Re: AMBER: Sander (Amber 9) crash on cortex

From: Evan Kelly <ebkelly.ualberta.ca>
Date: Thu, 10 May 2007 19:10:46 -0600

Hi Ross,

Yes, you are entirely correct with your first suggestion: I am using
PBC with cut=20 and PME is on (ntb=1 for this stage). I just tried
running the simulation with cut=10 and it ran successfully. So, this
is a fix. Now, my understanding is that the higher the cutoff value
the more accurate the energy values for the simulation. However, I
understand that PME does compensate for the cutoff. In your opinion,
will the change from cut=10 from cut=20 make that much of a
difference in this simulation? The goal of the simulation is to
examine protein geometry, and the actual energy values are not of
great concern. Please let me know what you think.

Thanks!


On 10-May-07, at 2:09 PM, Ross Walker wrote:

> Hi Evan,
>
> This means that an allocation failed. In this case likely the
> allocation of ipairs. Note you have an extremely large number of
> pairs (178,595,354) which suggests that you are using an extremely
> big cut off. You are running with periodic boundaries and PME
> turned on yes? In which case you only need to set cut to somewhere
> between 8 and 10.
>
> If you are running with a solvent cap and really want a large cut
> off then you will likely need to recompile sander with 64 bit
> support. On AIX (I assume you are running AIX) it is by default
> only built for 32 bit support. On AIX this limits the addressable
> memory space to 256MB (why 256Mb I have know idea - typical
> IBM...). The -bmaxdata:0x80000000 on the FC and LOAD lines in the
> config.h file increase this to 2GB but you are likely still
> exceeding this or some system wide limit. Note often by default the
> max datasize in AIX is limited to 1048575 kB by enviroment
> variables. Your 178 million long pair list takes 681 MB on it's own
> so you are probably pretty close to this limit as it is. Try
> running 'limit' to see...
>
> Anyway, compiling for 64 bit should address this although it may
> harm performance.
>
> cd $AMBERHOME/src/
> make clean
> ./configure -mpi xlf90_aix
> edit config.h and delete all occurances of -bmaxdata=0x80000000
> add -q64 to the beginning of the FFLAGS and FOPTFLAGS lines and
> after mpxlf90_r on the LOAD line.
>
> Then
>
> make parallel
>
> This should hopefully work although I have not actually tried it.
>
> All the best
> Ross
> /\
> \/
> |\oss Walker
>
> | HPC Consultant and Staff Scientist |
> | San Diego Supercomputer Center |
> | Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
> | http://www.rosswalker.co.uk | PGP Key available on request |
>
> Note: Electronic Mail is not secure, has no guarantee of delivery,
> may not be read every day, and should not be used for urgent or
> sensitive issues.
>
>
>
> From: owner-amber.scripps.edu [mailto:owner-amber.scripps.edu] On
> Behalf Of Evan Kelly
> Sent: Thursday, May 10, 2007 11:57
> To: amber.scripps.edu
> Subject: AMBER: Sander (Amber 9) crash on cortex
>
> Hi everyone,
>
> I was recently met with the following error in the output file
> after a sander crash. The simulation I called for was an energy
> minimization at the force-field level of a small (148 residue)
> protein in a larger box of water (16000 water molecules). The crash
> occurred running in parallel on 4 cpus on an IBM 595 cluster. The
> error is found below:
>
> | Memory Use Allocated
> | Real 3301101
> | Hollerith 318114
> | Integer 1271200
> | Max Pairs 178595354
> ASSERTion 'ier == 0' failed in sander.f at line 223.
>
>
> Please note that the same simulation ran successfully on my iMac G5
> with only one cpu. Does anybody have any suggestions as to what
> might cause this sort of crash? Please note that I have
> successfully run different sander simulations on this same cluster
> (QM/MM of a smaller system).
>
>
> Thanks!
>
> ----------------------------------
> Evan Kelly
> ebkelly.ualberta.ca
>
>


--------------------------------
Evan Kelly
ekelly9.gmail.com




----------------------------------
Evan Kelly
ebkelly.ualberta.ca



-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun May 13 2007 - 06:07:31 PDT
Custom Search