First up among the benchmark tests is Futuremark's 3DMark06. This benchmark is roughly two years old or so, and runs at a default 1280x1024, so the processor can have a marked effect on the total score. Since the same CPU was used in all tests, and those scores were all within a hundred points of each other, I've excluded those from the graph. What we do see from 3D06 is that the 8500GT scores twice as high as the 8200 onboard, while the two combined in Hybrid SLI net around a 20% gain in performance.


Next is Futuremark's latest benchmark program, Vantage. While Vantage does test the processor like 3D06, and also runs at a default 1280x1024, it stresses GPU performance more heavily than CPU. Unfortunately I could not get Vantage to complete a run of all four tests successfully using the onboard. The frames per second on the graphics tests were so bad it would lag and eventually stop responding. Comparing the 8500GT performance to the Hybrid SLI mode shows that there is little advantage in having the onboard share some of the load. It seems the 8200 chipset is simply outmatched by this benchmark.


The last program used in these tests is oZone3D.net's FurMark OpenGL benchmark. This benchmark was run at three different resolutions, 1280x1024, 1680x1050 and 1920x1200. While the 8500GT shows a solid 35% improvement over the 8200 across the board, there was absolutely no advantage to be seen when using GeForce Boost. As often seen in standard SLI/Crossfire configurations, there are some benchmarks/games that simply do not take advantage of multiple GPUs. This appears to be the case here.

Now let's compare these cards in some real world games.