erikchen@ or ginzton@, can you give more details about you system and hangout session, such as your OS version, Chrome version and number of participants in the call?

sprang@, can you help triage as original report points to increase in SW encode path? It might simply be due to higher resolution though. Also, there might be hangouts experiment going on.

ginzton@ reports the same issue with the minimal webRTC app.

"""
4) appr.tc drives my computer crazy too (sufficient cpu usage to turn on the fan)
appr.tc in "waiting for someone to join" (without loopback):
"""

Deleted: unnamed.png 37.2 KB

	unnamed.png 37.2 KB View Download

System is a maxed out 2015 Retina MacBook Pro (15", 2.8 GHz i7, 16GB RAM, AMD R9 M370X 2GB GPU), running OS X 10.11.6, Chrome 55.0.2883.95.

This has been reproducible with every recent Chrome version for months maybe years (and on a similarly speced 2012 Mac), and I've been just putting up with it, but a call yesterday where I had trouble hearing other participants over the fan noise urged me to try to get to the bottom of it.

This is fully reproducible in a 1-person hangout. Even a 0-person hangout (go to hangouts.google.com, click the "video call" button which opens a new tab with the GVC interface, then click "start or join hangout", enter a hangout name, hit enter to start the call, and you get the "you are the only one here" screen, with a little video thumbnail of yourself at the bottom right.

The 0-participant hangout generates substantial CPU load, enough to start my CPU fan. On the order of 65% for the "Tab: Google Hangouts" worker process and 40% for the "Browser" process. If I then join the same hangout from my phone (making it a 1-participant call the load gets somewhat higher (~85% for the Hangouts tab and still ~40% for the Browser process).

Both the cpu usage and the fan noise take a little while to ramp up -- the CPU usage is immediately higher (step change) when I launch the hangouts tab but then continues to go up slowly before stabilizing at the numbers I'm giving here (maybe 30 seconds to ramp up); the fan takes maybe another 30 seconds and comes on quietly at first but gets louder when the high CPU usage is sustained.

That's all to say, if you're just opening and closing tabs and following repro steps quickly to try to repro, the fan noise might not seem so bad, but if you try to use Hangouts as a real user (i.e. carry on a call for more than a minute or two) it gets real.

sprang@: Is this still valid?

I still see the same problem on the same machine (now running OS X 10.13 and chrome 65).