[Opensim-dev] Problem with the bots

[Mister Blue]

We added a new command to TestClient to create a swarm of independently
acting bots (http://www.sciencesim.com/wiki/doku.php/vwperf/testclient_swarm<%28http://www.sciencesim.com/wiki/doku.php/vwperf/testclient_swarm>).
The modified TestClient is distributed with the tests. Email me for a patch
if you want to build sources.

The new TestClient command creates one bot which randomly wanders within a
defined area. Each of the bots is managed separately and independently
(motion, direction, target). As mentioned before, we start 25 if these
independent bots per TestClient instance.

Our goal is to create a close to realistic load on the simulator so it is
important to drive the bots from a client program. If you are not interested
in load testing the simulator, running bots from a region module is a
solution others have used.

-- mb

On Mon, Sep 6, 2010 at 9:07 PM, <it at nii.ac.jp> wrote:

> Hello,
>
> Thanks a lot for the answer Dan. Now I know what to expect. And I have
> another question.
> You are talking about independant bots, do you mean that you can control
> each bot in a single Testclient instance independantly ? (for instance, make
> a bot move forward while another move backward, or don't move ?)
>
> To do so, I wasn't using testclient put the opensim Scenepresence Object
> (scenepresence.controlingclient) that enable to actually have indepedant
> bots.
>
> I would like to know how independant are you bots.
>
> Thanks you very much
>
> IT
>
>
>
>
>
>
> It takes TestClient 5-15 minutes to bake 25 clients if they are all logging
> in at the same time (depends on hardware and other load on the client box).
> In our experiments, we use several multi-core servers each running multiple
> TestClient processes where each process hosts 25 bots. We can usually log in
> 500-1000 bots to our grid within 20 minutes and get all of them fully baked
> and ready for action. Instead of running 100 bots in a single TestClient,
> consider running 4 TestClients and have each one connect 25 bots. TestClient
> uses lightweight threads to handle packets from the simulator and if you
> have 100 bots moving around then there are a LOT of packets to process. A
> machine with more cores or hardware threads will improve your experience
> dramatically when simulating lots of independent bots in TestClient.
>
> ~Dan
>
> -----Original Message-----
> From: opensim-dev-bounces at lists.berlios.de [mailto:
> opensim-dev-bounces at lists.berlios.de] On Behalf Of it at nii.ac.jp
> Sent: Monday, September 06, 2010 3:01 AM
> To: opensim-dev at lists.berlios.de
> Subject: Re: [Opensim-dev] Problem with the bots
>
> Hello,
>
> Actually I would like to know a little bit more about the test you have
> done to assess the number of bot that can be runned on a simulation.
> My main concern is the time we have to wait to log several bot.
>
> Indeed, I can log bots and ensure their appearance will be Ok, but it
> requires a lot of time. (in OpenSim 0.7.0), for instance, to log 25 bots, it
> takes me almost 10 minutes if I want everybody to be baked.
>
> I am using the Testclient Object (inherited from GridClient).
> What about the time it took you to have 100 avatars on a sim ?
> And another question, according to
> http://www.sciencesim.com/wiki/doku.php/vwperf/profiling_069_070
> The test have been made on 6.9  ?
> What about 7.0 ?
>
> Thanks a lot for your help
>
>
> IT
> _______________________________________________
> Opensim-dev mailing list
> Opensim-dev at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/opensim-dev
> _______________________________________________
> Opensim-dev mailing list
> Opensim-dev at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/opensim-dev
> _______________________________________________
> Opensim-dev mailing list
> Opensim-dev at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/opensim-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20100907/2384adfc/attachment-0001.html>