Firefly and PC GAMESS-related discussion club


 
Learn how to ask questions correctly  
 
 
We are NATO-free zone
 



Re^9: Parallel geometry optimization with XMC-QDPT2

Alex Granovsky
gran@classic.chem.msu.su


Dear Dawid,

you need to use the following keywords:

 $smp np=number_of_threads_to_use smppar=.t. $end

and optionally:

 $smp httnp=1_or_2 mklnp=number_of_threads_to_use_in_dgemm_code $end

$smp group is fully documented in current manual, you'll find detailed
comments on these keywords there.


Normally, np should be set to the number of processes in xp group.

Hope this helps.

Best regards,
Alex Granovsky



On Thu Dec 29 '16 4:26pm, Dawid wrote
-------------------------------------
>Dear Professor Granovsky,

>>There is nothing wrong with your calculations.
>>Displacements are distributed between all xp groups, but only grand
>>master process (this process always belongs to the first xp group)
>>is allowed to write to the output file. Hence output file contains
>>a log of work performed by the first xp group only.
>
>
>Thank you!

>>By the way, do you know that MCQDPT and XMCQDPT code is capable
>>to run using multithreading? For optimal performance, each xp group
>>must be run in smppar mode. Do you use these features?

>I was not aware of that. How to use these features, then?

>Best regards,
>Dawid Grabarek

>


[ Previous ] [ Next ] [ Index ]           Fri Dec 30 '16 3:17pm
[ Reply ] [ Edit ] [ Delete ]           This message read 154 times