Parallel run Slows with Field average
 

Hallo All,
I just discovered something strange. I am doing something wrong but I am not able to figure it out.
I will first define my problem. I have a mesh of 3M cells, which I am running on 216 processors in parallel. The decomposition is hierarchical (6 6 6).
I don't have any issues with the simulation until I start fieldAveraging over time, the simulation suddenly slows down a lot. To give you an idea if 4 corrector steps without averaging were taking 2.4 s approx, and with averaging the time goes to 33-37 s approx.

And this has been bugging me because if I kept going on this pace, I will finish the simulation in 6 months.
Please help.
Regards,
Awais