How hot does your m2 MBA run? Doing what?

[jdb8167]

@jdb8167 I have also purchased a very cheap cooler (which is actually very quiet) for testing purposes and largely confirm what you report. It shouldn't be surprising that fanless computers which lose all the heat from the casing are much more sensitive to ambient temperature and the conditions around the casing, than internal cooled machine which pump the heat out in airstreams.

What I haven't got my mind round yet is that the fan cooler enables performance to stay high, but I also see the CPU temperatures staying high throughout long tests. Why doesn't it show them cooler?

I was reassured by the two hour Chess analysis and other long high load tests which showed that the CPU temps reported by Mx Power Gadget and iStat are being being reduced to a sustainable c80C. The 30 minute Cinebench run I just did with the cooler kept them over 102C. Not sure I feel comfortable about leaving it running for long periods like this, at least without understanding it.

Apple didn't design it to be sitting on a cooler so maybe it affects the throttling logic which enables it to protect itself in normal use during prolonged high load runs.

High temperatures shouldn't be a problem. What I think is happening is that the fans are removing the heat which allows the SoC to remain at a high level of performance. Remember heat is not temperature—they are related but not the same thing. As long as the heat is being removed why worry about a temperature sensor reading? If the temperature stayed high longer than allowed by the design the SoC would throttle. The M2 is also used in the 13" M2 MacBook Pro and that has a fan so you can't say the M2 was designed to work without a fan.

 

[Mike Boreham]

High temperatures shouldn't be a problem. What I think is happening is that the fans are removing the heat which allows the SoC to remain at a high level of performance. Remember heat is not temperature—they are related but not the same thing. As long as the heat is being removed why worry about a temperature sensor reading? If the temperature stayed high longer than allowed by the design the SoC would throttle. The M2 is also used in the 13" M2 MacBook Pro and that has a fan so you can't say the M2 was designed to work without a fan.

Thanks. Yes good point..it would be very interesting to know what the CPU temps are in the 13" M2 MacBook Pro with the same processor, but cooled by an internal fan. If they are c100C Apple obviously OK that as an operating temp.

I agree heat is not temperature but when considering component life, temperature is probably the more important. High heat flows probably mean big temperature gradients which can result in thermal fatigue.

There is still a piece of jigsaw missing for me. If 100C is deemed OK (because if wasn't the CPU would throttle to protect itself)....why does an ordinary run on a wooden table throttle back much more quickly? I am sure I am oversimplifying and there are good answers.

It is amazing (to me) how the M2 MBA responds to a simple fan cooled mat. The half hour Cinebench run I just did was 8644 on first run and 8312 on the last (19th) run..... a throttling drop of only 4% ! CPU core average was 104C on first run and 102C on last.

I should add that during the extended run high load tests I have done, my only concern has been about the internal CPU temps....the actual outside casing is never more than warm.

 

[jdb8167]

There is still a piece of jigsaw missing for me. If 100C is deemed OK (because if wasn't the CPU would throttle to protect itself)....why does an ordinary run on a wooden table throttle back much more quickly? I am sure I am oversimplifying and there are good answers.

This is a guess but it probably is because there are multiple sensors that are being used to determine when to throttle. If one or more of those sensors stays at a relatively low temperature there is no reason to throttle since multiple sensors will detect heat saturation. The fans in the cooler are preventing heat saturation and that means just having a single temperature sensor number probably doesn't mean very much.

I should add that during the extended run high load tests I have done, my only concern has been about the internal CPU temps....the actual outside casing is never more than warm.

(Another guess. It might actually be as simple as a single sensor that is detecting the case temperature. Since the case isn't getting very warm, no throttling.)

I've written my own menu bar temperature sensor application and there are dozens of sensors. You can watch them react differently depending on the load. I'm sure that the throttling behavior is complex and not something that can be understood from a single temperature sensor. Apple doesn't document any of this so everyone is just guessing.

 

[Mike Boreham]

This is a guess but it probably is because there are multiple sensors that are being used to determine when to throttle. If one or more of those sensors stays at a relatively low temperature there is no reason to throttle since multiple sensors will detect heat saturation. The fans in the cooler are preventing heat saturation and that means just having a single temperature sensor number probably doesn't mean very much.



(Another guess. It might actually be as simple as a single sensor that is detecting the case temperature. Since the case isn't getting very warm, no throttling.)

I've written my own menu bar temperature sensor application and there are dozens of sensors. You can watch them react differently depending on the load. I'm sure that the throttling behavior is complex and not something that can be understood from a single temperature sensor. Apple doesn't document any of this so everyone is just guessing.

Thanks. Yes certainly agree that throttling behaviour is complex!

So your bottom line opinion is that cooling pads safely allow operating at higher performance, and won't defeat the designed throttling protection?
Intuitively it would feel wrong that removing heat could make critical temps higher! but that is what it looked like superficially. I accept other explanations are possible.
Like you I don't intend to use my cooling pad as I am happy with the standard performance, but I do like to understand.

 

[jdb8167]

Thanks. Yes certainly agree that throttling behaviour is complex!

So your bottom line opinion is that cooling pads safely allow operating at higher performance, and won't defeat the designed throttling protection?
Intuitively it would feel wrong that removing heat could make critical temps higher! but that is what it looked like superficially. I accept other explanations are possible.
Like you I don't intend to use my cooling pad as I am happy with the standard performance, but I do like to understand.

I would use the cooling pad without worry if there was anything that I needed it for. I do occasionally use Handbrake so that is one possible use sometime in the future. Even then though, it is likely a run it and go do something else so I might not care how long it takes.

Everything I do for actual work runs fine on the M2 MacBook Pro and I never see temperatures higher than about 80 °C and then it is very brief and then the temps go back down to 35 °C almost immediately.

 

[JamesMay82]

Haven’t noticed any heating. I do most of my work on a web browser or on LibreOffice, but I occasionally do some work on Photoshop and Illustrator as well.
I come from a 2019 MBP 15” that felt like a burning stove only from being connected to an external display. My MBA now uses that display.

Ha can I buy your old one! It can heat my living room instead of me paying the gas company

 

[Mike Boreham]

It is amazing (to me) how the M2 MBA responds to a simple fan cooled mat. The half hour Cinebench run I just did was 8644 on first run and 8312 on the last (19th) run..... a throttling drop of only 4% ! CPU core average was 104C on first run and 102C on last.

Since a picture is worth a thousand words here are plots from the above runs. The fan cooled mat I used was this one.

 

[whizkid11]

Maybe it’s because I got 16gb of RAM but mine never gets warm even when flipping through a handful of apps and having 25 tabs open in Firefox.