No domain decomposition that stays within memory bounds is possible

From: Eric Emsellem <emsellem_at_obs.univ-lyon1.fr>
Date: Thu, 24 Apr 2008 15:34:19 +0200

HI,

I am setting up some basic tests for Gadget2, using reasonable initial
conditions for an isolated galaxy (gas+stars). I am using 4 proc (for an initial
test) and it runs fine until ~ step 200:

==> After about 38 Myr = 213 steps, I get a:

# Begin Step 213, Time: 0.0380859, Systemstep: 0.000244141
# domain decomposition...
# NTopleaves= 470
#
# No domain decomposition that stays within memory bounds is possible.

I am using 100 000 stellar (disc/bulge) particles and 15 000 gas ones.

I guess I must be doing something basically wrong/naive with the input
parameters. If someone can help me on that..
(I have checked the allocated memory and I am factors of 10 above the required
memory for the number of particles I am using so far)

thanks a lot in advance!

Eric Emsellem
Received on 2008-04-24 15:32:30