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.
M1 Pro and M1 Max CPU Temperature Forewarning
- Thread starter hopkins802
- Start date
- Sort by reaction score
What causes capacitors to fail and pop? Excessive heat so it does play a role while you claim it doesn't.
Trying to get out of that hole you dug for yourself?
As an anti-Apple evangelist, I'm sure you're familiar with Louis Rossmann, who specialized in logic board repair for Apple laptops. It's pretty clear, at least for Apple's laptops, that 99% of cap failures are due to shorts caused by liquid damage.
I'll let mr_roboto's previous comment answer your apparent gotcha further:
This is what I sit at all day long, every day with my 16 inch MacBook Pro M1 Max, and I don't care. I trust Apple and their engineers enough that I don't lose sleep over this kind of stuff. If there's something wrong with the temps, the system should take care of it. If there is some catastrophic failure for some reason, that's why I have a Time Machine backup and AppleCare on the device.
I have the settings to System Controlled and notice my temps and fans are still at zero RPM. It just works! I've heard the fans kick on less than 10 times that I know of. Don't know how they do it, but I'm still continually amazed at the power and capabilities of these new chipsets.
I have the settings to System Controlled and notice my temps and fans are still at zero RPM. It just works! I've heard the fans kick on less than 10 times that I know of. Don't know how they do it, but I'm still continually amazed at the power and capabilities of these new chipsets.
Presumably those are degrees Fahrenheit, in which case it is barely warm.
If they are Celsius you would be having a meltdown
LMAO.
Polymer capacitor - Wikipedia
Wikipedia: "A polymer capacitor, or more accurately a polymer electrolytic capacitor, is an electrolytic capacitor (e-cap) with a solid conductive polymer electrolyte."
Back to me: as a rule, you can assume that any capacitor which has a polarity is a type of electrolytic capacitor. The electrolyte can be a liquid, as in old-school cylindrical aluminum electrolytics. Or it can be a solid, as in polymer capacitors. There's a huge number of different materials that have been used as the electrolyte in an electrolytic cap, and a huge array of different sizes and shapes of devices.
Do you get it yet, or do we need to go through yet another round? You seized on that one word "electrolytic" and tried to use it to make me look ignorant, but all along you've been doing nothing but exposing how ignorant you are.
Also, what a heat pipe does is provide a very low thermal resistance path between a heat source and a radiator. It doesn't directly cool other components like caps, it indirectly cools them by not having them get as hot in the first place. Without the heat pipe to channel most of the SoC's thermal energy output into the radiator, it would spread out and heat up stuff like the capacitors quite a bit more.
By PC world you mean Intel and AMD right? Because even Microsoft's own SQx Processors are ARM based and the temperature for that CPU will be different than Intel / AMD.
You are making assumptions which is why the first page of replies (this is how far ive made it on this thread) are the way they are.
You dont know the operating temp for Apple's M1 chips because I dont think they have published them. I did see you can push them to 110-115 before they start throttling based on some testing someone did with probes and stuff.
Either way, Anything you know about "traditional" processors I would ignore when it comes to ARM based CPUs
Okay, well, there is an easy solution to this debate-- @mr_roboto , just let your hardware bake and bake, run at 100 C for long, long stretches each day, year after year after year.... and after 10 years, please dig up this post, and let us know how your hardware is doing.... You seem enthusiastic about letting your computer run at 100 C, non-stop, fans barely kicking in, for years and years--so I say, go for it, dude! Rock that hot CPU, work it to the bone non-stop, cook that laptop of yours, and let us know how it is doing after 10 years.... Better still, give us an update after 5 years. I hope that this laptop gives you many years of productivity, joy and titillating, warm thighs. Whether or not it dies after a few years, it will definitely keep you warm on those cold winter nights, so that will be a plus either way.
Last edited:
I think we obviously have different approaches to our hardware. I'd expect that, that's why there's options out there. But, while I understand that you're happy there are systems that let you tweak all the different parameters to see what you can squeeze out, I'd ask you to understand that there are people like me who want to keep things as simple as possible.
Maybe you don't remember the days of having to set your own IRQs and then discovering months later that all those frustrating crashes were because you introduced a subtle incompatibility. Or the days when MacOS would start and people would have a dozen system extensions that left everything configurable, but far less stable. I'm happy to leave that behind and just use a very good computer at very nearly optimal performance.
To me, 5-10% performance gains will be outdone by technology improvements in 6 months or so.
What does 5% better on a gaming desktop even mean? Instead of running at 60fps, it runs at 63fps? It sounds like bragging rights more than actual value for the effort. There's not anything wrong with that-- people have been hot rodding cars for decades because they enjoy it. But for daily driving, I don't want to have to search for my basic controls among a bunch of options I, and most people, don't care about. There are already way too many options in System Preferences as it is.
My computer is a tool. I buy it, I use it, eventually I replace it. I don't have the time or interest to spend hours polishing it to get 5% extra performance. Honestly, if a job takes an hour to run, that saves me 3 minutes. Am I monitoring my machine carefully enough for the 57 minutes it's busy to do anything useful with it in those 3 minutes? No. I'm on another machine doing something productive until it's done, I'm getting coffee while I wait, I'm sleeping while a batch runs.
If the job is short enough that I'm willing to sit and watch the progress bar, then 5-10% saves me so little time it's essentially imperceptible.
In a few years I'll get a machine that's 50% faster or more, often much more on specialized tasks. That makes a difference in my workflow in the sense that it changes what I can do while I wait versus what I do in the background, 5-10% doesn't do that.
And I'm certainly not willing to jeopardize reliability in the short term (kernel panic) or long term (failed components) to gain those few percent.
I don't think you are understanding what I am saying or perhaps purposefully ignoring it? I am not saying that every user should be tinkering or overclocking or what not. I was responding to some absurd points you were making.
Like this one
I had already explained using BIOS as just one example how it is completely possible to let advanced users experiment if they want to and more casual or uninterested users completely ignore. Feel free to use your computer at stock settings, that is perfectly fine, but saying stuff like having hidden away and completely ignorable advanced options hurts you just by existing is laughable.
Then don't. You don't have to OC or change settings you don't want to just because they are there.
And stuff like this where you pull anecdotal evidence out of your behind. What an absurd thing to say. How do you know that people who overclock don't benefit from it. Having a 5-10% (in some cases even more depending on your setup) of free performance ALL the time is enormously useful for many use cases. Stuff like using your machine as a build server compiling code all the time, dedicated video encoding machines, training AI models, just as few examples, essentially any sustained workloads that can fully load your cpu or gpu cores or benefit from memory throughput will benefit from overclocking.
Having OC options available also have the benefit of undervolting, i.e. if you have a really good quality (binned) chip you could even undervolt it for the same amount of performance for even less power.
In any case, like I've mentioned, there's nothing wrong with using your system stock because you either don't have the time or the inclination to tinker, or for your use case that 5% is not worth it, or its too hard for you, whatever, its completely fine, nothing wrong with it. But saying stuff like 'Having them hurts me' or 'people who overclock don't really understand what they are doing and are just parroting' only makes you look like a fool.
Search for "polymer electrolytic capacitor" on Digikey and you get nothing
https://www.digikey.com/en/products...BwPYBsBPOAUyUxPxIGM0kDC0BLazalbFaptXJEALoBfIA
You have to search for "polymer capacitor" and under type it even says polymer.
https://www.digikey.com/en/products...YFcBOBDALigBABwPYBsBPOAUyUwGMVsUKBLNXJEAXQF8g
If you don't think temperature plays a role open your eyes and look at "lifetime @ temp" column. Majority are 1000 hours @ 105C. What do you think happens at >105C? It shortens lifetime or pops. Basic stuff even a hobbyist knows.
Be careful, he will throw more wikipedia links at you
. Apparently the goto resource for Engineers.
Jebus, this place has gotten toxic...
Then you go on to completely ignore what I'm saying... Not to mention that you keep talking about overclocking when the OP is clearly concerned about overheating and long term reliability.
But I do have to ask about this, though:
If you can point me to an article somewhere where a non-hobbyist is using overclocking in a serious way, for more than a YouTube headline, I'd be interested to see what they've found. Is Amazon tuning up the machines in their server rooms to maximize their fully loaded CPUs and GPUs? Is Microsoft setting BIOS parameters and stability testing their build machines? Pixar? They buy a ton of hardware, so I'd imagine if it was beneficial at the granular level then that benefit would scale up.
Or are they all fools too?
Then you go on to completely ignore what I'm saying... Not to mention that you keep talking about overclocking when the OP is clearly concerned about overheating and long term reliability.
But I do have to ask about this, though:
They will benefit from a higher clock, but I remain unconvinced that anyone seriously pursuing those workloads benefit from overclocking. Do you imagine this is ever done in a production environment? By which I mean somewhere with the resources, capital and labor, to invest in purchasing or developing the right tools.
If you can point me to an article somewhere where a non-hobbyist is using overclocking in a serious way, for more than a YouTube headline, I'd be interested to see what they've found. Is Amazon tuning up the machines in their server rooms to maximize their fully loaded CPUs and GPUs? Is Microsoft setting BIOS parameters and stability testing their build machines? Pixar? They buy a ton of hardware, so I'd imagine if it was beneficial at the granular level then that benefit would scale up.
Or are they all fools too?
Lol, I am glad you mention Microsoft and Amazon. If you seriously think that Azure and AWS just plop their hardware into cases and turn it on without spending massive effort on tuning and optimization then I don't know what to say. Big companies have been doing things like spending millions into RnD on optimized cooling, tuning and the whole 9 yards to squeeze more from their servers and they do it on scale.
Microsoft literally has research specifically into cost effective overclocking - https://www.microsoft.com/en-us/research/uploads/prod/2021/04/Zissou-Overclocking-ISCA21.pdf
Here are just few of the many resources you can easily find on the topic (either already deployed solutions or top tech companies like Microsoft doing serious RnD)
Overclocking the Cloud? Immersion Cooling Could Enable Faster Servers
Microsoft has been test-driving the use of overclocked processors running in immersion cooling tanks, and says the combination can boost server performance by 20 percent. The ...
Some quotes from this article -
Gamers have long used the combination of overclocking CPUs and water cooling to squeeze peak performance out of PCs. Can that same approach create more powerful cloud computing platforms?
New research from Microsoft suggests that it might. The company has been test-driving the use of overclocked processors running in immersion cooling tanks, and says the combination allows servers to perform at a higher level.
“Based on our tests, we’ve found that for some chipsets, the performance can increase by 20 percent through the use of liquid cooling,” said Christian Belady, distinguished engineer and vice president of Microsoft’s datacenter advanced development group. “This demonstrates how liquid cooling can be used not only to support our sustainability goals to reduce and eventually eliminate water used for cooling in datacenters, but also generate more performant chips operating at warmer coolant temperatures for advanced AI and ML workloads.”
This also goes to show that for many modern chips, the cooler it is, potentially more performance can be had from it. Sure a chip can run fine at 100C, but cooling it down a bit more can give you more boost clocks and thus more performance (not talking specifically about m1 here but in general with modern chips, thermal headroom can literally mean more performance a lot of the time)
Glad that you mention tuning as well, because as I mentioned in multiple posts, OC tools are not always about overclocking, they are also about tuning. Not only do the major compute providers invest heavily to do their own tuning but they also expose ways for the end user to do hardware level tuning themselves. Here are some AWS docs that detail how their EC2 instances allow for fine tuned and granular tuning of both the cpu and gpu to suit your workload.Processor state control for Amazon EC2 Linux instances - Amazon Elastic Compute Cloud
Some Amazon EC2 instance types provide the ability for an operating system to control the processor C-states (sleep levels) and P-states (CPU frequency).
docs.aws.amazon.com
Optimize GPU settings on Amazon EC2 instances - Amazon Elastic Compute Cloud
Learn how to configure GPU setting optimizations on NVIDIA GPU instances.
docs.aws.amazon.com
Another non-hobbyist serious scenario are big crypto farms that have millions worth of GPUs running with overclocked memory and finely tuned ASICS with full time staffed responsible for it.
I guess all of these are not serious enough things. You have to realize that just because you might not be aware of something, it does not mean that thing does not exist. The world is bigger than your home office
Last edited:
Good than that M1 series produces 1/3 of the heat of comparable x86 machines. If you are talking about temperature instead, most capacitors are rated up to 105C or even 125C.
My 2012 15" MBP that runs at 100C day after day is still running strong. ?♂️Okay, well, there is an easy solution to this debate-- @mr_roboto , just let your hardware bake and bake, run at 100 C for long, long stretches each day, year after year after year.... and after 10 years, please dig up this post, and let us know how your hardware is doing.... You seem enthusiastic about letting your computer run at 100 C, non-stop, fans barely kicking in, for years and years--so I say, go for it, dude! Rock that hot CPU, work it to the bone non-stop, cook that laptop of yours, and let us know how it is doing after 10 years.... Better still, give us an update after 5 years. I hope that this laptop gives you many years of productivity, joy and titillating, warm thighs. Whether or not it dies after a few years, it will definitely keep you warm on those cold winter nights, so that will be a plus either way.
Okay, well, there is an easy solution to this debate-- @mr_roboto , just let your hardware bake and bake, run at 100 C for long, long stretches each day, year after year after year.... and after 10 years, please dig up this post, and let us know how your hardware is doing.... You seem enthusiastic about letting your computer run at 100 C, non-stop, fans barely kicking in, for years and years--so I say, go for it, dude! Rock that hot CPU, work it to the bone non-stop, cook that laptop of yours, and let us know how it is doing after 10 years.... Better still, give us an update after 5 years. I hope that this laptop gives you many years of productivity, joy and titillating, warm thighs. Whether or not it dies after a few years, it will definitely keep you warm on those cold winter nights, so that will be a plus either way.
I spent almost a decade being professionally responsible for IT of a mid-size university department. We almost exclusively use Mac laptops. I am talking about folks who run heavy duty statistical simulations in their laptops overnight and sometimes for days at a time. With all this, I don’t recall a single hardware failure that could be linked to heat. I still have my old 2015 MBP work lying around that has been pushed should all kind of crap, including prolonged gaming sessions, and it still works well. Of course, I don’t use it anymore as newer laptops are faster.
Your computer will be long dead of unrelated causes before temperature-induced damage becomes a factor. Laptops have less thst 50% chance getting to that 10 year mark you mention the first place. If you want your computer to be working after a decade, all you need is luck - your computing behavior plays only a secondary role here.
Okay, well, there is an easy solution to this debate-- @mr_roboto , just let your hardware bake and bake, run at 100 C for long, long stretches each day, year after year after year.... and after 10 years, please dig up this post, and let us know how your hardware is doing.... You seem enthusiastic about letting your computer run at 100 C, non-stop, fans barely kicking in, for years and years--so I say, go for it, dude! Rock that hot CPU, work it to the bone non-stop, cook that laptop of yours, and let us know how it is doing after 10 years.... Better still, give us an update after 5 years. I hope that this laptop gives you many years of productivity, joy and titillating, warm thighs. Whether or not it dies after a few years, it will definitely keep you warm on those cold winter nights, so that will be a plus either way.
You'd have a point if we didn't have 16 years of Macbooks running at high temperatures approaching or at 100c day after day. If what you're saying is true you'd see a point where older Macbooks start failing in serious numbers due to heat related issues. Only, you don't, with the exception of specific failures like the bad GPUs in the 2009, 2011 models or the bad SSDs in the 2015 Macbook. Instead you have a bunch of armchair electrical engineers who mistakenly assume that they know more than the people who designed them because they read something on a forum or a tech youtuber said something alarmist, or other reasons. You're looking for a problem that doesn't seem to exist. Let it go.
You truly are a joy to talk to...The world is bigger than your home office
Lol, I am glad you mention Microsoft and Amazon. If you seriously think that Azure and AWS just plop their hardware into cases and turn it on without spending massive effort on tuning and optimization then I don't know what to say. Big companies have been doing things like spending millions into RnD on optimized cooling, tuning and the whole 9 yards to squeeze more from their servers and they do it on scale.
Microsoft literally has research specifically into cost effective overclocking - https://www.microsoft.com/en-us/research/uploads/prod/2021/04/Zissou-Overclocking-ISCA21.pdf
Here are just few of the many resources you can easily find on the topic (either already deployed solutions or top tech companies like Microsoft doing serious RnD)
Overclocking the Cloud? Immersion Cooling Could Enable Faster Servers
Microsoft has been test-driving the use of overclocked processors running in immersion cooling tanks, and says the combination can boost server performance by 20 percent. The ...datacenterfrontier.com
Some quotes from this article -
Gamers have long used the combination of overclocking CPUs and water cooling to squeeze peak performance out of PCs. Can that same approach create more powerful cloud computing platforms?New research from Microsoft suggests that it might. The company has been test-driving the use of overclocked processors running in immersion cooling tanks, and says the combination allows servers to perform at a higher level.“Based on our tests, we’ve found that for some chipsets, the performance can increase by 20 percent through the use of liquid cooling,” said Christian Belady, distinguished engineer and vice president of Microsoft’s datacenter advanced development group. “This demonstrates how liquid cooling can be used not only to support our sustainability goals to reduce and eventually eliminate water used for cooling in datacenters, but also generate more performant chips operating at warmer coolant temperatures for advanced AI and ML workloads.”This also goes to show that for many modern chips, the cooler it is, potentially more performance can be had from it. Sure a chip can run fine at 100C, but cooling it down a bit more can give you more boost clocks and thus more performance (not talking specifically about m1 here but in general with modern chips, thermal headroom can literally mean more performance a lot of the time)
Glad that you mention tuning as well, because as I mentioned in multiple posts, OC tools are not always about overclocking, they are also about tuning. Not only do the major compute providers invest heavily to do their own tuning but they also expose ways for the end user to do hardware level tuning themselves. Here are some AWS docs that detail how their EC2 instances allow for fine tuned and granular tuning of both the cpu and gpu to suit your workload.
Processor state control for Amazon EC2 Linux instances - Amazon Elastic Compute Cloud
Some Amazon EC2 instance types provide the ability for an operating system to control the processor C-states (sleep levels) and P-states (CPU frequency).Optimize GPU settings on Amazon EC2 instances - Amazon Elastic Compute Cloud
Learn how to configure GPU setting optimizations on NVIDIA GPU instances.
Another non-hobbyist serious scenario are big crypto farms that have millions worth of GPUs running with overclocked memory and finely tuned ASICS with full time staffed responsible for it.
I guess all of these are not serious enough things. You have to realize that just because you might not be aware of something, it does not mean that thing does not exist. The world is bigger than your home office
Ok, so I hadn't realized if I used the right BIOS setting, I'd suddenly find a 2-phase immersion cooler in my MBP, but I've learned something: you can make liquid from register settings.
So, let's see how your Googling lines up with your claims:
"Microsoft isn’t yet building entire data centers of immersion tanks"
So the answer to that one is no... As you say yourself, this is research.
“We anticipate observing an increase in the company’s hash rate and productivity through 2022, without having to rely solely on purchasing additional ASICs”
Cards based on custom ASICs, not standard PC hardware...
"The default C-state and P-state settings provide maximum performance, which is optimal for most workloads."
So changing from default isn't giving you even the 5-10% improvement you quoted, but it does mean you need to change the settings differently for different situations, as I suggested.
Macs don't use Nvidia cards, so there's not going to be a way to disable Nvidia autoboost, but there are ways to disable Intel's turbo boost if you want to:
How to Enable or Disable Turbo Boost on Your Mac
Want to increase your Mac’s battery life by up to 25%? Find out how to manually enable and disable Turbo Boost in macOS.
www.maketecheasier.com
I do appreciate you actually answering something with information in addition to the bile though. Yes, I concede that when the scale gets big enough, companies will invest in bespoke hardware and massive thermal dissipation systems to be able to run at higher clocks. They're doing it for 20% improvements on custom ASICs though, not 5% on a store bought PC. We know the big datacenters have been moving to custom hardware and massive cooling, that's not news.
That diversion is on me though. I should have been more precise. I thought it was obvious that custom hardware and dunk tanks was not what was being discussed. In this thread about laptops, what was being discussed was whether Apple should give you a setting to change to let you muck with low level processor settings like your windows machine does.
My feeling was that it's not worth the trouble in off the shelf hardware. It's not of value to enough people-- most people don't understand what they're doing and to really benefit you'd need to tune to specific workflows and, in the cases you're describing, keep your machine in a cryotank.
Yes, that is why "performance/watt" is really just "performance". Just about every system from iPhone to Datacenter is thermally limited, so therefore performance is limited by the thermals.
"Cooling it down" from 100C doesn't let you boost the clocks though. Extracting heat at a higher rate lets you run the clocks at a higher rate while keeping the temperate at or below 100C. If you run the fans more slowly, the processor heats up to 100C at a lower clock rate and throttles. If you run the fans more quickly, the processor can run at a higher clock rate before the temperature reaches 100C and it is forced to throttle. It'll run to 100C either way.
As I have told you here:
here:
and here:
Apple does give you a setting to run the fans at a higher speed to open up more performance.
That's fine if you're asking a question; but if you are "warning people" of standard, within spec and expected behaviour, its a little less so.
There was no question in the original post, just unwarranted fear mongering ("the CPU is about to melt"), so it was called out for what it was. ?♂️
No harm no foul, guy now knows better, play on.
It's OK to be wrong, but if you are, and posting it like it is a fact then expect to be called out for it.
When a computer dies mysteriously, are any of us really sure the exact reason, unless we study the board or components with some kind of specialized tools?
I had an iPad Air that died while gamiing, 2 weeks after I bought it..... It was certainly a defective product, but it was interesting that the constant gaming (which of course is very hot) over two weeks, seemed to bring forth its defect. Apple promptly replaced the device, and the one I have now still works great.
My MBP 2011's hard drive finally crashed after weeks and weeks of highly intensive tasks at the start of the COVID pandemic. It was constantly hot as h%ll, and at the end of a two week period, the hard drive died.
I also had a 2006 MacBook, where the board died, after a lot of consecutive heavy tasks with the CPU. Also blazing hot for several weeks consecutively on its deathbed.
So it is true that I don't know for certain if heat was the scientific reason for any of these failures... But all of these devices failed when they were noticeably hot, hot, hot--and had been for several weeks consistently. I would imagine that some of the devices that you looked after for years were also noticeably hot when they failed.... We can't say for certain that excessive heat was the cause, unless we study the board and the components with special tools, but the devices were certainly blazing hot and for long stretches when they failed, so there is some kind of connection.
I use a laptop fan stand now, repasted my other devices, and have adjusted the fan curves on my remaining Mac devices, some of which are very old. I haven't had any problems with hardware failure since then. Maybe there is a psychological comfort in keeping devices as cool as I can while using them, but everything that I have read about thermals and computers points to: a cool CPU and cool system is a happy CPU/system, and device longevity is connected to thermal performance/adequate cooling. I feel good about the fact that I air on the side of caution with thermals now and strive to keep my devices cool. It is comforting, and I think that I am taking good care of my devices by doing that, and hopefully they will last and last. My MacBook Pro 2011 with an SSD, my MacBook Pro 2012 with an SSD are still rocking, and I hope that my intel MacBook Air 2020 will benefit from the tweaks that Ive made to it to improve its cooling and thermal performance for the next decade.
But don't take it from me..... Take it from intel:
last sentence from the link below,
"....PC cooling isn’t just good practice. It’s also important for getting the best performance from your build, and for potentially increasing the lifespan of your components."
https://www.intel.com/content/www/u...g-the-importance-of-keeping-your-pc-cool.html
It doesn't matter if it is intel or apple silicon. You want to keep computers cool while they are running. It can increase the lifespan... Just like intel mentions in the above link. Apple most definitely prioritizes fan noise over the best possible cooling. If inadequately cooled components fail after 5 years instead of adequately cooled components failing after 10 years, this is great business for Apple, and keeps us coming back to buy new stuff, faster than we might actually want to......