This benchmark does have a connection to virtual reality, as part of my inspiration to make it came from my experience using vr.js and WebGL for virtual reality in browsers. The Web isn't yet ready for VR due to latency issues, but I hope to change that!
On that note, are you guys interested in browser support for the Rift? I believe the latency issues are surmountable with effort and some cooperation from browser vendors, and I'd love to see something official in that direction (in fact, I'd love to work on something official).
 This is what I made: http://www.youtube.com/watch?v=pWYlTIt4j0Y&hd=1&t=1m11s
Firefox (except Mac)
Chrome and Mac Firefox
Windows 8.1, SSD, i3, 16GB RAM (63% used)
Chrome 30.0.1599.101 m: 3.4/10
Chrome Incognito: 4/10
Firefox 25: 5.4/10
FF Private Browsing 3/10 O.o
All tests done at least twice with similar numbers.
But Chrome has been running for a few days while FF was just started. Guess I'll leave FF running a bit and see if it changes anything.
Also interesting, FF takes around 2-3 times as long to run the tests than Chrome.
The reason for the low Chrome score is this:
CSS: 1.0 frames jank, Scrolling: 170.1 frames jank
> FF takes around 2-3 times as long to run the tests than Chrome.
There are a couple of reasons for this. Firstly, lower latency is more difficult to measure, so it takes more trials to get the same number of good measurements. Secondly, Firefox has longer scroll momentum than Chrome, and the benchmark waits for the scroll momentum to die down between each scroll latency measurement.
Thanks for the explanation, makes a lot of sense.
Can you explain the extreme FF private browsing results?
Measuring the input latency of individual websites would be interesting too. I'd love to incorporate that into the benchmark if I could figure out a way to do it.
Did anyone else have difficulty getting it to work on a multi-monitor setup? It gave me an error about trying to find the reference image and suggested I switch monitors, but that didn't work. I finally had to disconnect my secondary monitor altogether before it would work.
I was initially in Win7's "enhanced graphics" mode (can't remember actual feature name) but when I tried to run the benchmark I was thrown into Basic mode. I had a fair amount of other applications running, though, so I may have just run out of memory.
3.0/10 on MPBR early 2013 w 16GB and SSD.
Authors should consider a way to collect and present the data.
3.4 when the window is sized to it's smallest size.
2.5 when it's at maximum resolution (1600x1200).
Running on OSX 10.9.
Spoiler: Safari actually gets a very good score (relative to other browsers).
Other scores - Chrome 30: 2.4/10. Firefox 24 and 25: 1.2/10.
> Failed to find test pattern on screen. Ensure that your browser's zoom level is set to "100%", and the top-left corner of the window is visible. If you have multiple displays, try moving the browser window to the main display.
* Zoom 100%: check.
* Main monitor (of 2): check.
* Top-left visible: check.
I give up.
Edit: It's more likely that it's just buggy support for multiple monitors. I will look into fixing it; for now the only workaround would be disabling your second monitor.
 - http://jankfree.org/
Grew up in the midwest fwiw
It's been used in this context since at least 2010 by the Chrome team: http://en.wiktionary.org/wiki/Talk:janky
Given that all the other numbers on the page are ones for which smaller is better, I think the main score itself should be accompanied by a note saying explicitly that higher scores are better.