I'd say it's more like benchmarks don't tell the whole story. They should still be good for catching obvious regressions before any code hits a real user's machine, but having real world data should help to identify a whole bunch of bottlenecks that were unknown before, and determine if performance tweaks are actually helping.
It's a shame browsers focus so much on JavaScript performance. I tried FF4 a few weeks ago and the user interface itself just feels slow whereas Chrome's interface is much snappier. It'd be great if all of the browser vendors realised that JavaScript doesn't tell the whole story!
Fairly obvious that it's slower than IE? Says who?
You treat "repaint" as if it's a single thing that's either faster in all cases or slower in all cases. Browser rendering is a very complex thing and some pages render faster in one browser while other pages render faster in another.
Sorry man. I thought it was clear that it was says me. I'm just one data point obviously, but the speed differences on the sites I go to/make was clear. I removed the earlier comment, it probably sounded more authoritative than it should.
And I was referring to whatever Chrome web developer timeline calls 'paint'. FF could be faster in other parts of rendering, I don't know. But paint matters to me the most right now and FF disappoints there.
