ABINIT Discussion Forums

Hi all,
I remember parallelisation over dataset was considered at some point in Abinit, but I've never heard of it again. Is it still among the goals, or has it been abandoned ?

Thanks,

Emmanuel

It was working partially but never in production afaik. It is less interesting, as it is somewhat trivially parallelizable. It's for comfort only, you can run the datasets as different batch jobs and it will be the same. K-point and band parallelizatino are much more interesting as they allow you to scale and to reduce memory per core.

ciao


Matthieu