bit-tech revists PhysX in UT3

This isn't a news story as such, but more of an update on a review we did a little while ago.

Back at the end of last year, we reviewed the phenomenal Unreal Tournament 3, with a page of our extensive review focusing on the Ageia PhysX content which is supported through the game. We took a very close look at the two PhysX-accelerated levels which Ageia has helped design to create scenarios which fully showcase the capabilities of the PhysX hardware.

Our impressions of the levels weren't very good. We found the games to have very poor performance, averaging around 7 FPS, and the PhysX content itself failed to mesh with the game world cohesively, in our opinion.

Shortly after the review was published, Ageia contacted us to let us know that there was a bug in the game which meant that the PhysX content may appear to be enabled, when in fact it isn't. Ageia was content with of our personal opinions of the PhysX content, but was concerned we were not getting optimum performance. In an effort to bring our readers the most honest, reliable and up-to-date content, we agreed to have a second look at the game - but were unable to find out why the performance was so low.

Ageia and bit-tech continued to look into the matter, adding updates to the original article when needed. Things were inevitably delayed a little over the Christmas and New Year period, but we eventually tracked the problem down to the hardware itself. Components of any type can only take so much stress and it seemed that at some point over its lifetime, our PPU reached the limit and failed. Thankfully, Ageia sent us a new card and we were ready to resume our investigation.

We set up a new system to test to the game in, one with proven hardware which matched the recommended system specs for the PhysX levels, installed the PhysX card and had another look at Unreal Tournament 3. This time round the performance was dramatically improved and the framerates increased to a smoother 25 - 30 FPS, though still with occasional stuttering.

The actual PhysX effects however still looked the same and the look of the game matched our previous opinions;

"Firing three rockets into the planks would often only damage the single plank directly hit, not the planks all around it...if you did hit the plank then the debris which came out was always odd looking. It was all small bits, oddly coloured to be slightly lighter than the surroundings and there was always far, far too many of them."

However, what had previously been said about the performance of the game was now revealed not to be a problem with the PhysX hardware itself, but with the specific card we were using. The faulty card is now sitting in an electro-static bag on my desk as I write this, ready to be discarded. The new card is the one which we shall use in all PhysX reviews going forward as we continue to bring readers our honest and complete thoughts on the games we play.

Here at bit-tech we always encourage developers and manufacturers to enter into dialogues with their users and we always try to bring the most honest and reliable articles to our readers. As such, the original UT3 article will remain as it is in order to communicate the process we went through in the review and to let readers know our full opinions. It will however be updated once more with a link to this story, in order to correct the mistakes made when judging the performance of the game.

If you have any questions or comments on the matter, or simply want to let us know how your PhysX card is holding up, then head to the forums and tell the world.