[petsc-dev] Bad scaling of GAMG in FieldSplit

[Mark Adams]

>
>
> What does Telescope do on the fine grids? Is this a structured grid and
> Telescope does geometric MG?
>
>
> No, it’s a MatMPIAIJ. Telescope just concatenates local matrices together,
> makes some process idle (throughout the complete setup + solve), while the
> others do the work.
>

So you are reducing processors on the fine grid. You are way into strong
scaling.

GAMG has a parameter for how many equations/proc to aim for in process
reduction. So the 2nd grid will be reduced if you like. It can also
repartition these grids because concatenation is not likely to be very
good.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20180727/681eb154/attachment.html>