Nvidia: GPU PhysX Requires Tight Collaboration Between GPUs, Rendering on ATI Radeon Not Possible

Nvidia Ceases Support of Ageia Physics Processing Units, Says “No” to Non-Nvidia GPUs for Rendering

by Anton Shilov
05/05/2010 | 10:46 PM

Nvidia Corp. recently said that it had discontinued support of dedicated Ageia physics processing units (PPUs) due to undisclosed reasons, which forces owners of dedicated PhysX accelerator cards to use Nvidia GeForce instead. In addition, the company claimed that the work of GPU-accelerated PhysX actually requires an Nvidia graphics chip to do rendering, which is why it is not possible to use a GeForce for PhysX processing and an ATI Radeon for graphics rendering.

No More PhysX PPUs


“The current PhysX driver unfortunately does not support Ageia* PPU hardware which you're already aware. Support for Ageia PPU ended after the 8.xx.xx driver but like you we've receive other users asking to include support in current PhysX releases. There are discussions to release a stand-along PhysX driver that will support Ageia PPU but that is still in discussion,” an Nvidia technical support specialist is reported to have said.

Nvidia’s own web-sites confirms the claim of the technician: Ageia PhysX processors users should use and install an older PhysX system software such as version 8.09.04 that will enable Ageia PPU acceleration version 2.8.1 SDK or earlier and Windows Vista and Windows XP only. The current version of PhysX SDK is 2.8.3, hence, Nvidia dropped support for PPUs both in Windows 7 and in the latest software development kits.

Although Nvidia’s move is not pleasant for owners of Ageia PhysX accelerators, it is completely logical for Nvidia, which stopped selling PPUs as soon as it acquired Ageia in early 2008. Usually technology companies tend to support products for about three years after their release, but the history knows a number of exceptions.

No ATI Radeon

On another occasion, Nvidia also explained why it does not enable PhysX GPU acceleration on Nvidia GeForce graphics processing units (GPUs) whenever another graphics solution – such as ATI Radeon HD 5000-series graphics board – is used for rendering. Apparently, GPU PhysX requires a tight collaboration between graphics chip that renders a 3D game and the one that does processing of physics effects.

“There are multiple technical connections between PhysX processing and graphics that require tight collaboration between the two technologies. To deliver a good experience for users, Nvidia PhysX technology has been fully verified and enabled using only Nvidia GPUs for graphics,” an official claim of Nvidia reads.

Interestingly, but until late July, 2009, and ForceWare 186.x drivers, Nvidia did not disable PhysX whenever an ATI graphics card was used for graphics rendering. Even though it was definitely not a trivial task to make both ATI Catalyst or Nvidia ForceWare drivers or Nvidia PhysX system software work on the same system on Windows Vista at the same time in order to run graphics rendering on ATI Radeon and PhysX on Nvidia GeForce or Ageia PhysX physics processing unit, some people could still use such a configuration. When Nvidia disabled such was of operation, it said there were “development expense, quality assurance and business reasons” behind the move. Enthusiasts, however, managed to re-enable GPU PhysX support on systems with ATI Radeon inside.

Quite interestingly, Nvidia fully allows using GPU PhysX acceleration on multi-GPU systems featuring a number of GeForce GPUs, which means that there are no bandwidth or any other technical issues with using one GPU only for PhysX and the remaining GPU(s) for graphics rendering. Instead, there are indeed business reasons behind disabling PhysX on ATI Radeon-powered systems.

“A new configuration that’s now possible with PhysX is 2 non-matched (heterogeneous) GPUs. In this configuration, one GPU renders graphics (typically the more powerful GPU) while the second GPU is completely dedicated to PhysX. By offloading PhysX to a dedicated GPU, users will experience smoother gaming,” an explanation by Nvidia reads.

According to the company, the minimum requirement to support GPU-accelerated PhysX is a GeForce 8-series or later GPU with a minimum of 32 cores and a minimum of 256MB dedicated graphics memory. However, each PhysX application has its own GPU and memory recommendations. In general, 512MB of graphics memory is recommended unless you have a GPU that is dedicated to PhysX.

It should be noted that there are only around fifteen games that can take advantage of PhysX processing on Nvidia GeForce graphics processors. 

*Interestingly, but an Nvidia technician referred to Ageia as to Aegia for three times, which is surprising. We decided to correct the quote.