Live data from Hacker News

Building a macOS app to know when my Mac is thermal throttling

stanislas.blog

121–123 of 123 posts

Re: Building a macOS app to know when my Mac is thermal throttling

#121
post #79

Hypothesis: it’s not the CPU, it’s the USB controller saturating the thermal mass. So instead of the CPU or GPU going bonkers, the body gets too hot and they lose thermal dissipation capacity for the normal usage and then overheat and then throttle. They go 100% on the graphs because the throttling reduces frequencies and then load becomes too heavy. Experiment other adapters and monitors.

I'm not sure if I entirely understand your hypothesis, but I don't think it's 100% on the hardware - I don't think it's due to the hardware of the hardware. I also have an M2 Air and a few things seem to consistently make my machine get hot and chug and dump the battery. Idling with my machine plugged into my 4k 144hz monitor isn't an issue, but going on a Zoom call or having one too many video streams open while doc…

In that case, zoom is transcoding video, which is a pretty damn heavy operation (note that this is why previous MacBooks didnt have 1080p webcams: they wouldn’t stand more than 5minutes on zoom without throttling).

My point here is that it’s not that the CPU (or the SOC to be exact) cannot handle such loads. It just that it cannot handle it when you have the device connected via USB-C and charging via USB-C too. Also, if using cheap usb-c to display port or adapters.

The corollary of this is that if that would even be possible, were your display wireless, or if you were generating frames by saving them on the disk, the CPU would not throttle.

Of-course that is of little utility here because the end result is the same: CPU gets slow, be it because of CPU, GPU or whatever.

Re: Building a macOS app to know when my Mac is thermal throttling

#122

Earlier quoted context omitted.

This could be total nonsense and I am definitely out on a limb here, but I remember from my amateur crypto mining days (almost 15 years ago wow) that keeping everything relatively consistent is better for the longevity of your hardware (GPU/CPU mainly). Some people obsessed over getting temps as low as possible but that’s not very productive, diminishing returns and all of that. There is just generally a healthy temp…

> Basically: swinging/fluctuating is not good, consistency is good. Yeah, I'd understand not wanting to go between 0C and 90C over and over. But my GPU idles at around 35C, maxes out at 85C or something, and going back and fourth will surely be preferable than staying with a single temperature but voltage clock the card. Especially considering performance. But again, I'm using my card for ML, number-crunching, simula…

My CPU/GPU basically lives between 35-60C, I don’t mess with voltage. Just a huge case with lots of air flow/fans.
Post reply on HN