[mbell]

Unfortunately the biggest problem by far is video playback and for whatever reason they seem to refuse to fix the problem.

Here is a simple test I just ran:

Using https://github.com/erkserkserks/h264ify for blocking VP9

2015 rMBP 13", 3.1Ghz i7, 16GB (defaults to software rasterization in chrome)

https://www.youtube.com/watch?v=nK9Im1eP3DI (1080p60 video)

Power is CPU Package Total (core + GPU).

Chrome VP9: 26W

Chrome VP9 + Force GPU Acceleration: 28W

Chrome h264: 19W

Chrome h264 + Force GPU Acceleration: 13W

Safari: 4W

Additional comments:

This is all in window at the default 'theater mode' size of youtube. I can't full screen either of the VP9 options without completely maxing out the CPU, dropping frames and during my laptop into a frying pan. Its really hot even in window with chrome VP9 using ~175% CPU.

[orbitur]

Glad to see some hard numbers on Chrome video performance compared to Safari. I like Chrome because I like the integration with Google services and separate user profiles so I have 2 sessions going at once, but I often have to take Hangout meetings in Safari because Safari doesn't turn my laptop into a hairdryer heating element.

I am curious, though, do the Chrome devs have access to the video APIs that the Safari people would?

[h43k3r]

Any idea how does it achieve that?

[mbell]

There is no reason for h264 decode to take much power at all, it's all done in hardware now (or should be done in hardware).

In other words, Safari's power number isn't surprising but I don't know how chrome manages to use so much processing power. Even my cell phone can decode 1080p60 without catching fire but chrome manages to bring a $2200 laptop to it's knees doing it.