Synthetic benchmarks are better than actual programs at singling-out any specific performance problems, even though their results can’t always be related to actual programs.
3DMark shows the extra calculating power of three GTX 260 Core 216 graphics cards, making it appear even more unfortunate that only one of our games benefited.
PCMark does show a slight advantage in system performance when using two, rather than three graphics cards. We didn’t see this difference in real applications however.
Unlike PCMark, Sandra shows zero difference in CPU or memory performance between our two configurations.