Disagrees with Intel Power Gadget

[smeylan]

osx-cpu-temp reports ~46C while Intel Power Gadget reports 54C. The discrepancy increases when the machine is under heavy load (osx-cpu-temp doesn't increase appreciably, while IPG goes to 100C). I'm using a 15" mid-2012 rMBP (MacBookPro10,1).

[roastchicken]

If you're still able to reproduce this, it might be worth testing the stock powermetrics command-line utility which (AFAIK) is the closest you can get to getting these values from the SMC directly. Perhaps Intel's Power Gadget is doing something special to get more accurate results?

sudo powermetrics --samplers smc -i1 -n1 will get a single (-n1) sample from the SMC (--samplers smc) after waiting 1ms (-i1) (normally it samples continuously every 5000ms)

If you can reproduce it with powermetrics, then we know the cause is not osx-cpu-temp but rather what the SMC is reporting.

[ahmetb]

I can +1 and say disagrees with Fanny Widget for macOS as well. Seems to be reporting around 6°C lower.

[WinkelCode]

If you're still able to reproduce this, it might be worth testing the stock powermetrics command-line utility which (AFAIK) is the closest you can get to getting these values from the SMC directly. Perhaps Intel's Power Gadget is doing something special to get more accurate results?

sudo powermetrics --samplers smc -i1 -n1 will get a single (-n1) sample from the SMC (--samplers smc) after waiting 1ms (-i1) (normally it samples continuously every 5000ms)

If you can reproduce it with powermetrics, then we know the cause is not osx-cpu-temp but rather what the SMC is reporting.

I tested it against powermetrics, it was reporting around 99°c, meanwhile osx-cpu-temp was reporting around 50-60°c. The 99°c number is believable, the MacBook was working hard and 2015 MBPs are notoriously hot, even with fresh thermal paste.