|
Testing:

To run the 4870X2 through its paces, we are going to run a plethora
of benchmarks. Now, GPU benchmarks generally fall into three
different classes, all with different strengths and weaknesses:
Synthetic: These benchmarks are generally not tied
to any specific game, but are generally designed to emulate what
gaming can be like. If the benchmark is universally accepted,
it can be used as a rough guideline as to performance. The
downside is they are a poor judge for any specific game title.
Timedemos: Time demos are usually pre-recorded
sequences within the game that are replayed, and the performance
measured. The advantage of time demos are that runs are exact
every time, and should give a good measure of a games performance.
The disadvantage is that not everyone owns all the games.
In Game Frame Counting: In game frame
counting usually involves the tester playing the game, and using a
tool like FRAPs to capture the frame rate. The advantage here
is that actual game play is being used, the disadvantage is that it
is hard to duplicate the exact same sequence more than once.
In order to paint the most complete performance picture, we will be
using all three methods. And in order to make things
interesting, we are going to compare the X2 to the reigning champ: a
factory overclocked MSI GTX-280, all on the following test bench,
using Vista SP1:
| CPU: |
Q6600 @ 3.54GHz, 1.35V |
| Motherboard: |
Foxconn Blackops Quantum Force X48 |
| GPU: |
MSI GTX-280 650MHz/2300MHz, 177.39 Beta |
| RAM: |
2x2GB OCZ Reapers, DDR3-1770 8-8-8-27-1T |
| Case: |
Danger Den Torture Rack |
| Sound: |
Onboard |
| Cooling: |
Swiftech H20-220 Compact |
| Hard Drives: |
1x500GB
Seagate Barracuda
2x150GB Raptor X |
| PSU: |
Corsair 1000HX |
Synthetic Tests:
First up on our list of tests will be the synthetic benchmarks.
In this category we will be using
Futuremark's 3DMark06 and 3DMark
Vantage. First up, the 3DMark results:

In my mind, 3DMark06 is too antiquated to really be of use anymore,
especially considering that it is run at 1280x1024. As time
goes on, it is just not going to scale like it used to. But, I
still run it because I know people will hoot and holler if I don't.
The 4870X2 takes down the 280 by a margin of 2,297 points, or
a gain of 13%.
Next, we will take a look at 3DMark Vantage results. First,
however I must add a caveat. The 177.39 beta drivers add PhysX
support to the GTX-280, which inflates the CPU score by a good
margin. As such, I won't report the CPU test findings here,
however, they do affect the total score, so keep this fact in mind.

First up from the Vantage testing is the "X" test. The X
preset runs graphics at the highest quality, and places the largest
emphasis on GPU performance. Thus, the X preset best
represents the high end market. The 4870X2 bests the 280 here
as well, by 39% in the total score and 42% in the GPU test.
Pretty impressive.


The Vantage "H" and "P" presets represent more mid-range type of
usage models. Even still, the 4870X2 prevails in each test by
at least 20%.
|