[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Model Performance Under "GO" vs. "Step"
From: |
Rob Kewley |
Subject: |
Re: Model Performance Under "GO" vs. "Step" |
Date: |
Thu, 23 Oct 1997 08:18:13 -0400 |
Jim Kresl wrote:
>
> Hello,
>
> I've been working with the E. Nelson version of the Village model, and
> have encountered performance problems that differ depending on whether I
> use "Go" or "Step". Granted the model is somewhat large (400x454
> cells with ~150 agents), but not unreasonably so. "Go" ultimately
> results in a crash where each time step takes considerably more time
> than the one previous (~10% increase each step), even though the number
> of agents and their interactions remain constant. Using "Time Step",
> the model runs through to completion and each step takes about the same
> time to finnish --( within +/- 5%).
>
> I've checked for obvious memory leaks, or other design quirks, but given
> my limited experience, I'm a bit confounded. If anyone has run into
> this situation before, or perhaps worked with Erik Nelson this past
> summer and could offer a suggestion or explanation, I would surely
> appreciate it.
>
I had a similar problem earlier this summer which was caused by a bug in
Swarm scheduling operations. See the archive entry below for the fix.
This bug is fixed in Swarm-1.0.3.
http://www.santafe.edu/projects/swarm/archive/list-archive.9708/0061.html
==================================
Swarm-Support is for discussion of the technical details of the day
to day usage of Swarm. For list administration needs (esp.
[un]subscribing), please send a message to <address@hidden>
with "help" in the body of the message.
==================================