AMBER Archive (2004)

Subject: Re: AMBER: NMODE COMPILATION

From: Anselm Horn (Anselm.Horn_at_biochem.uni-erlangen.de)
Date: Mon Mar 15 2004 - 05:34:50 CST

Dear Obdulia,

> I have changed parameter (MAXMEMX=350000000) in sizes.h, but I receive
> the following message:
> [...]
> real*8 x (MAXMEMX)
> ^
> Array `x' at (^) is too large to handle
> _nmode_.f: Outside of any program unit:
> _nmode_.f:50: size of variable `store_x__' is too large
> make: *** [nmode.o] Error 1/
>
> I have read about a related problem in previous messages, where it is
> advised to change compiler.
>
> I have tried with g77 and ifc (7.1), but the problem remains.
>
> Could anybody help me, please?
>
> (My PC has 2 Gb Ram and 1 Gb swapp memory).

You seem to have run into the 2 GB limit inherent to 32bit systems.
When you try allocating a double precision variable of size 350000000 with the
Intel compiler it complains (example program):

Error 530 : In program unit LARGE the size of array A exceeds the implementation
limit (2**31-1)

Comparing the numbers yields: 2^32 - 1 = 2,147,483,647 whereas 350,000,000 * 8 =
2,800,000,000. So, if you reduce the MAXMEMX parameter appropriately (e.g.
260,000,000 * 8 = 2,080,000,000), it should compile.
If this is not sufficient memory, you have to move to 64bit systems ...

Regards,

Anselm 

-- 
Abteilung f. Bioinformatik
Institut f. Biochemie
Friedrich-Alexander-Universitaet Erlangen-Nuernberg
Fahrstrasse 17
91054 Erlangen
Tel.: +49-9131-85-24682
Germany

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