AMBER Archive (2004)

Subject: Re: AMBER: memory prolem of the nmode calculation

From: David A. Case (case_at_scripps.edu)
Date: Tue Nov 30 2004 - 19:40:29 CST

On Tue, Nov 30, 2004, Xiao Jian Tan wrote:
>
> When I ran nmode calculation, it gave the following error message:
>
> -------------------------------------------------
> Total memory required : 406117959 real words
> ASSERTion 'ier == 0' failed in nmode.f at line 105.
>
> In fact we do have enough memory (4G), should be enough for this system.

Note that 4 x 10**8 double precision variables will take up 3.2 Gbytes
of memory. It certainly seems like that much might not be free on your
system.

You could create a simple program to allocate memory and see how much you
can grab, and whether it depends on other tasks running on your system.
You could also try to run smaller nmode jobs to see if they will work,
again seeing how large a system can really be handled.

If you don't need the eigenvectors, you could try setting ismem=1: that should
require somewhat less memory.

...good luck...dac

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