Doesn't seem to happen going by the graphs included, but one thing I look in benchmarks is not how fast or not a program was, but how frequent spikes and hiccups in speed are. Having played games at 11 FPS but that were consistent at that and seemingly weren't lagging (variable max fps?), big numbers don't tell much imo.

Also, statistically, one single benchmark, and from an unamed game at that, doesn't tell much either. If I might suggest, maybe do like the microblogging folks and start a responses/quoting thread of more tests?