Lost Planet: Extreme Condition

Next, we welcome a new addition to our benchmarking suite - Although it isn't yet available as a full game, its DirectX 10 sporting credentials make it an irresistible subject for testing.  Lost Planet: Extreme Condition has attracted controversy since the release of a demo around the time of ATI's Radeon HD 2900 XT launch, thanks to it being a part of NVIDIA's 'The Way It's Meant To Be Played' programme as well as ATI's lack of access to the title.  However, despite all that this is a playable demo, it does support DirectX 10, and it is a game that PC gamers want to play, making it well worth of examination in my book.

To test this title, we use the demo's built-in benchmarking functionality, which consists of two different timedemos - One utilising a largely outdoor-based snowy scene, and the other an indoor cave with plenty of weird creatures buzzing around.  Both demos show actual scenes that you'll encounter in the game, so these numbers should give a reasonable accurate idea of real-world performance.

At 1600x1200, the higher clock speed of the EVGA board translates into a smallish two frames per second advantage in both areas of the game.

Adding 4x anti-aliasing and 16x anisotropic filtering sees a big leap in the advantage shown by the Superclocked part - 10% in the snow scene, and over 11.5% in the caves.  Having said that, rendering performance outdoors had already slipped a little below what one would prefer in this sort of game genre.

At 2048x1536, both scenarios give around a 10% advantage to the EVGA board, although once more it gains more from its higher clock speeds in the 'cave' demo.

We definitely slip out of playable territory at 2048x1536 with AA and AF enabled, but both demos give EVGA's offering around an 11.5% advantage here.

S.T.A.L.K.E.R.: Shadow of Chernobyl

The last title in our suite is also amongst the newest - The recently released S.T.A.L.K.E.R.: Shadow of Chernobyl, powered by the DirectX 9 level 'X-Ray' engine, which also makes use of High Dynamic Range rendering.  As this game lacks timedemo functionality at present, we've recorded performance via run-through of part of the games open air section using FRAPS.

Unfortunately, this title doesn't support anti-aliasing with dynamic lighting enabled on any graphics hardware at present - Thus, for our testing with anti-aliasing enabled, we've had to drop to the lesser static lighting model.

S.T.A.L.K.E.R. is a tough little cookie rendering-wise with dynamic lighting enabled - You wouldn't believe it looking at these numbers though, as both boards make relatively light work of it.  EVGA's offering shows off over a 12% advantage at this resolution however.  Rather oddly, the game refused to run on NVIDIA hardware utilising dynamic lighting at 2048x1536, locking up as soon as our save game loaded - Thus, we've had to skip our test at this particular resolution.

Even with 4x anti-aliasing and 8x anisotropic filtering in use, moving to static lighting in its wake sees a huge increase in performance, bring both boards into (or close to) three figure frame rate territory.

Not much changes even at 2048x1536 with anti-aliasing and anisotropic filtering enabled - Once again, the Superclocked EVGA part finds itself in the lead to the tune of around 12% however.