Graphical Benchmarks
We get the graphics subsystem involved here, evaluating the performance drop experienced with the second MagicTwin station using up some of the graphics processing power, along with the load on the CPU and memory subsystems.
72% of the baseline performance shows how both CPU and GPU load on the second station effect overall performance when the graphics card needs to get involved, should it be being used on both stations at once. Of course running a 3D game on each station would cause performance to drop much further. The switch between separate rendering contexts presented by different applications is costly, never mind the likely CPU resource contention on top of that.
It's only 83% with UT2003, showing it's CPU bound nature on Pentium 4 2.4GHz, given a fast enough graphics card.
82% in X2 shows similar results to UT2003.
With our small test sample of applications, the performance difference ranges from 72% to 83% of the baseline figure. It's what you'd expect given the load on the second station.