Physical vs Non-Physical movement
|
|
Ralph Doctorow
Registered User
Join date: 16 Oct 2005
Posts: 560
|
12-04-2006 18:53
Has anyone done any lag testing of moving stuff via multiple llSetPos vs the physics engine?
|
|
grumble Loudon
A Little bit a lion
Join date: 30 Nov 2005
Posts: 612
|
12-06-2006 04:24
In sims like mine where there are too many scripted prim updates you are better off using the physics engine as the client only needs to be told how fast the object is moving when it changes speed.
|
|
Ralph Doctorow
Registered User
Join date: 16 Oct 2005
Posts: 560
|
12-06-2006 07:19
From: grumble Loudon In sims like mine where there are too many scripted prim updates you are better off using the physics engine as the client only needs to be told how fast the object is moving when it changes speed. Well, surely the server has to do a lot of processing for physical objects motion, it can't just be a client side rendering. Collisions for example wouldn't work, and all the other physics characteristics would have to be calculated on the server I'd think. |
|
Tyken Hightower
Automagical
Join date: 15 Feb 2006
Posts: 472
|
12-06-2006 07:48
From: Ralph Doctorow Has anyone done any lag testing of moving stuff via multiple llSetPos vs the physics engine? Why not try doing some tests and analyzing it yourself, then post a little of the results here?  |
|
Ralph Doctorow
Registered User
Join date: 16 Oct 2005
Posts: 560
|
12-06-2006 08:12
From: Tyken Hightower Why not try doing some tests and analyzing it yourself, then post a little of the results here? Oh, sorry to disturb you, apparently asking for group knowlege is frowned on. I had somehow thought that was the purpose of this forum. I'd suggest you just put me on your ignore list. |
|
Sterling Whitcroft
Registered User
Join date: 2 Jul 2006
Posts: 678
|
12-06-2006 14:30
Ralph,
I don't think Tyken was being snooty.
He agrees with your curiousity--this seems to be an untested question.
If either you, Tyken, or Grumble test this out, I'd be interested in knowing, too.
|
|
Eloise Pasteur
Curious Individual
Join date: 14 Jul 2004
Posts: 1,952
|
12-07-2006 02:18
I posted a reply to this yesterday that seems to have gotten chewed.
I've not done it in a while, the last time I tested sims weren't clamped to a 45 fps max, so things might be different now.
Anyway, the answer was then it depends on your location.
In an essentially empty sim physics hurts more than llSetPos.
In an otherwise script-busy sim with no physics, it's hard to see a difference between the two methods.
Physics seems to have a non-linear relationship between the number of physical objects and impact on the sim - the more there are the higher the impact of adding an additional extra one. I guess that's reasonable because the number of potential impacts rises non-linearly too.
All that said, I'd love to see the numbers if someone does the testing under the new conditions.
|
|
Lazink Maeterlinck
Registered User
Join date: 8 Nov 2005
Posts: 332
|
12-07-2006 02:47
I believe if you made it either volume detect, or phantom, the physics engine doesn't have to work so hard since the collision box is not really calculated on those prims.
|
|
Ralph Doctorow
Registered User
Join date: 16 Oct 2005
Posts: 560
|
12-07-2006 08:32
It's interesting that scripts that are in the .2 sec timeout are still listed as "Active" AFAICT. I kind of doubt that they take much sim time, but who knows?
A priori I'd think that the llSetPos would be less of a load on the sim since it doesn't have to do collision detection with non-physics stuff, and there is a lot less overhead with buoyancy, drag, etc. But it clearly has more script overhead.
I'm still playing with this.
|
