As this article is only meant to measure the GNU/Linux Intel driver performance (which is a Kernel-matter), the desktop environment shouldn't be the bottleneck here.
Ubuntu's Unity is a _huge_ one and we should avoid it.
I should've been clearer and went too specific: Michael should compare Ubuntu's overall performance with another distribution.
Beforehand, we'll have to wait for judgement day.
Comparing various DE's running _on top of_ Ubuntu doesn't make sense. In my opinion, Ubuntu itself is the problem.
It may be very popular, but it's a mess.
16-Way Linux OS Performance Comparison, the benchmark results are mostly similar across distributions although there are some obvious problems with some (what is going on with PostMark on BSD?? And Nexuiz on Debian?).
While in OpenArena and Unvanquished, Windows had a slightly higher FPS than Linux, its max latency was higher than Linux's, thus you would actually have a better experience using Linux. I would consider this scenario to be win for Linux even though the numbers show a "loss".