[Q-e-developers] global_kpoint_index

[Paolo Giannozzi]

The reason why it is done in such a dumb way is, if I remember correctly
(cannot verify now), that it is not easy to figure out which pool computes
which rotated wavefunctions. There are utilities for pool operations
(poolscatter etc), you may take inspiration from those. Paolo

Il 18/ott/2016 10:48 AM, "Lorenzo Paulatto" <lorenzo.paulatto at impmc.upmc.fr>
ha scritto:

Nevermind,
we have one now..

By the way, I'm trying to do a broadcast from one pool to all the others;
i.e.
assuming the only a certain pool, not necessarily rthe first one, has the
correct data and I want everyone to have a copy of it.

I'm finding this annoyingly difficult, I'm a bit lost with which are the
correct
communicators to use. I would like to avoid doing a huge mp_sum, is there
any
place where it has been implemented already?

This would also be used in the EXX code: at the moment each pool rotates the
wavefcuntions for the points it has, then a massive all-to-all mp_sum is
called to broadcast the data, this looks like a massive bottleneck to me
(maybe it is not?), I'm only one step away from fixing it, but I need the
above..



On Tuesday, October 18, 2016 10:16:36 AM CEST you wrote:
> Hello,
> do you know if we have already a function that does the inverse of
> global_kpoint_index(ik,nkstot)
>
> i.e. that gives you the pool-local number of the k-point, give the global
> number?
>
> cheers


--
Dr. Lorenzo Paulatto
IdR @ IMPMC -- CNRS & Université Paris 6
phone: +33 (0)1 44275 084 / skype: paulatz
www:   http://www-int.impmc.upmc.fr/~paulatto/
mail:  23-24/4é16 Boîte courrier 115, 4 place Jussieu 75252 Paris Cédex 05

_______________________________________________
Q-e-developers mailing list
Q-e-developers at qe-forge.org
http://qe-forge.org/mailman/listinfo/q-e-developers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20161018/2096e1f2/attachment.html>