Written by Thomas Gribble | Original posted here

When AGEIA introduced PhysX technology to the world in 2006, the reception was not exactly warm. At a time when prices of processors and graphics cards soared to ridiculous levels, the thought of spending an extra chunk of change on a card that didn't appear to do very much was a difficult one to swallow. There was a marked lack of notable games that supported the technology, competing physics APIs that were showing a lot of promise, and an overall unwillingness of the industry to accept that in-game physics was indeed going to be the next big thing. Of course, there were champions of the technology; in fact, it is very likely that there was no bigger advocate of AGEIA's PhysX than yours truly.

I raced through my review of the sample that was sent to me with shear excitement and, in light of the less-than-favorable results that I obtained in my first round of testing, remained optimistic. My second round of testing turned out to be everything I had hoped it would be. Running the demo of a new game called CellFactor on PhysX technology was absolute bliss. Not only was this groundbreaking technology at my fingertips, but the game play I experienced could only be described as revolutionary. Indeed, the future of in-game physics and this relatively inexpensive piece of hardware from AGEIA seemed almost unimaginably bright. Granted, a huge amount of my optimism for the technology was based on the promises I had received from various AGEIA personnel stating that adoption into mainstream games was right around the corner. There was an impressive list of developers who were on board with the technology, and the games they were sure to create would almost certainly propel this company and the product it produced into the stratosphere.

So after being probably the strongest supporter of PhysX technology, why am I writing this article? How did my opinion of PhysX change so dramatically? Well, the answer to these questions requires compound response. My love for PhysX reached its absolute peak when Immersion Games released CellFactor: Revolution in May of 2007. The game provided everything that I could possibly want: Stunning graphics, unprecedented environmental interaction, revolutionary game play, and an absolutely visceral fragging experience. It was certainly not the best game I had ever played. Nor did I pull any punches when criticizing its flaws. But the game was free. The game demonstrated the power of in-game physics and PhysX technology itself. If nothing else, CellFactor: Revolution would be a beacon to which all developers could look to when producing their own physics-enabled games. Or so I thought.

Since May of 2007, there have been several major games released that utilized PhysX technology. By far the most notable of these games was Unreal Tournament 3. UT3 was one of the most anticipated titles to launch in 2007. I was particularly anxious for the release not only because I was a fan of the franchise and wanted it to be a good game, but I also sincerely hoped that this would be the savior of AGEIA's PhysX technology. Unfortunately it was neither. The game turned out to be pretty damn good, but it was not adopted into competitive play with nearly the fervor its predecessors were - a testament to its longevity (or lack thereof). It was, however, a terrific implementation of PhysX technology. In PhysX levels, you could go through areas and blast the hell out of floors, ceilings, and walls in order to create shortcuts from place to place. You could use the shock rifle's secondary attack to pummel an enemy with upended debris as well. This was reminiscent to CellFactor and I was pleased to see it done so well (see images and clips @ PC Perspective). Unfortunately, this experience could only be enjoyed in two levels of the game. Additionally, these two levels of the game did not actually come with the game itself; you had to apply a patch that basically downloaded them as extra content. Moreover, you could only play these levels if you had the actual PhysX hardware. So, although the PhysX implementation in the game was pretty cool, it was only being exposed to people who A) had a PhysX card and B) had the wherewithal to download the additional content. Limited exposure plus limited implementation do not a savior make.

Other games have had PhysX as well. These games, however, do not really use the technology in way that affects the actual game play experience rather than just add to the visual effects. Effects augmented by physics processing include things like particle systems, volumetric fog, cloth behavior, debris manipulation, and extended environmental interaction. So called Effects Physics is the most frequented application of PhysX technology.

NVIDIA, who bought AGEIA in early 2008 in what even I hailed as a brilliant move, has completely eliminated the necessity of a dedicated physics processing unit. NVIDIA has converted PhysX into a GPU-based technology that runs with the aid of CUDA, NVIDIA's GPGPU language. While getting rid of the hardware requirements was probably a step in the right direction, the real hope that came out of the acquisition was that NVIDIA would be able to use its presence in the gaming world to persuade more and more developers to implement PhysX into its upcoming titles.

Perhaps the first significant to title to incorporate PhysX technology since NVIDIA took over is a game called Mirror's Edge. The free-running game takes place mostly on rooftops and involves ninja-like manipulation of the landscape. It is a fascinating game that I had the opportunity to play on the Playstation 3 when it first came out. It's adaption on the PC bumps up the graphics and brings PhysX into the mix. The problem, of course, is that it is all effects physics. There is a slight impact on game play, insofar as lines of sight can be impaired, but other than that… Well, just take a look for yourself:



In order for PhysX, or hardware accelerated physics in general, to be successful there is going to need to be a movement by game developers to incorporate the technology into games in a way that actually affects game play. Having effects physics in games as optional eye candy is not going to promote a large scale adoption by consumers, especially when they see videos like the one shown above that make it very clear that the impact is insignificant. Yes, the effects physics being produced now can increase the graphical splendor of a game. And yes, there is definitely something to be said about the overall appeal of more realistic games. However, in the long term there will be limited success for API’s like PhysX that necessitate hardware acceleration when competing standards can replicate the same sort of effects through software. Note to NVIDIA and conspiring game developers: unleash the real power of PhysX if you want it to be successful.

Developing games that really take advantage of PhysX is surely not an easy task. The added difficulty here will likely increase production time, which will probably put a damper on NVIDIA’s plans for PhysX. In light of the company’s recent announcement that major game developers EA and Take Two will be integrating PhysX into upcoming titles, it is definitely possible that the technology will become more widespread over time. While it is nice to see the GPGPU industry moving forward, the argument for buying NVIDIA video cards over superior offerings from AMD solely for their PhysX capability will remain weak until we start seeing some really strong titles emerging that make use of the technology. AMD’s commitment to Intel’s Havok and its software accelerated physics could prove more beneficial in the short term and provide some serious competition to PhysX when newer titles materialize. Until then, I have one question for NVIDIA about PhysX: Where’s the beef?