ATi's X800 Pulls Off Another Coup in the Graphics Performance War

Test Setup

Not only did ATi’s reference cards reach us very late, they also came with the wrong BIOS versions. At 473/445 MHz, the X800 Pro ran slightly slower than specified. The X800 XT, on the other hand, was markedly overclocked, running at 526/574 MHz instead of 520/560 MHz. This was enough to boost the card’s performance by up to 3% - enough to falsify our measurements. Unfortunately, we only realized this after we had already completed our suite of tests, making it necessary to completely re-run all tests. For this corrected run-through, we used an ATi-internal clocking tool (please don’t ask for it !) to adjust the cards’ clockspeeds to the correct values.

Not to be outdone, NVIDIA also added to the hectic pace of this author’s testing procedure by sending a new ForceWare beta driver v61.11 at the last minute (late Friday afternoon). This driver is supposed to improve performance in several applications. According to NVIDIA, the driver does not influence image quality to this end. Unfortunately, we no longer had sufficient time to run detailed image quality tests to verify these claims at this point. Therefore, we chose to simply show the results of the new driver alongside those of the driver version 60.72. Interestingly, we also noted several cases in which the performance actually worsened when using the new driver, as we will point out in the benchmark section.

Benchmarks

Unreal Tournament 2004

For our tests in UT2004 we used our own timedemo on the map Assault-Torlan (no bots). All quality options are set to maximum.

Unreal Tournament 2004 - Normal

Without quality enhancing features such as FSAA enabled, all cards are more or less CPU limited. The ATi cards seem to take a little more of the load off the CPU, however, as their frame rate is just a touch higher. In 1600x1200, the X800 Pro begins falling behind the X800 XT and the GeForce 6800 Ultra.