AMBER Archive (2005)

Subject: RE: AMBER: AMBER8 limitation of memory in QM/MM calculation

From: Ross Walker (ross_at_rosswalker.co.uk)
Date: Mon Aug 01 2005 - 21:52:29 CDT


> allocate( x(lastr), ix(lasti), ipairs(lastpr),
> ih(lasth), stat = ier
> )
>
> ! allocate( x(lastr), ix(lasti), ipairs(lastpr), ih(lasth) )
>
> REQUIRE( ier == 0 )
>

Did you really remove the stat = ier? That is a complete hack that is not
going to help you at all. The allocate command returns the result of the
allocation in the ier variable. You can't just ignore it and hope things
will be ok. If it didn't manage to allocate the memory all that will happen
if you ignore the ier return value is that sander.QMMM will segfault with an
obscure error message later on.

> After I try to printout of lastr, lasti, lastpr and lasth.
> I got these
> values.
>
> 1621806 1081568 1132237491
                       ^^^^^^^^^^
                       !
------------------------
|
This suggests that you need 113,223,749 integers to hold the pair list. In
total on a 32 bit machine this allocation statement represents (1621806*8)+
(1081568*4) + (113223749 * 4) bytes of memory = 448.41 MB of memory. This is
just for the arrays in allocation statement, you will need more for the
other arrays. Note this inefficient use of memory is due to the way the
AMBER8 pair is handled. You will need to make many more changes to the code
to get it to work efficiently with the pair list as it is currently
implemented.

> I still can not run QM/MM MD because of limitation of
> memery. Can
> anyone suggest me how I can solve this problem. I'm
> appreciate for your
> suggestion.

There is not much you can do as there are many other limits in the amber 8
QMMM code. When I originally looked at improving things for amber 9 I
eventually gave up trying to remove all the limits from the QMMM code and
instead re-wrote it from scratch.

The only real solutions you can try for the moment are reducing the size of
your cutoff. Reducing the number of waters. Restricting your simulation to
just part of your system and/or trying things on a 64 bit machine with lots
of memory.

I doubt the last option will help but it is worth a try.

All the best
Ross

/\
\/
|\oss Walker

| Department of Molecular Biology TPC15 |
| The Scripps Research Institute |
| Tel: +1 858 784 8889 | EMail:- ross_at_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 ready every day, and should not be used for urgent or sensitive issues.

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu