|
||||||||||||||||||
|
||||||||||||||||||
|
|
||||||||||||||||||
|
3DMark 3DMark has been the standard amongst gamers for synthetic benchmarking for nearly a decade. For the purpose of this review, we will be looking at 3DMark06 and 3DMark Vantage.
Take a minute to gander at the above graph, and let's see what this data means. The CPU score scales as we would expect, with the Q6600 at 3.54GHz, the CPU score runs around 5200, and at 3.98GHz, it pushes to around 5700, which is about an 11% increase in score for a 12% increase in clock speed. Going from a single HD 4870 X2 to Quadfire gives us a 4.1% increase in total 3DMarks, a 1.3% decrease in SM2.0, and a 9.7% increase in SM3.0. However, going from Quadfire at 3.54GHz to Quadfire at 3.98GHz increases our total score by 11.6%. Overclocking the Quadfire gave us no added benefit. This obviously means that 3DMark06 is highly CPU bottlenecked at this point, which is no surprise considering that 3DMark06 runs at 1280x1024 with no AA.
Next up, we take a look at 3DMark Vantage X. If you are unfamiliar with Vantage, the "X" preset denotes "Xtreme" and runs the testing at 1920x1200 with 4xAA and 16xAF. Here the Quadfire rig actually scales quite well, with a 69% increase in overall score, and a 73% increase in GPU score over the single 4870 X2. Overclocking the CPU to 3.98GHz gives us an additional 4.5%. Overclocking the GPUs does not really add anything here.
The "H" preset for Vantage runs the test at 1680x1050, with 2xAA and 8xAF. As the settings are reduced, we would expect to see a CPU bottleneck have a larger effect on scores. In this case, going from a single to Quadfire gives us 50% more performance, overclocking to 4GHz gives us 3.8% more on top of that. Again, overclocking the GPUs gives us very little benefit.
Rounding out the 3DMark testing, we have what is probably the most popular preset for Vantage, the "P" or "Performance" preset. This runs very similar to 3DMark06, with resolution set to 1280x1024, 0xAA and Trilinear AF. As expected, scaling is adversely affected at his low resolution, gaining only 24% in this test. Overclocking the CPU adds another 10%, which would confirm the CPU bottleneck here. Again, overclocking the Quadfire yields no benefit.
|
