Benchmarks: Just Cause 2, Mafia II, Metro 2033
We used Just Cause 2's built-in benchmark tool, choosing the most demanding of three tests called "Concrete Jungle". Fraps was used to record the minimum frame rates and monitor the benchmark tool's coherence. The best in-game visual settings were selected with 8xAA/16xAF.
When we tested Just Cause 2 using a single monitor we had the GeForce GTX 590 taking the lead at 1920x1200, but resulting ~6% slower at 2560x1600. This margin was increased to 23% at 5040x1050 as the Radeon HD 6990 rendered 11 more frames. The gap widened at 5760x1200 as the dual-GPU Radeon was 60% faster and finally 100% faster at 7680x1600.
For measuring performance we used the game's built-in benchmark which can be found under the "Extras" menu. All settings were set to their maximum value while anti-aliasing was turned on. The only setting disabled what PhysX.
The GeForce GTX 590 showed a massive performance advantage over the Radeon HD 6990 in this game. At 5040x1050 the GeForce GTX 590 was 65% faster, 107% faster at 5760x1200 and then 79% faster at 7680x1600.
For testing Metro 2033, we used Fraps to gauge GPU performance based on 60 seconds from the "Chase" level. The game was tested with DirectX 11, but we disabled Tessellation as it hammered performance too severely. Anti-aliasing was also disabled and instead we used the default analytical anti-aliasing with 16xAF.
With a single 30 LCD at 2560x1600 the Radeon HD 6990 was 20% faster than the GeForce GTX 590. Now at 5040x1050 we found that the Radeon HD 6990 is just 5% faster and 3% faster at 5760x1200.