Meet The 2012 Graphics Charts: How We're Testing This Year

It's time to revamp the Graphics Charts section! For 2012, we're increasing the number of games and resolutions, dividing the results into three segments. But that's not all. We also include GPGPU benchmarking, power consumption, temperature, and noise.

This year, we massively expanded the number of benchmarks in our graphics charts. Just to give you an idea, a complete run through our benchmark suite now takes three hours and 12 minutes for the fastest board, not counting installing and uninstalling hardware, measuring sound, or taking thermal readings.

As a result, we're consciously splitting the workload up and focusing on high-end and premium cards. Don't worry, though. We'll add up to 20 new boards each month until the lower end of the performance range is filled out, too. As this quarter progresses, we also plan to add the fastest dual-GPU cards and reference models of lower mid-range and entry-level cards, as well as custom-designed variants of already-tested reference cards. It's a long process that takes a lot of planning, but if you don't see what you're looking for today, you can be sure we're in the process of adding it!

Due to numerous requests, and in response to feedback in the forum, this article strives to explain the charts, the underlying test methodology, and the settings we use to generate the results you'll see in the charts section. This should increase transparency and make comparisons easier. Although we can't anticipate and test all possible use cases, we're trying to include as broad a range of benchmarks as practically possible.

You can visit the 2012 Graphics Charts by clicking this link

The 2012 Test Rig

We were starting to see some evidence of a processor bottleneck using our Core i5-based platform in tests able to apply a heavy CPU load, like DiRT 3. Since the Radeon HD 7970 and GeForce GTX 680 push performance to new levels, demanding a more powerful processor, we chose Intel's Core i7-2600K overclocked to 4.5 GHz for this year's test rig. Hyper-Threading is enabled, in the event an application can make use of it.

We set off on this journey at the beginning of the year, meaning Z77-based boards weren't yet ready. There was also no reason to mess with the memory subsystem from last year's reference platform. However, we are using a power supply, Corsair's AX1200, which easily drives up to four graphics cards when we start paying more attention to SLI and CrossFire performance. We're also using Kingston's V100+ 256 GB SSD for all of our applications and games.

The test system is mounted on a rolling desk. This way, we can move the whole rig into a sound-proof test chamber for the acoustic measurements without installing the graphics cards in a second test rig. This is important, since we do all the benchmarks in an air-conditioned room kept at 72°F that is not suitable for sound level measurements below 32 dB(A).

Next, we'll take a look at our choices for monitor and resolutions.