Continued from previous page...

3DMark05
First up we use the venerable archetype of system benchmarking - Futuremark’s 3DMark05.

Run at its default settings of 1024x768 with no anti-aliasing (AA) or anisotropic filtering (AF) enabled, it’s quite heavily CPU dependant, but also does not utilise all four cores, unlike its newer sibling 3DMark06, so it’s a good way to see what improvements the K10 Phenom architecture offers over the K8 architecture of the Athlon before we take the extra cores into consideration. The added L3 cache will most likely come into effect here as well.

Here we can see that the Phenom X4 has a huge performance advantage over the Athlon X2. Still not enough to topple the Q6600 at its higher clock speed but a good start nonetheless. Granted, you probably don’t play a lot of games at 1024x768 these days, but this test is a good sanity check; a baseline if you will. Let’s move on and see if it bears out in some heavier situations.

3DMark06
Love it or hate it, 3DMark06 is the most widely used gamer’s benchmark on the scene right now. It’s true that comparing 3DMark06 scores between ATI and NVIDIA graphics solutions offers little real value, particularly at default settings, however it does provide valuable input for our test today because it is one of a growing number of applications that is able to utilise more than two CPU cores, thanks to the inclusion of some of Ageia’s PhysX software library. This time I’ve performed two different tests; 1280x1024 with 0xAA/0xAF, and 1680x1050 with 4xAA/16xAF.

Oh my gosh... Here you can see a textbook case of CPU bottlenecking with the Athlon X2. To get less than 10,000 points with this kind of graphics card is almost criminal. Overclocking the X2 scales the score up quite nicely, so you can see that 3DMark06 is fully CPU dependant at these settings. Once that bottleneck is removed by having the overclocked Phenom X4 in there, you can really see the score fly.

The Intel chip once again holds the crown thanks to the higher clock speed and bigger cache, although you can see its lead slipping once the resolution and AA/AF quality goes up, which reflects the fact that the benchmark is more GPU dependent at these settings.

Half Life 2: Lost Coast
Here’s another ubiquitous test for gaming setups. Even modest systems have been able to push an average of 60fps or higher out of Valve’s Source engine for quite a while now, but with higher resolutions and AA/AF enabled it stills scales very well with CPU/graphics grunt, and is also useful for finding bottlenecks in your system.

All settings are set to High, with vsync off and HDR fully enabled.

Nice – the universe is unfolding exactly how it should. Without any AA/AF, we see the FPS really shoot up with the overclocked Phenom, even at stock speeds it rivals the overclocked Athlon X2 which will probably be the enhanced cache structure coming into play along with a little help from the extra cores so this is great confirmation that a Phenom upgrade is really worth it for you Athlon users out there. With AA/AF enabled, we don’t see that much gain over the Athlon, but this most likely means that we’re seeing the graphics card starting to limit us.

And for once, the overclocked Phenom almost takes out the faster Intel chip. This begs the question to be asked: would it take out the Q6600 at the same clock speed? Probably, but again, that information offers no value. If you’re an overclocker, there’s no way in hell that you’re stopping at 2.75GHz town with a Q6600 under your hood.

Continued on next page...