The ABINIT Users Mailing List ( CLOSED )

[Fabien Bruneval <fabien.bruneval@polytechnique.fr>]

Hat trick! Three identical answers within 10 minutes! It seems that the 
GW team is perfectly self-consistent.

Fabien



Matthieu Verstraete wrote:
> Hello Juanma,
>
> Your cell has 92 atoms! I'll bet the KSS writing part is an "all in 
> memory" code, and you're exploding your RAM. The estimation for the 
> memory needed for abinit to run may also not take the kss writing step 
> into account. PMA's suggestions are the best: compile+run with 
> debugging, and compile option -C to see if memory is being overrun
>
> Perhaps someone from the GW team can confirm if mkmem 0 works in the kss 
> writing step. In that case, you can set it and the code should page the 
> wfct to disk during the writing of the kss. Still, GW on top of that 
> will be biblical.
>
>
> Matthieu