M1Max running HOT

[crazy dave]

To be fair it’s not common knowledge that these core sensors are different from what was typically being measured before. So I can understand why someone might be confused looking at TG Pro or other sensor outputs. Although even these kinds of CPU temps shouldn’t be an issue for the most part - chip makers have adjusted what top level CPU temps should be from 90-105 degrees for every chip generation based on a bevy of thermal characteristics that they want to control for but by and large these temps aren’t high enough to damage the chip or other components (unless something else goes wrong most shut off around 115 or so).

 

[Havelock_Vetinari]

If you watch the videos in my post above timestamped to the point where they discuss thermals your laptop’s behavior is not abnormal for the new macs and unlikely to be an issue.


However if you’d like to play it safe, yes you can tell TG Pro to turn your fans on higher. But again wrt core temps: you are not measuring what you are used to measuring.

Yeah, the brief part of him addressing temps is exactly what I've been seeing. But he didn't really dive into it, other than recommending to run the fans with TG Pro. So I guess this is how they all run, although no one knows if this is how they are supposed to run.

Maybe in a few months, when more people get their orders, we'll have a larger sample pool.

 

[mi7chy]

How is it different? x64 CPUs have had in-die temp sensors for awhile now. Sounds like another 8GB = 16GB moment.

 

[crazy dave]

How is it different? x64 CPUs have had in-die temp sensors for awhile now. Sounds like another 8GB = 16GB moment.

And when they have they’ve shown similar temperatures. That’s why Linus says in the video that in their experience hot spot core sensors measure 10-15 degrees higher than the chip temperatures - and the latter are what people commonly think about and measure. Hence why they made that statement.

 

[crazy dave]

Yeah, the brief part of him addressing temps is exactly what I've been seeing. But he didn't really dive into it, other than recommending to run the fans with TG Pro. So I guess this is how they all run, although no one knows if this is how they are supposed to run.

Maybe in a few months, when more people get their orders, we'll have a larger sample pool.

They said they’ll be testing all the different chassis.

 

[SpotOnT]

All the reviews I have read mentioned two things. 1) the temp. sensors in M1 are located deeper within the core compared to the old Intel MBP. So you will see higher temp readings with M1 than you are used to seeing, and that is fine. How much higher for M1 to equal the same temp. as an old Intel MBP isn't exactly clear, but it is probably within the 10-20 degree range. 2) Apple is still using a pretty conservative fan curve, that tends to prioritize noise levels over thermals.

Now how accurate that information is or how the reviewer's know where the temperature sensor is located on the M1 chip is ?‍♀️, but that seems to be the word on the street.

In short your M1 Max is operating the same as all the other M1 Max. If this is the correct operation or if Apple goofed with their fan profile is hard to say, but again, the word on the street is that this is operating how it should be operating.

 

[jdb8167]

And when they have they’ve shown similar temperatures. That’s why Linus says in the video that in their experience hot spot core sensors measure 10-15 degrees higher than the chip temperatures - and the latter are what people commonly think about and measure. Hence why they made that statement.

I’ve been playing around with reading the temps on my M1 and one of the decisions is what temp sensor to use. I read all the values from the various CPU temp sensors and do you average or report on the highest temperature? I think using the highest temperature makes sense but I don’t have any basis for that.

 

[ksj1]

To be fair it’s not common knowledge that these core sensors are different from what was typically being measured before. So I can understand why someone might be confused looking at TG Pro or other sensor outputs. Although even these kinds of CPU temps shouldn’t be an issue for the most part - chip makers have adjusted what top level CPU temps should be from 90-105 degrees for every chip generation based on a bevy of thermal characteristics that they want to control for but by and large these temps aren’t high enough to damage the chip or other components (unless something else goes wrong most shut off around 115 or so).

Well yeah, but the premise (and title of this thread) is that the M1 Max runs hot. In practice my 2018 15" i9 with 32gig and a 512 ssd would heat the room, while the 16" Pro Max with 32 cores, 64gb and 1TB ssd I have now is ice cold for the same tasks. And completes them much faster as a bonus (unit tests and compilation).

 

[Lihp8270]

I am surprised that @cmaier went as long as they did.

They obviously know what they’re talking about and I’ve learned from reading what they’ve posted.

It’s a shame OP dismisses what they have to say in favour of internet myths and rumours. Just wanted to say that I at least appreciated the information you posted.

 

[crazy dave]

Well yeah, but the premise (and title of this thread) is that the M1 Max runs hot. In practice my 2018 15" i9 with 32gig and a 512 ssd would heat the room, while the 16" Pro Max with 32 cores, 64gb and 1TB ssd I have now is ice cold for the same tasks. And completes them much faster as a bonus (unit tests and compilation).

That’s fair as @leman pointed out in another similar thread: heat and temperature are related but not the same. The M1 Max maxing out at 90W outputs far less total heat than an Intel chip with dGPU maxing out at 220W.

 

[SpotOnT]

Those temperatures are not high at all. CPUs are designed to run at 100 C continuously for the entirety of their operating lifespan. When we model them for performance and correct operation, we assume the temperature is 100C, in fact.

That is very interesting.
May I ask what is the assumed CPU lifespan in the models you use?

 

[Havelock_Vetinari]

Well yeah, but the premise (and title of this thread) is that the M1 Max runs hot. In practice my 2018 15" i9 with 32gig and a 512 ssd would heat the room, while the 16" Pro Max with 32 cores, 64gb and 1TB ssd I have now is ice cold for the same tasks. And completes them much faster as a bonus (unit tests and compilation).

Running at 98C with only 30% fan speed is hot. The bottom of the laptop at 110+F is hot.

I see that the disagreement comes from the fact that most people compare it to previous MBPs, which apparently ran even hotter. I can't comment on that. My MBP replaced an i7 ThinkPad, and while the fans ran when it was under heavy load, it stayed quite cool, both to the touch and internally.

Anyhow, I bought the TG Pro and manually cranked the fans to about 80%...well, I know now why Apple is running this laptop as hot as possible. At 80% this must be the loudest laptop I've ever heard! Of course, I don't know what the Intel MBPs sound like, but to me the M1Max with 80% fans is a server-level noise, no joke.

Now that I've actually heard the fans, I am doing a full 180. There is no way Apple would do anything to have the fans run at faster speed than they are now, about 30% when fully loaded. I can live with that and hope there are no short-term implications, but there is no way I'd want to hear those fans ever again

 

[magbarn]

My M1MAX 16 runs pretty hot on handbrake too. What's neat is that it's almost as fast as my 5900x 12 core AMD desktop when transcoding H264 files (My BMW's rear seat entertainment can't play H265.)

 

[cmaier]

That is very interesting.
May I ask what is the assumed CPU lifespan in the models you use?

You can ask

Long enough that it would be unlikely that the CPU would be the thing that kills your computer.

People who think 100C is high would be very panicked by what goes on during burn-in testing

 

[iPadified]

Keep in mind that electromigration is something we account for in designing the chip. In fact, I wrote the tool we used at AMD. It is primarily an issue for wires and vias where current flows only in one direction. In modern chips, which use static logic, this is very rare. The wires where it is a problem (typically analog circuits like PLLs, amps, etc.) are designed to accommodate sufficiently high unidirectional current by being appropriately sized.

For older chips, where dynamic logic was used, or other forms of logic other than CMOS, this was a much bigger problem.

So atoms in the crystal are actually moving with electromigration?

 

[iPadified]

Running at 98C with only 30% fan speed is hot. The bottom of the laptop at 110+F is hot.

I see that the disagreement comes from the fact that most people compare it to previous MBPs, which apparently ran even hotter. I can't comment on that. My MBP replaced an i7 ThinkPad, and while the fans ran when it was under heavy load, it stayed quite cool, both to the touch and internally.

Anyhow, I bought the TG Pro and manually cranked the fans to about 80%...well, I know now why Apple is running this laptop as hot as possible. At 80% this must be the loudest laptop I've ever heard! Of course, I don't know what the Intel MBPs sound like, but to me the M1Max with 80% fans is a server-level noise, no joke.

Now that I've actually heard the fans, I am doing a full 180. There is no way Apple would do anything to have the fans run at faster speed than they are now, about 30% when fully loaded. I can live with that and hope there are no short-term implications, but there is no way I'd want to hear those fans ever again

According to the TI paper that leman referred to chips can be designed to hold for 10 years running contantly at 105 Deg C. According to that paper, at 98 Deg C, the life time would increase by 50-70% Do you really think Apple will run chips so hot so computers will fail in say 3-5 years? That would be equal to financial suicide and gross PR stunt for AMD/Intel. Not going to happen.

Uncomfortably hot laptop when having it in the lap use is an entirely another thing.

 

[leman]

I have been building my own computers since the time Yahoo was still a thing, and this is simply not true. For example, in my latest build I put in an AMD Ryzen 5800X, which has a Tjmax of 90C. No one would be throttling a CPU at 90C if it could run just fine and trouble-free at 100C.

Desktop CPUs and mobile CPUs are usually packaged differently (integrated heat spreader vs. no heat spreader), with temperature sensors in different locations etc. That’s why desktop chips often have lower Tjunction. This is part of the myths origin. But internal temperature is the same.

Mobile CPUs tend to throttle at even lower temps.

False. See above. AMDs mobile CPUs specify Tjunction at 105C

This is Apple's first CPU, so I don't know what you are referencing, maybe Intel? This is a new system and there is nothing really that we could reference it against. Comparing it to an Intel Mac is unfair, as they are simply not the same.

Apple has been building CPUs for over a decade (yes, A-series are the same abs they also reach 100C). And yes, I am referencing Intel (and AMD). A semiconductor CPU is a semiconductor CPU, no matter who makes it. Same materials, same basic technology, similar properties.

The first thing to fail in the system will be the weakest link. With such sustained temps, every component is affected, as you can see from the screenshot, not just the CPU. Heat is Enemy #1 to anything electric. I can't imagine that keeping your battery toasty wouldn't affect its life, for example.

What is your evidence that other components are toasty? The CPU can be hot but it doesn’t mean that the rest of the system is. Temperature and heat are different things.

Would the CPU fail within a year or two? I honestly don't know. I personally have never had a computer that ran this hot, ever. And I've done my fair share of overclocking and liquid cooling. There is just not enough information and telling me that Apple "tested it" is not reassuring.

Well, if you have a background in overclocking it’s no wonder you would worry about temperatures. It’s the overclocker community from where these myths about CPU longevity have originated. It’s exactly this kind of bro science that doesn’t let us have good things, because users start freaking out about perfectly fine running devices, which makes them prone to marketing BS.

Everything else aside, what I really wish I could find out is whether this is expected behavior with the M1 Max chips (not MBP in general). Since Apple is being so secretive, I really don't know if this is by design or if there is an issue with my MBP.

Yes, this is expected behavior. It occurs on every single device with Apple Silicon (A-series, M1,M1 Pro, M1 Max) and also on Apple devices with Intel processors. Apple has been running their CPUs (no matter the maker) hot for at least last 10 years. My source: I’ve been professionally managing hundreds of Macs in an environment where they are pushed to their limits.

Your machine is fine. It won’t hurt it’s lifespan in any meaningful way.

 

[Fomalhaut]

I like to keep temps below 70C and get concerned when it hits 80C. Around 100C is just nuts. That's what happens when marketing overrides thermal and reliability engineering so they can market it as fanless when it's really overcooked. Reminds me of how Jobs wanted the early Apple II revision and also III to be ventless for aesthetics and it caused overheating and longevity issues. Later II revisions corrected overheating with vents and work fine still over forty years later while III ended up in landfill.

Just pay for TG Pro and set your own fan profile to keep it no more than around 70 to 80C. Use Blender CPU as stress test to find upper temperature.

Please send us your resume showing your years of high-end CPU design experience, and we'll consider you for our next head-of-engineering position /s

 

[bousozoku]

Those temperatures are not high at all. CPUs are designed to run at 100 C continuously for the entirety of their operating lifespan. When we model them for performance and correct operation, we assume the temperature is 100C, in fact.

That's good. My 3rd generation quad-core i7 in a MacBook Pro would run at 100 degrees C when working on video. I don't see that much from my 7th generation quad-core i7 in an Omen by HP laptop, but I hear the fans more.

 

[Andropov]

So atoms in the crystal are actually moving with electromigration?

Yes. This is why bi-directional wires would have less problems with electromigration, as the movements caused by flow in one direction would be countered when current flows in the opposite direction, essentially averaging out* over time.

*for the wire as a whole. The position of any individual atom still probably can be roughly modelled as varying with something like the square root of time under current.

 

[Ifti]

People need to stop worrying and just use their systems. Don't look for issues if they do not exist....
Same with iPhone users stressing about batteries etc. You ruin your own enjoyment of your item....
IMO of course.....

 

[dimme]

On my M1 iMac the temps reported by istat pro and TG pro are different when the cpu is stressed. With istat pro giving an average reading. The average reading is lower and smother. I'm not sure why but just wanted to add it to the mix. I am enjoying this thread, and I am less concerned about the brief times my machine is at 100C.

 

[hefeglass]

C'mon, man, really? It's not the same and you know it.



False equivalency - just because a chip is throttled in order to avoid errors, does NOT mean it won't be damaged by high temps. I can attest to that, because I have literally fried a couple of chips myself.

Again, I'm not worried that these temps will damage the chip, I don't know if this would be the case. I know, however, that whatever Apple is using to draw the heat from the chip, let's call it "paste", will fail quickly with such temps. Anyone who has owned a MBP for a couple of years could probably relate to this - fans kicking in as soon as you open a browser, can't keep the damn thing on your lap, etc.
Now, for a lot of people this may be acceptable, and that's fine. My experience with non-Apple workstations is that it tends to happen 4-5 years into ownership, by which time technology has advanced enough to justify a new laptop (although you could re-paste it yourself, if you don't want to fork the money).

And there are other components affected by the heat, not just the CPU, most importantly the battery. I just hit the back of the MBP with an infrared thermometer (while under the same load) and the outside of the case is at 112F, with the laptop sitting on a ventilated aluminum stand, not a desk or a lap.

I just don't think it's intended to work like this. I am willing to bet that as more people get their MBPs, the OS will have an update that would tweak the fans behavior. Or not, we are talking about Apple here I am keeping my screenshots to compare them at some point in the future, though.

I love how this dude just doubles down on his made up theories to challenge someone who obviously knows quite a bit on the subject. "I build computers since yahoo was a thing"...wtf is that even supposed to mean? yahoo is still a thing. I have built computers since windows 3...I remember being super excited when windows 3.1 dropped
I SHOULD BE ARGUING WITH CMAIER

 

[Havelock_Vetinari]

I love how this dude just doubles down on his made up theories to challenge someone who obviously knows quite a bit on the subject. "I build computers since yahoo was a thing"...wtf is that even supposed to mean? yahoo is still a thing. I have built computers since windows 3...I remember being super excited when windows 3.1 dropped
I SHOULD BE ARGUING WITH CMAIER

Haha, if you don't know when Yahoo was a thing, I don't think you were even born when Win 3 was around. You shouldn't be arguing with anyone

 

[hefeglass]

Haha, if you don't know when Yahoo was a thing, I don't think you were even born when Win 3 was around. You shouldn't be arguing with anyone

dude..I was on Prodigy...then Compuserve...then AOL (before yahoo and the web )
...trust me id like to say I didnt use all those services and am not old af