AMBER Archive (2002)

Subject: Re: Memory and swap in sander

From: David A. Case (case_at_scripps.edu)
Date: Mon Aug 12 2002 - 10:55:55 CDT

On Mon, Aug 12, 2002, William Wei wrote:

> Hi, Amber funs, I run parallel sander on Onyx. A 16 processors and 8
> processors parallel jobs and two other one processor minimization. They cost
> about 28G swap, but only 3.0 G memory. In my computer I have totally 35G
> memory, I want to use more of them to speed up my dynamics simulation. Can
> anyone give ma a suggestion for how to modify the source file(s), and how
> much memory should I create for the sander? Thanks, and have a good day.
>

Except on really gigantic systems, Amber doesn't require very much memory.
Memory usage is printed out near the beginning of sander runs. Reals take
8 bytes, integers 4 (on most machines, i.e. unless you compiled with -64
on the Onyx.)

>From the distribution:

  DHFR + water (23000 atoms): about 45 Mbytes
  rt_polymerase (141000 atoms): about 300 Mbytes

These are for a single processor; values will be somewhat smaller for
parallel jobs.

If we knew of any code modifications to make that would speed up simulations,
we would have already put them in :-)

..dac 

-- 

==================================================================
David A. Case                     |  e-mail:      case_at_scripps.edu
Dept. of Molecular Biology, TPC15 |  fax:          +1-858-784-8896
The Scripps Research Institute    |  phone:        +1-858-784-9768
10550 N. Torrey Pines Rd.         |  home page:                   
La Jolla CA 92037  USA            |    http://www.scripps.edu/case
==================================================================