Mirrors Edge Game Testing

Testing:

To test out Mirrors Edge I will be using a fly-through demo to allow each card to have a shot at a consistent run through the time demo. Each card will be run 3 times through the run and an average of the three runs will be tallied for the final marks. Each video card was run with Phsyx enabled and disabled to see what impact this has on gameplay and the realism of the effects. Resolutions tested will be from 1280x1204 to 2560x1600. The game settings for each run will not change with the exception of the resolution. The drivers used are the latest releases for each manufacturer. The Nvidia camp used the 181.22 and the ATI camp used the 9.1 catalyst suite. All settings in the respective control panels are left at setup defaults.

Testing Setup:

• Processor: Intel Core I7 920 150x20
• Motherboard: MSI X58 Eclipse SLI
• Memory: Mushkin HP3 12800 7-7-7-20
• Video Card(s): See Below
• Power Supply: Mushkin 800 watt Modular power supply
• Hard Drive: 1 x Seagate 1TB SATA
• Optical Drive: NEC DV5700
• OS: Windows Vista Ultimate Edition SP1 64bit

Comparison Video Cards:

• Sapphire HD4870X2
• Diamond HD4870 1GB
• Nvidia GTX 295
• Nvidia GTX 260-216

Settings used:

• 8X AA
• Physx On /Off
• All other in game settings to maximum

What is clearly evident in the graphs above, is that when Phsyx is disabled in the game, the competitors are relatively close in performance. More so between the GTX 260 and the HD4870 1GB card. But when Physx is enabled in the game, the green camp just pulls ahead by large margins, bordering on double the performance of the ATI cards. You could have tricked the HD4870x2 into running better by renaming the EXE to one that offers Crossfire support, but that really is not the way to test, as most people are not going to get that involved. When Physx is enabled, the ATI cards offload the calculations to the CPU, causing a massive performance drop that is not evident with the Nvidia cards, as the Physx calculations are handled on the Nvidia GPU.