Problems occurred when I would attempt to change resolution (in game) or switch from windowed to fullscreen mode. A number of games had mouse issues.
I'll see if I can install U12.04 on some real hardware and see what happens with a uengine demo.
The whole point of graphing results is to allow easy visual comparison. If the quantities are not actually comparable, then the chart is useless.
It's the equivalent of comparing the speed of two people by putting one on a bicycle and one on a motorcycle. Does it tell you which person is fastest? Yes, the one on the motorcycle obviously. Is that useful? Of course not!
IMHO it would be far more accurate to compare the desktops which don't redirect fullscreen on one graph, and those that do on another. Otherwise articles like this invite simplistic conclusions such as "Unity and KDE are slow", when this may not be necessarily true (though of course, maybe it is!).
BTW.. please take this as constructive criticism.. I'm glad Phoronix exists to produce these kinds of benchmarks!
I have a good idea. How about everyone makes suggestions on improvements and then post your results using the same testing tools that phoronix provides. ;)