LOSTCIRCUITS

Gunmetal

Often considered as the house-benchmark for nVidia graphics cards, we can see some advantage of the GFFX 5900 here. The benchmark was run with the application specific 4xAA at default settings in the control panel. All other settings were set to the highest quality and detail levels (subtitles disabled).

Average Frame Rate

Yellow: 9800XT, Blue: 9800 Pro, Pink: GFFX5900. The GFFX5900 takes a marginal victory here.

One issue we did not go into yet is the question of average frame rate vs. playability. For playability, the average frame rate is some kind of an indicator but what really counts is the lowest framerate. On the other hand, the lowest frame rate depends on where the measuring interval boundaries fall within each run, meaning that what appears a higher min fps value for one card may not really be representative of the actual performance of the card since the slowest sequence may fall into two separate measuring intervals where the numbers are then averaged with faster portions of the same run. On the other hand, the average value is really the total number of frames divided by the runtime, which is very reproducible but may be meaningless. Another classical case of Heisenberg.

Minimum Frame Rate

Yellow: 9800XT, Blue: 9800 Pro, Pink: GFFX5900. Keep in mind that the GFFX and the 9800XT are not "3 x faster" than the 9800 Pro, the latter just happened to "get caught" inside a single mesuring interval at the slowest sequence.

Aquamark3

Yellow: 9800XT, Blue: 9800 Pro, Pink: GFFX5900. The average fps are scaled down by a factor of 10 to match the scale of the graph.

Final Fantasy XI Benchmark2

Yellow: 9800XT, Blue: 9800 Pro, Pink: GFFX5900. Sometimes heralded as a new wonder benchmark, FF-XI does show a few differences but all in all it is still more CPU and system memory-dependent than anything else.

Next Page:    => CodeCreatures, X2 =>