Hardware Test: Can A High-End PC Achieve A Performance Break-Through?
In these tests, the Gaming PC 2 and Gaming PC 3 come into play. Based on an overclocked Intel Core 2 Quad 6600 and an AMD Phenom II X4 940 BE, with clock rates at 3.0 GHz and 3.6 GHz, we tested the two platforms with GeForce GTX 260 and GTX 285 graphics cards. In addition, we also tested two GeForce GTX cards on the AMD system, both in SLI and in dual, independent card configurations, both for graphics output and PhysX acceleration. All graphics settings were either turned on (where no value settings were available) or set to their highest levels. The test runs were further separated to compare Shader Model 3.0 and 4.0.
Intel Versus AMD
From the get-go, the difference between the two CPUs was unmistakable. At the same clock rates, our old and overclocked Intel quad-core outperformed the AMD Phenom II, whether at 3.0 GHz or at 3.6 GHz. The performance boost going from 3.0 to 3.6 GHz had an obvious impact on the minimal frame rates. Although the average frame rate also climbed, it didn’t improve as much as the minimum values. Because both systems delivered more or less identical results, we also decided to perform some more tests on the AMD system (Gaming PC 3) to observe the impact of SLI. Here, we have to give AMD a significant nod for its Phenom II X4, because its older dual cores painted a much bleaker picture.
Quad-Core Versus Dual-Core
We compared our first set of benchmarks, taken with two cores deactivated, against another set taken with all four cores turned on. This led to the somewhat sheepish realization that this game does not benefit much from the addition of additional processing cores. What doesn’t work with two cores, still doesn’t work with four cores, either. This is too bad, because the PhysX calculation on the CPU could have benefited from this scenario.
Benchmarks: High Settings
Next, we focused on high graphics settings with PhysX and Shader Model 3.0.
…and then, with Shader Model 4.0.
Results:
- For Shader Model 3.0, frame rates are slightly higher than they are for 4.0
- Frame rates scale well above 3.0 GHz as clock rates climb, for both processors
- SLI confers somewhat higher frame rates, but doesn’t scale well for this game
- The combination of a graphics card for rendering plus a second card for the PhysX accelerator doesn’t do much
- As before, the CPU remains the measure of all things for this game.
Playability
On any of these configurations at high settings and with all four tested graphics cards, game play is fluid throughout. But not everybody has a PC that belongs to this performance class. This puts the game into a two-class model (the haves and the have-nots). We can only hope a future patch might remedy the situation.
Eye Candy
The visible differences between these settings and their mid-range counterparts are not terribly striking, as you have to look carefully to see any improvements. That said, shadows and steam look much more natural at first glance. Even here, hardware investments and results don’t seem to show a meaningful relationship. By comparison to FEAR 2: Project Origin, the benefits are very noticeable.
