Where is the data that it does?
My 2014 MacBook Pro runs hot as hell is still fully functional. It’s well past it’s software update life cycle now.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
14" M1 Max - 100C on Cinebench R23, within seconds. YT reviewers get 51C.
- Thread starter elmalloc
- Start date
- Sort by reaction score
My 2014 MacBook Pro runs hot as hell is still fully functional. It’s well past it’s software update life cycle now.
This is talking about CPU temperatures. A beefier fan and heasink is still going to heat the chassis where the battery is. It’s a good thing these M1 chips run much cooler for most tasks than Intel chips. So the battery heat concern isn’t a problem.
Quite surprising, right? That a laptop SoC would stop exactly at the temperature that all CPU manufacturers of the last 10 years agree is the maximal safe operating temperature of CPUs. What a wondrous coincidence
You are entirely right: high temperature has indeed detrimental impact on batteries. But this is where the heat vs. temperature debate comes in. These chips might run at high internal temperatures, but they are still producing relatively little heat that can be efficiently extracted by the heat exchangers. The battery never needs to get hot. Even on my 16“ i9 that runs fairly hot most of the time (as I wrote above, it pulls more power doing casual stuff like text editing than an M1 Max under full CPU load), the battery temperature never exceeded 35C which is still in the acceptable region…
I know it was talking about that, but it doesn't translate well to a laptop, at least for me, i've had too many battery failures. Yes, they were all intel based machines, but I haven't had an M1 long enough. (2 laptops last year, one MBP, and one Windows laptop)
If you look at failure rate models it is clear that probability of failure for an IC increase based on temperature, power usage and environmental conditions. You can look at sources like MIL Handbook 217 that provide such models.
However the mean time to fail is orders of magnitude longer than the expected life span of the MBP.
Fun fact, electronics is not considered to age, probability of failure is the same at the first hour as the last.
I think this illustrates my point very well. People tend to misuse the term „throttling“. Running hot is not the same as throttling. In the Cinebench test, it is very clear that the 14“ still has a lot of thermal headroom, since the fans are not running at maximal speed. And it doesn’t look like running fans at full speed would get you more performance since the CPU is already running at full power. The chassis temperature is practically identical between the two laptops. The 16“ obviously has a more powerful cooling system, so it can run the test without turning the fans on. The 14“ has to try a bit harder. But again, you can see how Apple is utilizing the entire thermal operation range of the chip by only cooling when it’s really necessary. This gives you maximal performance, lowest possible fan noise and minimal battery expenditure on fans. It’s just a smart way if cooling the system. And the rest (omg my cpu runs too hot!) is just in people heads because they lack education.
Except for the weight difference (4.8 vs 3.5 pounds) you're better off getting the 16" for quieter computing due to lower fan noise, longevity since it doesn't cook itself like the 14", longer runtime with larger 100Wh vs 70Wh battery and lower fan RPMs drawing less power, smaller notch relative to screen size and better sound. All we need now is to see more data to decide if 64GB makes any difference in memory bandwidth and performance over 32GB RAM.
Last edited:
Lol lack of education. Big ego small ears that's for sure.
One more try lol
Lets use GPU as example
What's Thermal Throttling and How to Prevent It
Heat is an inevitable byproduct of work. Heat is also prevalent in electronics, and when it comes to graphics cards you can manage it using passive cooling,...
www.techspot.com
And maybe this
I get this is about Intel and it will take a while (if ever) if we we know how apple does it but there's no magic lol. It scales and gives better headroom but ultimately you will hit the barrier.
The way this machine uses fans is interesting. You can have a lot of safari tabs with forex data (dynamic) running and the fan won't even move. Download tradingview (app) and run it by itself (you can have 3-4 safari windows running this site without an issue) and all of a sudden fans go to 2600-2800 even though TG Pro is still showing similar temps.
Photoshop also kicks my fans into 2400+ with TG Pro temps looking stable. Very similar behavior.
Seems like CPU has much better way of managing heat than GPU. Big difference in fan behavior just cause of 1 simple app. Don't see anybody mentioning this in the test or here on the forums for that matter (or maybe I missed it) but it seems to me that everybody is following the same set of generic test scenarios instead of actually using this machine.
***I will add that this is with 38" connected. I will try without external screen later to see if it makes any reasonable difference.
Attachments
Last edited:
I'm nobody, but for what it's worth, @leman seems to be one of the brighter ones around here.
One more try lol
Lets use GPU as example
What's Thermal Throttling and How to Prevent It
Heat is an inevitable byproduct of work. Heat is also prevalent in electronics, and when it comes to graphics cards you can manage it using passive cooling,...
From the article: "In order to maintain performance, you need to control heat, but not all graphics cards experience throttling to the same degree, or even at all."
This points out exactly what @leman has been saying: Apple's implementations typically* allow for running the card at peak (advertised) performance, so messing with the cooling system seems like wasted effort, unless your aim is to overclock your card.
You'll only hit the barrier if the cooling system is unable to move heat at a high-enough rate, which doesn't seem to have been shown as an actual issue [they'd need to show actual performance drop below advertised rate for this to become empirically validated].
*My 2019 16" 2.3 GHz i9 has suffered from throttling (moving below the advertised 2.3 GHz) several times during heavy loads, but this generally has not been the case with most designs. I'm nearly convinced Apple released this 16" knowing it would throttle, as an additional way to gloat about how cool Apple Silicon is.
From what I am understanding he doesn't have the machine. Seem you don't either.
How are you guys determining this from generic reviews?
When they say not all graphic cards they are just clarifying that the term graphic card means same way for tiny chips to full blown behemoths.
My 14" is triggering the fans with system in the 54C DEPENDING on what I am doing.
I didn't even run real work on yet but I know already that in order for this thing to sustain performance hours at a time it will have to manage the heat.
Last edited:
Seems like there is a lot of talking past each other here.
I am simply going to accept that a manufacturer such as Apple will design their devices to specs that will not damage them prematurely when operated normally. So I find it unlikely that they would allow the temperatures to exceed levels that would prematurely age the devices. I guess premature here is a bit in the eye of the beholder, but empirically I think its clear Macs have a deserved reputation of longevity - I certainly have been using every one of mine for over 10 years, starting with my first Mac IIci, and hated to let them go in the face of increasingly painful performance delta's vs current models.
If that is true, it makes little sense, IMHO, to try to tweak operating temps lower than designed for. High school physics tells you that the efficiency of thermal transfer is directly proportional to the Temp difference between source and target, so all you are doing by lowering the source Temp is making the cooling systems work harder by making them operate at lower efficiency.
Even when trying to go beyond the design envelope, like overclocking or extreme use cases, cooling to lower than standard Temp would be inefficient, and I would guess you would do better to devise strategies to transfer heat at the higher, rated Temp.
But I'm in biomedicine, so what do I know?
I am simply going to accept that a manufacturer such as Apple will design their devices to specs that will not damage them prematurely when operated normally. So I find it unlikely that they would allow the temperatures to exceed levels that would prematurely age the devices. I guess premature here is a bit in the eye of the beholder, but empirically I think its clear Macs have a deserved reputation of longevity - I certainly have been using every one of mine for over 10 years, starting with my first Mac IIci, and hated to let them go in the face of increasingly painful performance delta's vs current models.
If that is true, it makes little sense, IMHO, to try to tweak operating temps lower than designed for. High school physics tells you that the efficiency of thermal transfer is directly proportional to the Temp difference between source and target, so all you are doing by lowering the source Temp is making the cooling systems work harder by making them operate at lower efficiency.
Even when trying to go beyond the design envelope, like overclocking or extreme use cases, cooling to lower than standard Temp would be inefficient, and I would guess you would do better to devise strategies to transfer heat at the higher, rated Temp.
But I'm in biomedicine, so what do I know?
Clearly CPU heat management is superior on this machines.
GPU seems to need much higher airflow to maintain temperatures.
Anybody know a tool that can monitor clock speeds and CPU usage in M1 Macs?
Activity Monitor is moving between 93 and 98 usage durning cinebench render and iStats is freezing when I run the test.
Guess need to wait for it to be updated.
Last edited:
Their test methodology was awful, unfortunately. They do not appear to understand the concept of experimental controls.
They needed to script the test load, make sure the displays were all at the same brightness setting throughout the test, and so forth. But they did the opposite of that, so their results don't mean a lot.
They did get some interesting information from Apple which confirms a couple suspicions I had:
I expected M1 Max to use more power at idle than M1 Pro. Not because there's twice as many GPU cores, but because there's twice as many LPDDR5 pcakages and twice as many LPDDR5 data pins. You can't fully turn off DRAM without losing its contents, so some extra power overhead is always going to be there.
The extra GPU cores can be fully turned off while idle, though. They shouldn't cost any energy until they're actually needed.
I also expected the thing about sometimes getting better battery life out of M1 Max on specific GPU loads. The power vs frequency curve of a core (CPU or GPU) is always nonlinear - to hit high frequencies you have to boost voltage, and power consumed by switching CMOS logic is proportional to both frequency and V^2.
Because of this nonlinear curve, when you use twice as many GPU cores but still only need the same throughput, you can downclock the cores, reduce voltage, and enjoy the benefits of using less power.
Now you might be asking "why not run faster instead?" The common place where you might not need to is graphics. There's not much point drawing faster than the screen refresh rate, and Apple's power management control loops seem to do a great job of automatically matching GPU frequency to demand.
Obviously, this isn't something you can depend on being true all the time. Like they said, "specific multi-core GPU workloads all day".
Normally, laptops would start to throttle when hitting 106 C temperatures out of self protection.
That it doesn’t throttle is not necessarily a good thing.
That it doesn’t throttle illustrates that it’s operating as expected. Laptops throttle when reaching over 105C (as this is the maximal temperature deemed safe by the manufacturer). The fact that the 14” can hover at 105C for a while without reducing performance shows exactly how incredibly precise Apples engineering is. The CPU never throttles. They keep fans exactly at a level to maintain maximal reachable performance without wasting any energy on cooling the system more than necessary. This is most impressive. Nobody else is this precise.
Found this analysis by Microsoft of why PC's fail. It contains some interesting conclusions, see section 9.
This dude is checking PMU temperatures instead of the CPU.
?
Last edited:
M1 Max benchmark in Luxmark. Beats the 5700 XT easily and according to this review it should be about the same as a 6700 XT.
Attachments
Can you show me how did you manage to determine the fact that this chip never throttles? Or that the performance isn’t actually reduced at certain levels?
I do see load fluctuations at higher temperatures but it’s not really scientific till I can actually measure / check voltage, cpu during sustained load and etc.
I wanna use this tool to but I don’t know where to get it.