The Rage Benchmark Conundrum: Video Comparison
Whenever we benchmark hardware, we strive to isolate variables, using different components at the same in-game quality settings. This lets us generate the most meaningful results. But does it make sense to benchmark a game that adjusts quality to achieve 60 FPS, especially if it's dynamically changing settings in the process?
It can, we believe, if the approach is correct. First, we plan to benchmark the game with several different graphics cards to gauge just how much hardware muscle the Tech 5 engine needs in order to hit its 60 FPS target. We doubt it's able to yield the same frame rates driven by today’s entry-level discrete graphics hardware. So, at the very least, we can show you the minimum you need to get to hit 60 FPS.
Secondly, while we don’t have control over what the Tech 5 engine does to dynamically alter the graphics load, we can present the results in video form and let you judge for yourself whether they're comparable. We do this by recording in-game footage from four different graphics cards, all of which sustain an average 60 FPS, but drop to different minimum frame rates. The delta between the minimums suggests that the slower cards struggle compared to the higher-end cards. So, we'd expect the game engine to drop its load in order to achieve similar average performance.
The problem with this approach follows the Heisenberg uncertainty principle. The act of observing causes a change in the conditions of the experiment, and in this case recording video impacts the CPU load during game play. Subject to variance or not, this experiment is worth running because it could demonstrate significant differences between rendering output on various graphics cards.
The results reveal few visual distinctions between the components on our test bench. The engine's load balancing is largely undetectable, meaning you can expect a very similar in-game experience regardless of the graphics card you choose, so long as your PC is fast enough to sustain smooth frame rates. This mirrors my testing experience, as I wasn't able to notice the difference between various configurations.
Additionally, the output from GeForce and Radeon cards looks very similar, so it’s nice to know that you’ll see a consistent result, regardless of graphics hardware.