This seems bogus but I can't really tell, and it hasn't gotten the attention of
any kernel developers yet.
https://bugzilla.redhat.com/show_bug.cgi?id=924570
Like I mention in comment 7, I get:
[ 900.693256] f20c.localdomain mcelog[581]: Processor 1 heated above trip temperature.
Throttling enabled.
[…]
[ 900.704776] f20c.localdomain mcelog[581]: Processor 1 below trip temperature.
Throttling disabled
So in about 1/100ths of a second the temperature is too hot, and then OK, which
doesn't seem possible if the times are correct.
This logging constantly happens anytime the system is even moderately used, it gets hot,
fans are like hair dryers, but no other anomalies. It takes significantly more aggressive
tasks to get OS X this hot. I'd say 1 out of 3 boots, merely boot+startup to
gnome-shell makes the system hot enough to kick the fans to full and report these mcelog
messages.
The claim of hardware error seems like a real problem, but I'm not finding any way to
manually throttle or limit the CPU top end to prevent these errors.
Chris Murphy