PHYSX NEWS PHYSX SDK
PROJECTS TABLE
GPU PHYSX
GAMES INFO
PHYSX
ARTICLES
PHYSX WIKI FORUM
РУССКИЙ ENGLISH


:: Back to news index ::

The Evolution of PhysX SDK, performance-wise

with 6 comments

A quite interesting, unexpected and a little emotional article – The Evolution of PhysX – was published today by Pierre Terdiman, senior software engineer in NVIDIA and one of the developers of the original NovodeX engine.

Update: Multithreaded performance scaling in PhysX SDK

The article provides in-depth performance comparison between various versions of PhysX SDK (2.8.4, 3.2 and 3.3 Beta), using well-known open-source Bullet physics engine as as a reference point.

The performance tests were performed using PEELPhysics Engine Evaluation Lab, a specialized tool that is using within NVIDIA to research behaviour and performance of various physics engines using a set of standartized scenes.

“The Evolution of PhysX” article is devided into 13 chapters, each one filled with graphs and detailed technical explanations of the test results:

  1. Introduction and PEEL overview
  2. Rigid bodies – stacking (boxes)
  3. Rigid bodies – stacking (convexes)
  4. Rigid bodies – piles of objects
  5. Rigid bodies – compounds
  6. Rigid bodies – collisions with triangle meshes
  7. Joints
  8. Scene queries – raycasts (simple shapes)
  9. Scene queries – raycasts (meshes)
  10. Scene queries – sweep tests (simple shapes)
  11. Scene queries – sweep tests (meshes)
  12. Final results and conclusion
  13. Addendum

The results of the tests are quite revealing:

Firstly, even the relatively old PhysX SDK 2.8.4 is showing, in many cases, more consistent performance than Bullet 2.81 engine, while accused by many for being crippled and highly un-optimized for CPU execution (Please note that we are not talking about performance of GPU PhysX effects here).

Secondly, it is clear now that PhysX SDK 3.x offers stable performance improvements over the previous generation, while latest PhysX SDK 3.3  is significatly faster (sometimes by several orders of magnitude) than any of the PhysX SDKs or the competition. One more reason to sign up for SDK 3.3 Beta testing.

And finally, the conclusion from Mr. Terdiman

The point is simply, again, that contrary to what people may still believe, PhysX is actually very optimized and a perfectly fine CPU physics engine. In fact, if some competitors would not prevent me from publishing the results, I would happily show you that it often beats everybody else.

I invite curious readers to create their own benchmarks and see for themselves.

Written by Zogrim

May 12th, 2013 at 12:37 pm

6 Responses to 'The Evolution of PhysX SDK, performance-wise'

Subscribe to comments with RSS

  1. Very interesting read, surprised how well 2.8.4 holds up and for the most part and performns better than Bullet on the CPU. PhysX 3.3 is just well ahead of everything else.

      

    Spets

    12 May 13 at 5:37 pm

  2. Spets:

    Very interesting read

    Indeed, it is quite nice.

    However, it would be even more interesting to get Havok results, since Havok Physics is the main competitor of PhysX SDK, not Bullet.. unfortunately, that does not depend on us.

      

    Zogrim

    12 May 13 at 6:26 pm

  3. Zogrim: However, it would be even more interesting to get Havok results

    It would have been great if they allowed it, but I guess this statement speaks for itself:

    “In fact, if some competitors would not prevent me from publishing the results, I would happily show you that it often beats everybody else.”

    Sounds like Havoc does not want the results up for a reason :P

      

    Spets

    12 May 13 at 6:35 pm

  4. Spets: Sounds like Havoc does not want the results up for a reason

    Well, maybe, but as I have noticed, many companies do not like public performance comparisons, or, for example, public information on pricing.

    Not a big surprise here.

    Also, if SDK 3.3 “often beats everybody else” (I assume everybody is Havok 2012 here), I would really love too see Havok 2013 (“next-gen”) comparison here.

    That one should be a tough guy :)
    And maybe Havok will decide to show us how tough it is.

      

    Zogrim

    12 May 13 at 6:46 pm

  5. Maybe you could ake ypur own tests? Or the last statement is incorrect?

    “I invite curious readers to create their own benchmarks and see for themselves”

    Cant you?

      

    Timmy

    12 May 13 at 7:31 pm

  6. Timmy: I invite curious readers to create their own benchmarks and see for themselves

    I can only guess, that either Mr. Terdiman plans to release PEEL for public or simply suggests you to try and make your own version of this tool.

    Upd: PEEL won’t be available for download at this time

      

    Zogrim

    12 May 13 at 7:42 pm


Leave a Reply

*
Copyright © 2009-2014. PhysXInfo.com | About PhysXInfo.com project | Privacy Policy
PhysX is trademark of NVIDIA Corporation