The benchmark actually has a pretty reasonable workload: rendering, playing sound effects, doing some basic flood-fill searching, etc. And it's much faster in IE10 than Firefox for me, so...
With graphics-intensive things, I tend to find that IE9 and IE10 always come out on top, followed by Chrome and Safari. Firefox is usually horribly slow.
Yes, they added a lot of bells and whistles to make it relevant for a benchmark. I am just surprised that they didn't use something that is more dynamic by design.
I can't tell for FF, but Chrome has no problem with that demo.
