http://www.imtalkathon.com?source=TXT_EML_WLH_SeasonOfGiving>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:49:37 CST