Issue 543628  has been merged into this issue.

 Issue 522175  has been merged into this issue.

It's also conceivable it improves performance by optimizing paints. I agree we should look into it. Is this something we can use chrome://tracing/ for?

Yeah, there may be performance/memory tradeoffs.

Tracing has memory data in it too. But it's not amazing for A/B performance testing. Will require some fair bit of analysis.

Normally you'd write telemetry tests to look at performance changes, and then you can monitor over time also.

I chatted with vollick@ on friday and he gave me some direction on how to get some rough metrics using chrome://tracing. He was hopeful that these should be conclusive enough that we shouldn't need to explore telemetry tests.

Stay tuned for progress.

Possibly related to performance of the ripple animation, I came across  issue 596289  today (though it looks to be Windows-specific). 

I've come across a couple of reports about flashes and flickering in the browser UI (issue 593449 and  issue 596122 ). I am also seeing this myself on Chrome OS M-50 beta if I open a ton of windows and move the cursor quickly over top of elements which have MD hover; the issue does not reproduce with the MD flag turned off.

Let's bump up the priority of this investigation.

vollick@, or danakj@, do you have any idea what might be causing these flickers or a direction on how to investigate the root cause?

vollick@, what are the chances the tracing metrics we discussed last week would have markers/signals for this issue?

No, but tracing is the tool to use to debug stuff.

This did not make M-50, but what I've heard at the weekly meetings makes me think this is still a high-priority issue.

This is definitely still a high-priority and some of the preliminary findings were the motivation for fixing  issue 604415  ASAP.

I have completed my investigation and AFAIK there are no concerns here anymore.  I just need to document one more part of the investigation before opening it up for review.

Moving this nonessential bug to the next milestone.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Ben, it's been a bit over a month since comment 15.  Can this bug be closed?  What remains to be done?

I still haven't been able to capture my discussions about memory costs however all findings show there there is little to no concern. 

Perhaps we should reduce the priority here and maybe even remove the milestone.

I agree this is no longer a cause for concern. Removing the milestone but posting the findings would be good still.
Also changing description to state that this is specifically about ripple cost to not be too general.

Where is a good spot to 'advertise' the findings?

graphics-dev and maybe chromium-dev