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


:: Back to news index ::

PhysX SDK 3.3 Closed Beta Testing begins

with 4 comments

PhysX SDK 3.x, topped by latest ’stable’ SDK 3.2.4 release, represents in many ways better and now, after three releases, relatively mature alternative to a prooven PhysX SDK 2.8.x branch.

Update: PhysX SDK 3.3 Beta is available for public

Recently, the PhysX SDK team began to offer a preview of the upcoming version, PhysX SDK 3.3, to advanced PhysX users — professional developers, who have the time and experience to try out the latest offering, test it and provide feedback to the PhysX SDK team.

If that describes you or your team, do not hesitate to contact PhysXlicensing@nvidia.com and use the words ‘beta-3.3 request’ in the subject line to apply for the SDK 3.3 Closed Beta Testing.

PhysX SDK 3.3 – Feature Highlights

Performance and stability optimizations for rigid body solver

Rigid body collision performance was improved up to 15-20% in comparison to SDK 3.2, while memory footprint was reduced.

Please note that we expect more performance improvements in final release

Alternative broad-phase algorithm – Multi Box Pruning (MBP)

Multi Box Pruning is now offered as an alternative broad-phase algorithm to Sweep And Prune (SAP). MBP shows improved performance and reduced overhead when all objects in scene are moving or when large numbers of objects are inserted.

Please note that we expect more performance improvements in final release

However, the benefit of MBP shows up primarily in scenarios where many objects are awake and moving simultaneously; in cases where most objects are sleeping most of the time, the standard SAP configuration will provide better baseline performance.

Also, it is worth noting that to use MBP, the application will have to make the extra step of configuring the world grid and bounds.

Updated Persistent Contact Manifold (PCM) contact generation mode

Persistent Contact Manifold, a fully distance-based collision detection system, is now often faster and more robust than the available legacy collision detection path.

Improved behavior and performance of Continuous Collision Detection (CCD)

Continuous Collision Detection algorithm has undergone major optimizations, and as a result the behavior of the CCD system has improved, while processing overhead has been reduced.

In addition, CCD now supports contact modification and notification.

More robust scene queries and contact reports

Scene Query performance was significantly improved in a variety of scenarios, while the API underwent major changes in order to improve consistency.

Please note that we expect more performance improvements in final release

Support for sharing shapes among rigid bodies

When the scene contains many instances of actors with identical geometry, the new shape sharing feature will reduce the memory costs of the simulation.

However, shared shapes have a very strong restriction – one cannot update the attributes of a shared shape while it is attached to an actor.

Enhanced cloth solver

PhysX 3 cloth solver has received several new features, including: support for self-collision and inter-cloth collision, new parallel GPU solver, anti-stretch tether constraints, triangle mesh collider support and more.

GPU acceleration improvements

The GPU code was ported to CUDA 5 and now supports Kepler GPUs (SM version 3.0 and 3.5). In addition, both cloth and particle solvers now support CUDA-Graphics interoperability feature, which can be used to increase the performance of the GPU simulation by eliminating the unnecessary CPU kernels.

SIMD library now supports ARM “NEON” extensions

NEON SIMD instructions, specific for ARM processors, are now utilized by the PhysX SDK to improve performance on ARM-based mobile platforms.

And many more, including bug-fixes and minor improvements.

Written by Zogrim

May 3rd, 2013 at 12:28 am

Posted in PhysX SDK

Tagged with , ,

4 Responses to 'PhysX SDK 3.3 Closed Beta Testing begins'

Subscribe to comments with RSS

  1. Looking great :D

      

    Spets

    3 May 13 at 12:23 pm

  2. what processor were the tests run on?

      

    physxkid

    3 May 13 at 2:28 pm

  3. physxkid: what processor were the tests run on?

    Those particular ones – i7 2600K

      

    Zogrim

    3 May 13 at 2:54 pm

  4. Will this run with c++ 11, or do i have to keep using “PhysX-3.2.1 (13457563)” that somehow works with it?

      

    Santiago Pacheco

    13 May 13 at 8:47 pm


Leave a Reply

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