8 bytes for reals (Real Max Rstack), 4 bytes for everything else, no
guarantees that absolutely every byte allocated gets accounted for (I am not
an expert on this in sander; in pmemd I don't yet attempt to account for
stack allocations, but know it is really easy to alloc/dealloc something
deep in the code without keeping track of the fact that it bumps up the
total memory requirement). I would consider it a nice first level estimate.
Regards - Bob Duke
----- Original Message -----
From: "Ed Pate" <pate.math.wsu.edu>
To: <amber.scripps.edu>
Sent: Friday, April 14, 2006 6:45 PM
Subject: AMBER: amber memory usage
> Dear Amber users:
>
> I am trying to determine the memory usage from an Amber simulation as I
> try to move the simulation to another computer.
>
> The relevant part of the Sander output (I think) is
>
> | Memory Use Allocated
> | Real 7390458
> | Hollerith 683261
> | Integer 8078098
> | Max Pairs 30967142
> | Max Rstack 7500000
> | Max Istack 5000000
> | Duplicated 0 dihedrals
> | Duplicated 0 dihedrals
>
> Do I determine useage as (Real+Hollerith+Integer)*8 ?
>
> Thanks for the help.
>
> Ed Pate
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Apr 16 2006 - 06:07:07 PDT
