Ping! Any specific insights for this benchmark?

Shoot, sorry - I missed that I was going to get an animometer benchmark to investigate (although I have been looking at some others).  I will get on this.

Chrome performance on this has more than doubled in recent builds on Mac, perhaps due to Ganesh now being enabled.

Still, a significant fraction of the time (~25%) is spent parsing CSS transforms of the form: "translate(877px, 205px) rotateZ(0deg)".  Issue 605792 tracks the larger issue here, but it looks like https://codereview.chromium.org/1904633002/ may be a hack to get us a bunch of benefits.

It's kinda hard to measure the impact of improvements on this benchmark as the results seem to vary wildly.  On my macbook even in Safari it reports as +-20% but typically varies overall by 50% (between 400 and 600).

Running for 120s and "maintain target frame rate at 50fps" I can get much more stable results.  Using Elliott's patch (https://codereview.chromium.org/1904633002/) gives me about a 20% increase in score.

In particular, it cuts the time spent in JS (which is almost entirely just setting the transform on all the nodes) in half (eg. from 6ms/frame to 3ms/frame).

I don't think there's anything left here other than continuing to improve the CSS fast paths (issue 605792).  Let's continue tracking the work there.