"out of Topnodes" vs "maximum number of tree-nodes reached"

From: Mark Baumann <markb_at_physics.utexas.edu>
Date: Sat, 5 Dec 2009 00:13:12 -0600 (CST)

Hello,

As I've steadily increased the number of particles in my runs, I've run
into the following error:

"We are out of Topnodes. Increasing the constant MAXTOPNODES might help"

This sounds to me like I might've run out of processor memory. Is that
the case, and if so what is the best course of action? Should I actually
change MAXTOPNODES and recompile? Or should I just increase the number of
processors?

I've tried doing the latter (increasing the number of processors) because
it seemed like the easiest solution. But then I ran into the "maximum
number [number] of tree-nodes reached" error. I presume this is because
my # of particles per processor dropped too low. So perhaps I've gone
from having too few processors to too many processors.

I've played with the number of processors and the TreeAllocFac to try to
find a happy medium to make things run, without success. Here are details
of the various combinations that I've tried:

num particles = 1e6

num processors = 256 up to 320, in increments of 16
error is "out of Topnodes"
TreeAllocFac = {0.8, 1.2, 1.5}

num processors = 336 up to 512, in increments of 16
error is "max number of tree-nodes reached" (in create empty nodes)
TreeAllocFac range = {1.2, 1.5, 2.0}

I would be very appreciative of any suggestions that might help get me off
the ground on a run containing 1e6 particles.

Thank you for the help!

Mark
Received on 2009-12-05 07:31:59

This archive was generated by hypermail 2.3.0 : 2022-09-01 14:03:42 CEST