[gmx-users] PME vs Cut off

[Hector Mtz-Seara]

Hi All User,
I'm simulation monolayer and to do it I put vacuum between the two sides 
of the system. Then you obtain:

        VACUUM
------------------------------
LIPID MONOLAYER
------------------------------
          WATER
------------------------------
LIPID MONOLAYER
-------------------------------
        VACUUM

Until today I was using cut off, but I want to use know PME as it seems 
to be more accurate. Then after reading the manual and some FAQ'S, there 
are some questions that I want to do.

First In a parallel environment  8 processors ( 4*(2hyperthreading)) is 
faster the cut off method or PME?
Which do you recommend?
Which are the correct value for fourierspacing, pme_order?
and the most important, as I don't want that the monolayer can interact 
by the vacuum which high must be use between monolayers to avoid 
interaction? and I have to use ewald_geometry 3d or 3dc? because I don't 
want to avoid all interactions among z direction only the interaction 
between the two monolayers.

With regards,

Hector