After presenting the related work in this field, we move on to our own benchmark experiment. In our experiment we hope to gain a richer understanding or appreciation of the benchmarking process through the development of our own practical implementation. We identify a number of fields within performance measurement that we feel need further exploration. These fields include the measuring of cost factors incurred when introducing graphical special effects. We record the performance of local graphics accelerators when these special effects are introduced. This includes the related strengths and weaknesses of each accelerator. The tests introduce the following special effects zbuffering, blending, fog and light. We apply these effects to the rendering of three basic primitives, namely the point, the line and the polygon. We measure the increases in performance time with the introduction of these special effects in order to determine the performance of various graphics accelerators. We rank both the special effects and the graphics accelerators. We also measure the difference between 2D and 3D rendering.
We conclude this paper with the following results: the tests rendered in 3D exhibited higher performance times than the 2D counterparts, The special effects in order of expense are Z-buffer depth testing; Blending; Fog and Light. The accelerators finished in the following order of performance: the Geforce2; Geforce Pro; Voodoo3 and Monster3D.