|
|
Radeon Shootout |
Join the community - in the OCAU Forums!
|
Dead Space, Stalker CS, Conclusions
Dead Space:
This game is getting a little long in the tooth, but it still looks excellent, and it generally scales pretty well. Until DOOM 4 comes out, I reckon it's the astro-horror sci-fi game to play! Testing this game requires the use of FRAPS, which inevitably brings a lack of reproducible results. The following graph gives an indication of the frame-rate variation of the testing undertaken, which was from a save game on the third level.
Whilst one can discern general trends, it isn't a proper apples-to-apples comparison. Nevertheless, it's the only way of testing performance for games without a benchmark utility or scripting support.
Given that the game is mostly indoors, grand open-plain spectacles of the kind rendered in Far Cry 2 are something the game engine doesn't have to worry about, which must surely be a contributing factor to the stellar performance of all three cards.
I told you it scaled well! The performance gaps to the two more expensive cards are remarkably similar at both resolutions.
The 4770 waves its arms frantically for attention, with a fantastic 24% improvement at 1600x1200. The 4890's performance increases are more modest, but who cares when your minimum frame rate is over 100?!
Stalker: Clear Sky:
In 2007 GSC Game World from the Ukraine came up with a winner in Stalker: Shadow of Chernobyl, and its prequel Clear Sky was released in August 2008. The first game was tough on graphics cards and the updated X-Ray 1.5 engine employed in Clear Sky is even more demanding. If you enjoyed these games, look out for the latest in the series, Call of Pripyat, due for release in November.
Whilst these results may seem a little underwhelming, it's worth bearing in mind that the quality settings are purposely pretty high here, to place as much strain on the GPUs as possible. As soon as anti-aliasing is enabled, there's – wait for it – clear sky between the 4890 and the cheapies. Feel free to groan. :)
DirectX 10.1 seems to be ineffective, even producing slightly negative trends here, but the minimum frame rates showed a positive trend across the board, although it was in the order of 5 to 10 percent at a maximum.
Continuing the trend of previous tests, the 4770 sits just below the 4890 in terms of performance, but falls away as resolution increases or anti-aliasing is enabled.
The 4670 and 4890 enjoyed modest 2 to 5 percent gains across the range of tests. The 4770 again takes advantage of the extra memory bandwidth and processing power available from its combined GPU and memory clock speed increases, especially at 1600x1200, where it produces 11 to 13 percent gains.
Conclusion:
The adage of "you get what you pay for" seems to be holding reasonably true, although the value at higher end of the scale can be questionable, depending on the resolution you game at. If you have an older 17” screen and you're at 1280x1024, the 4770 is a nice pick. For around AU$150 you'll get most of the performance of a 4890 for 60% of the cost.
Of course, if eye candy and high resolutions are more your thing, then you'll probably be well aware that you need to spend up on your GPU, and at $250 the 4890 seems a pretty good bet.
The aging 4670 is still a seriously good buy for a casual gamer; the performance of a $90 graphics card nowadays is nothing to sneeze at.
Thanks to AMD Australia for providing samples. Keep your eyes peeled for an SSD vs. Raptor comparison coming up soon on OCAU.
|
|
Advertisement:
All original content copyright James Rolfe.
All rights reserved. No reproduction allowed without written permission.
Interested in advertising on OCAU? Contact us for info.
|
|