Re: The limit of atom number on Blue Gene/L

From: Fred (Rui FENG) (fengruifree_at_gmail.com)
Date: Mon Jan 19 2009 - 17:42:03 CST

Hi Peter,

Thank you for your reply.

I tried to generate the compressed PSF but failed. I downloaded a
binary version for AMD64 and added the "genCompressedPsf on" in the
configuration file. When I ran the command ./namd abc.txt (the conf
file name), an error occurred:

ERROR: The following variables were set in the
ERROR: configuration file but are NOT VALID
ERROR: genCompressedPsf
FATAL ERROR: ERROR(S) IN THE CONFIGURATION FILE

It seems that the pre-compiled version of NAMD doesn't support the
parameter "genCompressedPsf", but in the wiki it says the function has
been integrated in the binary. Do you think it's necessary to compile
the source code? Thank you!

Fred

On Mon, Jan 19, 2009 at 8:41 AM, Peter Freddolino <petefred_at_ks.uiuc.edu> wrote:
> Hi Fred,
> you should be able to simulate meaningful systems that are a bit larger
> by using NAMD's compressedPsf features; please follow the directions at
> http://www.ks.uiuc.edu/Research/namd/wiki/index.cgi?NamdMemoryReduction
>
> Fred (Rui FENG) wrote:
>> Hi All,
>>
>> Is there anyone who tested the limit of atom number on Blue Gene/L? My
>> answer is 1 million (no bonds) on 512/1024/2048 nodes, no matter the
>> scale of partitions. For example, when I submit a job that has more
>> than 1 million particles (say, 1.1 million) on 512 nodes with VN mode,
>> the NAMD crashes. But When I submit the same job on 1024 nodes or 2048
>> nodes, it still crashes.
>>
>> Can you give me some experience about how to run millions of particles
>> (no bonds) on Blue Gene/L? Is there any compile option or argument I
>> can set? Thank you so much!
>>
>> Have a nice day!
>>
>> Fred
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:50:23 CST