Issue 685844  has been merged into this issue.

I don't see the problem on the two laptops and one desktop machine I tried. I'm wondering if it might be experiment-related - can you copy/paste your variations?

From about:version:

Variations	6a89113b-a7aa8ed
68ebfce2-f23d1dea
90757ebb-f23d1dea
1e528f0f-f23d1dea
918346b9-3f4a17df
38eb801c-3f4a17df
8364a5c2-eb6219f1
7c1bc906-8122a015
ba3f87da-a6404135
f3499283-ca7d8d80
31362330-3f4a17df
c70841c8-a2567007
9e201a2b-3f4a17df
5274eb09-3f4a17df
684d1cdf-f23d1dea
165e16d1-3f4a17df
9e5c75f1-f09cf7b1
6b121ae7-ca7d8d80
f79cb77b-3d47f4f4
b7786474-d93a0620
23a898eb-e0e2610f
74df3f1-3f4a17df
4ea303a6-8467a4eb
3d7e3f6a-2eb01455
f8c15c50-3f4a17df
9736de91-ca7d8d80
64005e71-fb487281
3326cd71-3f4a17df
f069f10b-f23d1dea
ad6d27cc-7075cd8
ca314179-ca7d8d80
69bf80fa-f23d1dea
b2f0086-93053e47
7fc902e8-f23d1dea
adda5502-3d47f4f4
3ac60855-486e2a9c
f296190c-6e597ede
4442aae2-7158671e
ed1d377-e1cc0f14
75f0f0a0-d7f6b13c
e2b18481-9d9eea77
e7e71889-4ad60575
828a5926-ca7d8d80

Thank you. Unfortunately, nothing jumps out as a cause for the flickering.

ccameron@, what are your thoughts about this bug?

There haven't been GPU changes in a while.

Do we have a bisect of when this appeared?

I am unable to reproduce it - thakis@, would you please bisect?

I tried and failed to find a bad revision to start at.

I tried:
tools/bisect-builds.py --verify-range -a mac -b 433059 -g 400000 -- --no-first-run --user-data-dir=/tmp https://webkit.org/blog-files/3d-transforms/poster-circle.html
# ^ current beta base rev as bad

tools/bisect-builds.py --use-local-cache --verify-range -a mac -b 444943 -g 400000 -- --no-first-run --user-data-dir=/tmp https://webkit.org/blog-files/3d-transforms/poster-circle.html
# ^ current dev base rev as bad

In both cases, both ends of the revision range are fine.

(Also, bisecting got really difficult compared to when I last tried to do it. I used to pick a rev by going to build.chromium.org and then click "continuous" and "mac", but that is now a new page design that loads extremely slowly and sorts revisions lexicographically; also it stopped updating a long ago as far as I can tell. https://www.chromium.org/developers/bisect-builds-py only says "go to omahaproxy to get end points" which isn't very granular. Does anyone know how bisecting works nowadays? Also, what's up with having to pass --use-local-cache for the tool to not take minutes to start up?)

Hm, I restarted chrome and now poster circle works fine. (Still at Version 56.0.2924.76 beta (64-bit))

Should this just be closed then if it doesn't repro anymore, and we can then reopen if other people see it or it starts happening again?

Re: bisect, I just bisect between 0 and ToT and it's usually good enough.  Alternatively I do something like git log 51.0.2662.0 to get some revision that's closer to now than 0.

I am seeing occasional flicker on 56.0.2924.87 Linux Stable.

I bisected a flickering issue on Ubuntu to: https://chromium.googlesource.com/chromium/src/+log/67480252d6f5fa0b58c67ed241f7025f302807c6..3e9082bfc63be6b304543ee499354fd71ea063ae

This may or may not be related to the issue reported by thakis@.  I'm seeing a very subtle flicker, which looks like 1/2 of a quad is missing for a frame.

It's rare and requires looking at the poster circle for ~10 seconds.  Sometimes it doesn't happen and requires closing and re-opening the tab.

Undoing above changes.  I just noted thakis@ mentioned the flicker he noticed was on all pages. These issues must be unrelated, so I will file a separate issue for Poster Circle.

Poster Circle specific flicering issue moved into  Issue 693826 .

I think we should close this out (and investigate vmiura@'s reported issue in the other bug), unless this has shown up again.

thakis@, have you seen this since the initial report?

cc'ing thakis@ for question in comment #17.

I haven't seen this recently.

Per #19, closing this for now.  Reopen if it happens again.