14" M1 Max - 100C on Cinebench R23, within seconds. YT reviewers get 51C.

[crazy dave]

just to add here. Right after the part about sensors he states that ramping the fans to max should "lock" max performance of the chip. Obviously that is due to lower temps.
Literally what I am saying here.
Add a minute here and there and by the time the computer is ready to be replaced you made that extra $$ just from having a machine run always at 100%.

It's not just about longevity or chip/parts failure it is also about keeping peak performance as uninterrupted as possible.
To me this is worth million times more than noise or potential extra fan wear across computers lifespan. Easier to replace a fan than trying to make up for lost business.

That’s why the 16” comes with a high power mode

The main point of disagreement seems to have been whether it caused any damage… I don’t think anyone was disagreeing that you can squeeze out some extra performance by running the fans at full … though it’s not clear to me how much it gets you for the different Pro models - earlier in the video Linus mentions that the sustained performance of the early Pro model he tested was pretty close to peak. There was some throttling but not too much. I dunno for the Max and Pro, so far the high power mode on the 16” doesn’t seem to get you that much.

But I can see for your use case that it could very well be worth it.

 

[ha1o2surfer]

Maybe the app is misinterpreting the sensors? Mine gets hot just web browsing but istat hasn't been updated to see the SoC temps yet.

 

[ha1o2surfer]

I simply have realistic expectations and some actual knowledge, having managed hundreds of professional machines over my career.



You are running a data center? I though we are talking about laptops?

Google/Intel actually did an article about running higher data center temps and saw no increase in failure rates..

Google: Raise Your Data Center Temperature | Data Center Knowledge


"Intel recently conducted a 10-month test to evaluate the impact of using only outside air (also known as air-side economization) to cool a high-density data center in New Mexico, where the temperature ranged from 64 degrees to as high as 92 degrees. Intel said it found "no consistent increase" in failure rates due to the greater variation in temperature and humidity. "

I'd argue the only thing that lowering the inside temp of a laptop does is increase battery longevity.

 

[Argon_]

What is fan helper? I wonder if MaxTech is running something that he didn't tell us, but yeah if we up the fans it's going to cool down, but 53C with no fans running on multicore....false advertising lol.

Not false advertising. The 14 and 16 have vastly different heatsinks, in terms of capacity.

 

[Grohowiak]

Google/Intel actually did an article about running higher data center temps and saw no increase in failure rates..

Google: Raise Your Data Center Temperature | Data Center Knowledge


"Intel recently conducted a 10-month test to evaluate the impact of using only outside air (also known as air-side economization) to cool a high-density data center in New Mexico, where the temperature ranged from 64 degrees to as high as 92 degrees. Intel said it found "no consistent increase" in failure rates due to the greater variation in temperature and humidity. "

I'd argue the only thing that lowering the inside temp of a laptop does is increase battery longevity.

We're talking 2008 so not that recently BUT last paragraph pretty much reiterates what I am saying in regards to peak sustained performance and temperatures.
"You've got to find that sweet spot. But it's not where we are today. It's not 65. I think it's probably around 80 to 85."

Iceland’s data centers are booming—here’s why that’s a problem

Bitcoin miners and data scientists love cheap, green Icelandic processing power. Maybe a little too much.

www.technologyreview.com

 

[magbarn]

BOY.
If these M1 Chips get that HOT 95C! and they are already as small as 5nm, going to 4nm or 3nm is looking REAL REAL SCARY to keep them from burning up and burning OUT!

I think I'll Stick with Intel. Getting something rendered a few minutes faster is no big deal if you can't keep your CPU from burning up.

You do realize that intel MacBooks routinely hit 100C also.

 

[leman]

Did you happen to notice this part in the link you provided?

Or this part

You intend to run your laptop at 24/7 365 100C for twenty years?

That’s optimistic

I noticed it. This was his claim:

Clearly that statement is nonsense. His own link spells that out quite plainly.

Sure might look like it if you quote one random sentence out of the paper without the context.

But then again, reading is hard and not everyone has the education to deal with sources.

And Apple has had plenty of heat related issues in MacBooks over the years, partly due to inadequate cooling. Even as recently as the 2018 i9 MBP and the 2019 MBA. I never presented myself as an expert. Just someone that knows a ******** statement when they see it.

I gave you a screenshot of an i9 MBP running above the rated spec in a sustained workload (that was the 16”, sure, but the 2018 model is not much different). As to all these heat issues, please link us to the data showing that Apple laptops fail in droves because they run too hot.

 

[Grohowiak]

Maybe the app is misinterpreting the sensors? Mine gets hot just web browsing but istat hasn't been updated to see the SoC temps yet.

It kinda depends on the website.
With 38" connected na just MR + Spotify I'm at 44 with fans off.
Once I open trade view for example it goes up to 50+
Ad few more dynamic charts and such and 60 kick in.

 

[matram]

I find this thread somewhat confusing and I have got a Ph.D. In thermodynamics.

The power dissipated will be balanced by the heat removed which is proportional to the temperature difference between the CPU and the ambient temperature. The heat is likely transported by a combination of conduction to the chassis, natural convection and forced convection, i.e. the fan.

It makes sense not to start the fans full blast immediately, because other means of cooling can handle the lower heat load. The fan control is likely set so that fans speed up gradually starting from 50-60 degrees and the speed then increase to maximum at 100 degrees.

The fact that the CPU can run sustained at 100 degrees is the normal strategy by Apple.

Probability of failure of an IC increase with temperature but in the automotive context where I have worked the mean time to fail is in the million hour range, so way beyond what you need. Fun fact, we used 6000 hours as the life of a car. Even if you run your MBP 24/7 there is only about 8000 hours in a year.

 

[leman]

I don’t think anyone was disagreeing that you can squeeze out some extra performance by running the fans at full …

I disagree ?

They were talking about modifying the fan response curve, that won’t do anything for performance besides having the fans kick in earlier.

Running at 100% fans was somewhat effective with Intel models but that’s because the system will instantly pull over 80 watts anytime you literally do anything. This is not a concern at all with the ARM macs.

 

[leman]

BOY.
If these M1 Chips get that HOT 95C! and they are already as small as 5nm, going to 4nm or 3nm is looking REAL REAL SCARY to keep them from burning up and burning OUT!

I think I'll Stick with Intel. Getting something rendered a few minutes faster is no big deal if you can't keep your CPU from burning up.

Good luck with that.

Alder Lake mobile: PL1 45W, PL2 115W, PL3 215W

Apple Silicon: PL1 40W, PL2 40W, PL3… 40W

I wonder which one will get toastier

(Note: Apple CPUs don’t have PL levels like Intel, what I give above is a rough comparison)

 

[crazy dave]

I disagree ?

They were talking about modifying the fan response curve, that won’t do anything for performance besides having the fans kick in earlier.

Running at 100% fans was somewhat effective with Intel models but that’s because the system will instantly pull over 80 watts anytime you literally do anything. This is not a concern at all with the ARM macs.

? I think if you look at the 13” MBP in the video it does throttle to keep the machine quiet as well as cool and the new 16” MBP do come with a high power mode which, among other things, turns on higher fan rates. But the effect of the thermal throttling is small: the video mentions a negligible performance impact and, so far, reviews point out that the high power mode really is a small improvement as well. So while running the fans at higher speeds should result in “greater sustained performance”, it will probably only be a little better based on the reviews I’ve seen. If he’s decided that’s worth it, then hey it’s his machine and his Pascal’s wager.

 

[leman]

I think if you look at the 13” MBP in the video it does throttle to keep the machine quiet as well as cool and the new 16” MBP do come with a high power mode which, among other things, turns on higher fan rates. But the effect on throttling is small: the video mentions a negligible performance impact and, so far, reviews point out that high power mode really is a small improvement as well. So while running the fans at higher speeds should result in greater sustained performance, it will probably only be a little better based on the reviews I’ve seen.

I couldn’t observe any measurable throttling on M1 MBP I had, at least not in CPU operation. Of course, it will run a slightly lower clock than the peak single core, but that’s just how Apple configured it. I can imagine throttling in CPU + GPU operation simultaneously as the chassis can only handle around 20W.

 

[crazy dave]

I couldn’t observe any measurable throttling on M1 MBP I had, at least not in CPU operation. Of course, it will run a slightly lower clock than the peak single core, but that’s just how Apple configured it. I can imagine throttling in CPU + GPU operation simultaneously as the chassis can only handle around 20W.

Yeah it’s true that there’s no way to get it to run at peak single core with multiple active cores even if you have the thermal headroom. That behavior’s hard coded.

 

[leman]

Yeah it’s true that there’s no way to get it to run at peak single core with multiple active cores even if you have the thermal headroom. That behavior’s hard coded.

Which is a shame, the SoC should be able to handle it. I wonder why this restriction is there.

Well, these things can be more tricky than amateurs like us can imagine This thread already illustrates all kind of crap people believe in as if it’s some sort of holy scripture. This reminds me of this great little piece I’ve read recently:

Willingness to look stupid

 

[crazy dave]

Which is a shame, the SoC should be able to handle it. I wonder why this restriction is there.

Well, these things can be more tricky than amateurs like us can imagine This thread already illustrates all kind of crap people believe in as if it’s some sort of holy scripture. This reminds me of this great little piece I’ve read recently:

Willingness to look stupid

I often look stupid, willing or not

Yeah I’ve no idea why they did that - some sort of holdover from the iPhone? Or something deeper? There are so many interesting little bits about this chip. Maybe years from now, Apple will release more information and we’ll find out which decisions had deliberate designs and which were “oh f*** it ship it already”. Some you can tell already :::cough::display controller::cough:::, some we might find out when we see the M2 and what lessons they learned.

 

[elmalloc]

Another YT reviewer responded to my request to run TG Pro on his new 14/16s, he verifies our temps.

Spoiler on cinebench r23 multicore, he hit 100C on 14" and 97 on 16". On average though, I think the 16" is running about 10C cooler. CPU didn't throttle on any of his tests, I think, the scores were nearly the same in the 4 benchmarks he ran.

In those comments, it seems like the VERGE said the max has a "Significant hit" on battery life. Don't know what that means yet. I believe they (or someone else?) mentioned 14" pro at 10 hours, 16" max at 10 hours, 16" pro at 16 hours, but didn't show 14" max. We could use math to deduce 14" max battery life, I have a 14" max. I could do this test myself lol...sigh.

Not false advertising. The 14 and 16 have vastly different heatsinks, in terms of capacity.

Yeah, it was false advertising from that guy's YT. You can watch the video I posted above.

Now on average it was probably 10C lower during this tests (16" vs 14"). Not 50C lower. That would be documented in reviews and change some people's buying reasoning, I'd think.

Running 10C cooler on the 16", that makes more sense.

Maybe the app is misinterpreting the sensors? Mine gets hot just web browsing but istat hasn't been updated to see the SoC temps yet.

TGPro was updated October 27th to support M1 Pro/Max.

I think that actually answers my original question, the reviewer in question was not using the latest TGPro. MaxTech uploaded another video yesterday and did it again (still not using the latest TG Pro) and he's showing 40-50C in cinebench.

So the rest of the thread is just discussing if it's OK to hit 100C, interestingly the other YT reviewer above did this, and although some of his cores went red, appeared to be no throttling at 100C and performance between 14 MAX and 16 was the "same".

 

[diamond.g]

You intend to run your laptop at 24/7 365 100C for twenty years?

That’s optimistic



Sure might look like it if you quote one random sentence out of the paper without the context.

But then again, reading is hard and not everyone has the education to deal with sources.



I gave you a screenshot of an i9 MBP running above the rated spec in a sustained workload (that was the 16”, sure, but the 2018 model is not much different). As to all these heat issues, please link us to the data showing that Apple laptops fail in droves because they run too hot.

Would you buy a used Mac from someone that used it to mine cryptocurrency?

 

[leman]

Would you buy a used Mac from someone that used it to mine cryptocurrency?

I probably wouldn’t buy a used Mac at all. Terrible use of money.

 

[edfoo]

I also just tried Cinebench multicore test, and the 10-core M1 Pro in my 14" MBP also hit 100C. I stopped the test early after just 3 runs, not sure I wanted to burn this brand new MBP ?

 

[leman]

I also just tried Cinebench multicore test, and the 10-core M1 Pro in my 14" MBP also hit 100C. I stopped the test early after just 3 runs, not sure I wanted to burn this brand new MBP ?

You too bought this laptop to be a fancy paperweight? It’s not a candle you know…

 

[UBS28]

In Max Tech his review, the 14” even reaches temperature of 106 C. That is quite high.

The 16” went to around 91 C I believe.

 

[Grohowiak]

In Max Tech his review, the 14” even reaches temperature of 106 C. That is quite high.

The 16” went to around 91 C I believe.

Can you post a link?
I wonder how long is the test
Cheers

 

[Pmperry]

What temps were you getting on multi core?

I don’t think the laptop was getting loud, but tg pro showed it was getting hot.

Thanks!

His heat camera map showed quite a bit of heat too!

 

[bobcomer]

Google/Intel actually did an article about running higher data center temps and saw no increase in failure rates..

Google: Raise Your Data Center Temperature | Data Center Knowledge


"Intel recently conducted a 10-month test to evaluate the impact of using only outside air (also known as air-side economization) to cool a high-density data center in New Mexico, where the temperature ranged from 64 degrees to as high as 92 degrees. Intel said it found "no consistent increase" in failure rates due to the greater variation in temperature and humidity. "

I'd argue the only thing that lowering the inside temp of a laptop does is increase battery longevity.

They weren't testing machines with LIon batteries -- that's probably the first thing that will fail from heat in one of our laptops.