[ODE] PhysX card

Misha Smelyanskiy mmaloff2002 at yahoo.com
Fri Jun 16 11:53:40 MST 2006

I am working with Ageia SDK. I am trying to run wall desrruction demo that runs perfectly well in ODE as well as PhysX software, but when I go above 2000-2500 bodies and try to run it on PhysX H/W it segfaults for some reason. Does anyone know what is the limit on # of bodies I can run in PhysX hardware and how to change this limit. My PhysX card has 128M of RAM.

----- Original Message ----
From: Jon Watte (ODE) <hplus-ode at mindcontrol.org>
To: pspcontest at erwincoumans.com
Cc: Misha Smelyanskiy <mmaloff2002 at yahoo.com>
Sent: Thursday, June 15, 2006 7:27:06 AM
Subject: Re: PhysX card

I see; that's too bad. I understand why raycasts wouldn't be accelerated 
(too serial), but triggers use pretty much the same callback interface 
as regular penetration constraints, so no accelerating that seems like a 
loss :-(

Thanks for the info -- are you just working with the SDK, or doing 
something for Ageia?


            / h+

pspcontest at erwincoumans.com wrote:
> Triggers (to use for detectors) are not accelerated in PhysX 2.4, 
> neither are raycasts, character control and vehicles (and a bunch of 
> other things).
> I will be tricky to isolate the cost of collisions (independent from 
> constraint solver).
> Erwin
> Jon Watte (ODE) writes:
>> If I recall correctly, the Ageia SDK allows you to create entities 
>> that are only colliders, not physical (to use for detectors, etc) -- 
>> you can create your scene entirely made out of these things, and see 
>> how fast it runs. I believe there will be significant warm-starting 
>> gains, though.
>> Cheers,
>>           / h+
>> Misha Smelyanskiy wrote:
>>> Hello I recently purchased Ageia PhysX card, and I am trying to use 
>>> SDK 2.4 to get a sense of how the card performs on various physics 
>>> benchmarks, compared to P4. I would like to see a profile about how 
>>> much time is spent on narrow phase collision detection as well as 
>>> physics, when both are run on PPU. Has anyone succeeded in gathering 
>>> this information? If so, can you please recommend how to do this?
>>> Thanks,
>>> Mike
>>> ------------------------------------------------------------------------
>>> _______________________________________________
>>> ODE mailing list
>>> ODE at q12.org
>>> http://q12.org/mailman/listinfo/ode
>> _______________________________________________
>> ODE mailing list
>> ODE at q12.org
>> http://q12.org/mailman/listinfo/ode
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://q12.org/pipermail/ode/attachments/20060616/c8eed25f/attachment.htm

More information about the ODE mailing list