For the sake of consistency, Far Cry (v1.1) was tested with all graphical options placed at their highest settings and sound disabled. The Research map was used, with testing consisting of playing completely through the map in God mode using the same basic path each time. Anti-aliasing was enabled via the game’s menu option, and anisotropic filtering via the ForceWare driver’s profile for Far Cry.
Like IL-2: Forgotten Battles, Far Cry’s engine also imposed a much steeper performance penalty for anisotropic filtering than anti-aliasing. The latter’s loss varied from 10% to 20% across the tested resolutions, while higher filtering cost the game 20-30% depending upon the resolution; again enabling the filtering optimizations resulted in a score of 70.3 for 8x AF at 1024×768, once more a roughly 20% performance improvement. Yet combined, AA and AF render Far Cry unplayable at the resolution of 1600×1200 when playing with all in-game settings placed at their highest option.
Call of Duty’s Dawnville demo was used to test this OpenGL title based on id Software’s Quake 3 engine. The included timedemo utility was used to record performance rather than Fraps, and the “com_maxfps” command was used to raise the default frame limit of 85 fps.
With scores as high as listed above, additional reviewer commentary hardly seems necessary. Once again, anisotropic filtering incurs a higher performance penalty than anti-aliasing, yet nothing that impacts the frame rate too terribly.
NASCAR Racing 2003 Season was tested using SimHQ’s crowded Daytona replay. The camera was fixed to Earnhardt’s cockpit for the view mode, and all graphical options were set to their highest.
NASCAR is clearly CPU-limited as its failure to scale with the test settings indicates. Regardless, the game remains playable at even the highest resolution and settings tested.