Firefly and PC GAMESS-related discussion club



Learn how to ask questions correctly


Re: Memory for CASSCF(8,8)

Alex Granovsky
gran@classic.chem.msu.su


Hi Pasha,

> Hi, step-by-step I converged MCSCF
> (cityp=aldet, 14 act. orb. and 14 electr) job.

This is actually CAS(14,14) not (7,7) as you designated it.

> Now i want increase active space and try 16 act orb + 16 electrons.
> I ran check run i got

And this is CAS(16,16).


>     --------------------------------------------------
>          AMES LABORATORY DETERMINANTAL FULL CI
>     PROGRAM WRITTEN BY JOE IVANIC AND KLAUS RUEDENBERG
>     --------------------------------------------------

> THE NUMBER OF DETERMINANTS HAVING SPACE SYMMETRY A  
> IN POINT GROUP C2   WITH SZ=  0.0 IS       82820900
> THE DETERMINANT FULL CI REQUIRES  1822685203 WORDS
>***** ERROR: MEMORY REQUEST EXCEEDS MAXIMUM.
>  1822696229 WORDS REQUIRED    100000000 AVAILABLE.

Note this is the memory per core, not the memory in aggregate.
Current builds of the forthcoming Firefly version 8.0.0 include
more advanced CI code that uses distributed rather than replicated
memory for CI step. It has already been tested with CAS(16,16)
and even with larger active spaces. Please contact me directly
if you are interested in getting access to prerelease binaries.


>Ok, i need setup 1823 MWORDS or 13909 Mb or ~15Gb. It is ok for me, because i am going to use 24 CPU and up to 24 Gb memory (1Gb per 1 CPU). I setup mwords to 1900 and got strange behaviour of Firefly:

> EXECUTION OF FIREFLY TERMINATED ABNORMALLY AT 10:32:21 LT  16-NOV-2010

Sure this is because Firefly cannot allocate 1900 MWords per core.
Nor it even tried to do that as it is impossible for 32-bit program.

Regards,
Alex Granovsky


[ Previous ] [ Next ] [ Index ]           Sat Nov 20 '10 12:52pm
[ Reply ] [ Edit ] [ Delete ]           This message read 856 times