![]() |
![]() PERFORMANCE TESTS How We Tested We used the same test bed for the PCI cards that we used in our November reviews: a 200MHz Pentium with MMX technology in an ASUS TX97-XE motherboard and 32MB of SDRAM. For the AGP reviews, we used an ASUS P2L97 motherboard with 64MB of SDRAM, running a Pentium II at 266MHz. (This is the same board that Falcon Northwest uses in its AGP rigs). Both systems ran Windows 95 OSR 2.1 with the USB supplement installed. AGP systems had the updated VGARTD.VXD file as well. The refresh rate for the P55C system was set at 75Hz; for the AGP system, 85Hz. Both settings are typical for similar systems. Refresh rate is an important factor when running 3D, so we had to be consistent for test purposes. You might play with your refresh rate to get a few extra fps. (Note that a higher refresh rate doesn't necessarily equate to a higher frame rate.) We also used the same test software as before, with the addition of Wipeout XL. On the AGP systems, we also ran the 3D WinBench large texture scene, but the test ran inconsistently (even from one test to another), so we don't report formal results here. Besides, the FireGL Pro AGP board has 8MB of fixed local memory, so the large texture scene, which uses only 5MB of textures, would not have tested its AGP behavior. The testing software we used included the following: Game Tests Flight Simulator 98 Wipeout XL Moto Racer (the Delphine shipping version, not the EA version) Quake at 640x480 (GL Quake where applicable, WinQuake for DirectDraw testing, and DOS Quake) Duke Nukem 3D Synthetic Benchmarks 3D WinBench 97 (final score) WinBench 97 Business WinMark (2D Windows GDI performance) TerraMark by Eidos Interactive (version 2) CBench SVGA test (DOS performance) Click here for full results |