Able to reproduce the issue on Windows 10, mac 10.12.6 and Ubuntu 14.04 using chrome reported version #63.0.3239.132, latest stbale #64.0.3282.119 and latest canary #66.0.3334.0.

Bisect Information:
=====================
Good build: 63.0.3215.0
Bad Build : 63.0.3218.0
Note: Got all bad builds for first time, hence increased the good build in order to get bisect results. Also could not run per-revision script due to perf errors. Hence, provided bisect using old bisect script.

Change Log URL: 
https://chromium.googlesource.com/chromium/src/+log/783d51fa63399dc0c7022ff06a73b2448e5632e8..96f4437916860e307ec9b8d02b098cb784430c79

From the above change log suspecting below change
Change-Id: I4866f2748dbdab800421561d8fc849d88c73280f
Reviewed-on: https://chromium-review.googlesource.com/668098

wez@ - Could you please check whether this is caused with respect to your change, if not please help us in assigning it to the right owner.

Thanks...!!

My change (https://chromium-review.googlesource.com/c/chromium/src/+/668098) just changes the way we report CPU usage under Windows to be consistent with other platforms; it doesn't affect actual CPU usage.

Loading this site in Chrome Canary, I'm not able to repro "a big performance drop after a few seconds" but I notice that:
- It consumes ~1GB of GPU-process memory, despite having very basic 2D graphics.
- When the game tab is minimized, it consumes a full CPU core.

Basically this looks like a buggy site.  I'd be interested to know whether it really performs any better under Chrome M62 on Mac, as per the original report.

Tested the issue on Win-10 using M62 chrome version #62.0.3170.0 and chrome reported version #63.0.3239.132. Observed that M62 build consumed very less CPU usage i.e below 10. Whereas using chrome reported version #63.0.3239.132, very high CPU usage was observed i.e more than 50.

PFA two screen casts for both behaviour #62.0.3170.0 and #63.0.3239.132.

Thanks...!!

Deleted: 805840@62.0.3170.0.mp4 3.6 MB

	805840@62.0.3170.0.mp4 3.6 MB View Download

Deleted: 805840@M63.mp4 1.6 MB

	805840@M63.mp4 1.6 MB View Download

wez@ - Please check the screencasts and please provide inputs on the same.

Thanks...!!

Re #5 & #6: As per my comment #3, this is working as intended - my CL https://chromium-review.googlesource.com/598750 fixed the way in which CPU is reported under Windows, to report %age-per-core, rather than %age-of-total-CPU, making it consistent with other platforms (e.g. Mac :).  Based on your videos, I would guess that you are testing on a 12-core Windows system?

The original report was from Mac OS X, whereas my change only affects how CPU is reported under Windows. I would recommend attempting to repro the issue under Mac rather than Windows.

Removing hasbisect tag, since the bisect result was incorrect due to the Task Manager CPU reporting change from M62->M63.

As per comment #3 and #7, tested the issue on mac 10.12.6 using chrome reported version #63.0.3239.132, latest canary #66.0.3340.0 and chrome version #60.0.3090.0.
Observed that on navigating to site: https://yggdrasilgaming.com/games/vikings-go-berzerk/ and playing the game increased the CPU usage to more than 20%. It remained stable after playing sometime but on minimising the tab the CPU plunged to 101%.

Attached a screen cast for reference.
Hence, marking it as untriaged for further inputs from dev team.

Deleted: 805840@M60.mp4 7.3 MB

	805840@M60.mp4 7.3 MB View Download

I've reproduced the issue when the tab is in the background it consumes ~101% of the CPU time. Seems they are using some fake timer via Window.postMessage to consume lots of time.

Deleted: trace_background.json.gz 2.7 MB

trace_background.json.gz 2.7 MB Download

Re #12: Thanks, throttling postMessage tasks is on our radar, this is a very good example why we should do this.

Closing this bug as a part of Blink>Scheduling bug review. Future work on throttling postMessage tasks will be tracked separately.