Why does Apple act like Intel is always holding them back?

[leman]

- Apple's laptop chassis were totally inadequate to run Intel CPUs without throttling: https://www.cultofmac.com/563617/macbook-pro-serious-thermal-throttling/

This article is factually wrong. 'Throttling" means that the CPU runs below it's base frequency. I have tested plenty of Macs myself and I have read a lot of reviews, and none were throttling.

However, when the 2018 model was initially released, they had a power management bug that would cause the power system to overheat and severely throttle the CPU. Early reviews quickly picked this up and Apple has fixed the issue in under two weeks. I assume that's where the throttling story begins and pretty much ends. And sure, a heavy gaming laptop can afford to run the CPU at much higher TDP (essentially turning it into a desktop CPU), but that's a completely different story.

 

[mode11]

Fair enough, I stand corrected. It seems the post-fix i9 MBPs were comparable (and quieter) to similar thin and light laptops from other manufacturers. I agree that a thick gaming laptop is an unfair comparison.

I would slightly dispute your definition of throttling, though. Laptop CPU clock speeds are designed to swing wildly (between 2.9 and 4.8GHz in this case) depending on workflow, to balance speed and battery life. Whilst a laptop that sits at base clock is meeting its spec, it's leaving performance on the table. A slightly larger cooling system (as was eventually used in the 16" MBPs and more recent Dell laptops) definitely bears dividends.

 

[leman]

I would slightly dispute your definition of throttling, though. Laptop CPU clock speeds are designed to swing wildly (between 2.9 and 4.8GHz in this case) depending on workflow, to balance speed and battery life. Whilst a laptop that sits at base clock is meeting its spec, it's leaving performance on the table. A slightly larger cooling system (as was eventually used in the 16" MBPs and more recent Dell laptops) definitely bears dividends.

It's not my definition though, it's the industry's definition The problem is, there are many ways to talk about these things. You can argue that the laptop is throttling while I can argue that the laptop is just returning back to it's normal speed after an opportunistic overclock, and technically we would both be right.

The core of the story is that couple of years ago mainstream CPU manufacturers had a big problem: they had no way to make chips that would be fast and power efficient at the same time. The solution was to introduce dynamic clocking (Turbo Boost as Intel calls it) — where the CPU would adapt it's performance and power consumption to the age scenario. This was a blessing, but also a curse in some ways, since power behavior of these chips became fairly unpredictable. For example, Intel sells some of the current CPUs as 15 watt parts, which would suggest to the customer that these chips run at 15 watts — but they can draw multitude of that power for brief periods of time, and they will draw much less power most of the time when the system is not doing anything useful. And of course users see these burst performance numbers and they confuse it for the sustained performance.

My point is that the dynamic overcooking was introduced in order to offer quick bursts of performance, where the CPU would operate above it's nominal spec. But the turbo clocks are not guaranteed — they are purely opportunistic tool. You can use a larger chassis that would allow the CPU to turbo indefinitely, but at this point you are not using a laptop CPU, you are upgrading it to the desktop level. Apple is already running the CPU at 50+ watts TDP in the 16" model, where it's nominal spec should be 45W... and some gaming laptops run those Cpus at 120W. But that can't be the goal. The purpose of these mine CPUs is well, to be mobile. If you buy a 15W CPU you should expect it to run at 15W and not at 50W. This confuses users, makes things difficult for the laptop manufacturers and allows the chip makers to use underhanded marketing tactics. For example, the Tiger Lake i7 could be either a 12, 15W or a 28W version, and you don't know which one you are getting, because they have the same model number. And the AMD 8-core CPU chips sold as 15W parts will run at 50Watts or more in multi-core benchmarks, long enough to produce impressive benchmarks scores. Of course, let them run for more than two minutes and they will clock down by 30-40%...

That's why I think that Apple Silicon is a breeze of fresh air in this department. It has a much more sane power behavior and it doesn't need ridiculous power skips to deliver good performance. Intel and AMD both rely on very high clocks to deliver speed, and as the results their cores need to use 5x the power between maintaining the nominal and the turbo speeds. Apple relies on "smart" code execution and their delta of sustained and turbo clocks are measly 200-300mhz.

 

[mode11]

Thermal throttling is really any situation where the CPU's built-in safety feature is reducing the clock speed to avoid damage, once it's reached its max safe operating temperature (e.g. 90°C). This could occur at relatively low clock speeds (due to a poor heatsink), or at high clock speeds (due to a not-quite-good-enough heatsink). In practice, the computer manufacturer will program the firmware to limit the clock speed / increase fan speed as necessary, so this doesn't need to happen.

If a laptop manufacturer limits a CPU's sustained boost clock to e.g. 3.1GHz, and its temperature therefore never goes higher than e.g. 85°C, then technically it won't have thermal throttled. This doesn't mean the cooling system is amazing though. A laptop with a better one may be able to hit e.g. 3.5GHz whilst staying below the same 85°C. Neither will have throttled.

 

[robco74]

If you want to talk throttling, then use a Windows laptop. Use one of the Razers that have very nice cooling solutions. Use it plugged in and marvel at the performance. Then use it on battery and wonder where that performance went. A Mac laptop will still give you near full performance on battery power - and likely still give you better battery life.

 

[mode11]

If you want to talk throttling, then use a Windows laptop. Use one of the Razers that have very nice cooling solutions. Use it plugged in and marvel at the performance. Then use it on battery and wonder where that performance went. A Mac laptop will still give you near full performance on battery power - and likely still give you better battery life.

Which Razer laptop, and which MBP? The MBP may give its full performance on battery, but how does that performance compare to e.g. the Razer Blade 15 Advanced, which clocks to 5.1GHz and has an RTX2070 Super..?

 

[robco74]

The Razer won't give you as powerful a GPU. The RTX series does perform better in gaming than the 55xxM in the 16" MBP. But look at battery life, and the Razer can handle little more than basic web browsing or video playback and still get decent battery life. Start using that GPU though, and it heats up and battery life drops considerably. You need to lug around the massive power brick with you if you want full performance. You can choose a higher performance power profile on the Razer, but battery life will suffer. You can read pretty much any review and they all come to the same conclusion, here's one: https://petapixel.com/2019/12/23/16...lade-15-studio-photo-editing-shootout-review/ There is a reason that Dell sticks to the GTX series for the XPS 15, and even then, it's the Max-Q version. If you want full performance that lasts on the go, then the MBP is the way to go.

When it comes to the Razer Book and XPS 13 with Intel's latest, they still can't match the M1 when it comes to GPU performance and battery life. There are several comparisons all over the interwebs.

 

[Ethosik]

This article is factually wrong. 'Throttling" means that the CPU runs below it's base frequency. I have tested plenty of Macs myself and I have read a lot of reviews, and none were throttling.

However, when the 2018 model was initially released, they had a power management bug that would cause the power system to overheat and severely throttle the CPU. Early reviews quickly picked this up and Apple has fixed the issue in under two weeks. I assume that's where the throttling story begins and pretty much ends. And sure, a heavy gaming laptop can afford to run the CPU at much higher TDP (essentially turning it into a desktop CPU), but that's a completely different story.

Thank you for stating this. People throw out the term "Thermal Throttling" nonstop. Guess what, to maintain a full 5.0Ghz in the i9-10900k ALL THE TIME you really need an AIO or liquid cooling and A LOT of tweaking.

 

[Ethosik]

Apple's mistake was not second-sourcing x86 chips from AMD, but since they're all in on their own chips now, this isn't going to be a problem anymore.

This really would not make sense. Apple started to focus on Apple Silicon in 2015.

Former Intel Engineer Explains Why Apple Switched to ARM

Apparently the quality assurance of Skylake processors was so bad Apple decided it had to make the leap.

medium.com

In 2015, AMD processors were not as impressive as they are now. I believe AMD really started getting running on Ryzen 2000. 8 cores for $299. This was in mid 2018. Why would Apple get a contract going with AMD for 2 years when they are transitioning to Apple Silicon anyway? Not to mention designing the OS to meet AMD processor (proper drivers and such) takes time. Redesigning the motherboards to fit AMD processors and more.

 

[mode11]

The Razer won't give you as powerful a GPU.

The Razer's GPU is much more powerful - it's the CPU in the review you posted that's not as good (i7 vs i9). But then PetaPixel is aimed at photographers: "while the GPU advantage of the Razer and other PCs might give them an edge in 3D rendering or some video editing tasks, it does very little for photographers"; "the most time-consuming task for most professional photographers, batch exporting, doesn’t even use the GPU". What if you use After Effects, or Blender, which do? The MBP is arguably a better balanced machine, but if you need lots of GPU power that's not really relevant.

 

[Argon_]

This article is factually wrong. 'Throttling" means that the CPU runs below it's base frequency. I have tested plenty of Macs myself and I have read a lot of reviews, and none were throttling.

However, when the 2018 model was initially released, they had a power management bug that would cause the power system to overheat and severely throttle the CPU. Early reviews quickly picked this up and Apple has fixed the issue in under two weeks. I assume that's where the throttling story begins and pretty much ends. And sure, a heavy gaming laptop can afford to run the CPU at much higher TDP (essentially turning it into a desktop CPU), but that's a completely different story.

That seems like a CYA definition the industry cooked up to diffuse customer anger. "Oh, it's not thermal throttling, it's just reducing it's power consumption closer to baseline because of thermal constraints."

CYA if I've ever seen it.

Here's the definition according to Merriam Webster for 'Throttle.'

According to that, one of these MBPs throttled because it reduced it's speed. Why did it throttle? Temperature. Seems like it... thermal throttled. The other machine just followed it's programmed clock speed curve without reaching 100C. It did not throttle.

Second image source: MaxTech

 

[mode11]

Exactly. The definition of "thermal throttling" [in a sustained-load situation] is not only when "the CPU runs below it's base frequency". It's when the CPU is forced to reduce its own clock speed (via bus multiplier) because the thermal solution isn't cutting it. In this state, it could still potentially be running above its base frequency. Of course, a properly designed machine won't allow this to kick in, because the clock / fan speed curves will have sensible settings in the first place.

When people complain about 'thermal throttling', though, they're usually suggesting the cooling system is insufficient. Perhaps the CPU or GPU can't run to their expected potential. Clearly, this is by design - as xWhiplash said, it's impossible to run most CPUs / GPUs flat out in a laptop, as they'll suck colossal amounts of power and require a cooling system to match. It's always a trade off, it's just that sometimes people wish Apple would put the emphasis more on performance and less on thinness. In any case, the M1 and its successors seem to have made this moot for the time being.

 

[robco74]

The Razer's GPU is much more powerful - it's the CPU in the review you posted that's not as good (i7 vs i9). But then PetaPixel is aimed at photographers: "while the GPU advantage of the Razer and other PCs might give them an edge in 3D rendering or some video editing tasks, it does very little for photographers"; "the most time-consuming task for most professional photographers, batch exporting, doesn’t even use the GPU". What if you use After Effects, or Blender, which do? The MBP is arguably a better balanced machine, but if you need lots of GPU power that's not really relevant.

I meant CPU. Razer sacrifices CPU performance to give you more GPU power. Given their target audience, it's probably the better choice.

Incidentally, check out what Pixelmator Pro can manage using the Neural Engine on M1, it's pretty impressive.

The point remains that in order to get the most performance out of a high-powered Windows laptop, you have to be connected to external power, which means, in most cases, also lugging around a large power brick.

But for the overall point of the thread, Intel has been missing their targets for some time now. Their old tricks of adding more cores and bumping up clocks and turbo speeds are having diminishing returns. The Nvidia RTX chips are impressive, but power hungry, even in Max-Q form.

Razer has done some impressive work around cooling, but battery life is still a problem. Most OEMs are limited to chassis design. They are unwilling or unable to invest in better components, and are content to cobble something together from off-the-shelf and then try to compete on price. Apple OTOH has been willing to invest in areas like battery technology and custom silicon.

 

[leman]

Thermal throttling is really any situation where the CPU's built-in safety feature is reducing the clock speed to avoid damage, once it's reached its max safe operating temperature (e.g. 90°C). This could occur at relatively low clock speeds (due to a poor heatsink), or at high clock speeds (due to a not-quite-good-enough heatsink). In practice, the computer manufacturer will program the firmware to limit the clock speed / increase fan speed as necessary, so this doesn't need to happen.

That seems like a CYA definition the industry cooked up to diffuse customer anger. "Oh, it's not thermal throttling, it's just reducing it's power consumption closer to baseline because of thermal constraints."

Ok, I think I understand where you come from now. The thing is, you are confusing the chicken and the egg.

There are many different kinds of throttling when applied to the relevant discussion: thermal throttling, power throttling, work throttling... You seem to see throttling as something negative, to be avoided, but throttling is an essential tool to manage the parameters (power, performance, thermals) of a system. In a realistic system, resources are finite. Throttling is a way to balance them. It's a limiting factor that lets you tweak the system's performance. You can throttle by temperature (you define a target temperature and start slowing your CPU down when that temperature is hit), you can throttle by consumed power (you'd fine the target draw power and start slowing your CPU down when that power draw is hit) etc. Many laptop manufacturers use the later approach. It's also the default for Intel CPUs (they actually have multiple throttling power level and an intricate algorithm to balance them out) — if you want to know more, this Anandtech article has a nice overview. Or you can read Intel documentation directly.

CPUs in Apple laptops are not hitting close 100C because Apple's cooling solution is inadequate. They are hitting 100C because Apple is deliberately using thermals as a limit to manage the power of the system. They completely unlock the Intel thrilling power levels and let the CPU run as fast as it can, relaying on their custom power management controller to throttle it instead (this is easy to see by querying the internal CPU registers responsible for power limits). If you look at the details: wattages, frequency curves, temperature, you will discover that there is a lot of linearity — which doesn't make much sense in a realm governed by quadratic relationships (I have posted exact graphs couple of years ago on these forums, sorry that I can't find them right now). This entire thing is engineered, and quite meticulously so. Apple designs their entire cooling system so that the CPU will hit exactly close to 100C (it's maximal safe operating temperature) while drawing its nominal TDP and running at it's nominal frequency. Why do they do that? Your guess is as good as mine, but I would speculate, its because it makes perfect sense. This allows them to make the cooling system as lean and quiet as possible, while spending minimal amount of battery power to keep the fans going and still being perfectly able to maintain the target performance.

Now, the reason why you might see a Dell hit 70C while offering the same performance is because Dell uses a different cooling solution and throttles by power instead. It's two different methods to achieve the same results, with their own set of drawbacks and advantages. The solution Apple uses is much more difficult to engineer, because you need utmost precision. You also get hotter running internal components, but who cares. This entire "oh my CPU is hot, it's going to damage computer" myth is long overdue to be put in a coffin, where it belongs. Apple has used temperature throttling for as long as I remember, probably as early as it's first Core 2 Duo Macs. There were never any reliability issues because of that. They use the same method to manage the power and performance of their M1 Macs as well.

In short: the useful definition of throttling is "the machine is running slower than advertised". Anything else doesn't make any practical sense. How performance is managed is a different matter. Apple has chosen to manage their performance by running CPUs as hot as they can be safely run. Other companies choose to be more conservative and limit the CPU by how much power they can consume. The end result is the same. The priority is different. That's all. Apple laptops are not throttling — they are performing as well as they should. It's just that people don't understand the technical details.

 

[Argon_]

Ok, I think I understand where you come from now. The thing is, you are confusing the chicken and the egg.

There are many different kinds of throttling when applied to the relevant discussion: thermal throttling, power throttling, work throttling... You seem to see throttling as something negative, to be avoided, but throttling is an essential tool to manage the parameters (power, performance, thermals) of a system. In a realistic system, resources are finite. Throttling is a way to balance them. It's a limiting factor that lets you tweak the system's performance. You can throttle by temperature (you define a target temperature and start slowing your CPU down when that temperature is hit), you can throttle by consumed power (you'd fine the target draw power and start slowing your CPU down when that power draw is hit) etc. Many laptop manufacturers use the later approach. It's also the default for Intel CPUs (they actually have multiple throttling power level and an intricate algorithm to balance them out) — if you want to know more, this Anandtech article has a nice overview. Or you can read Intel documentation directly.

I don't exclusively think that throttling should be avoided. It is obviously inevitable in passively cooled devices under sustained loads. And a bit of it is reasonable enough in an actively cooled consumer device where sustained loads are uncommon. I do think that a device marketed as PRO should not hit 100C under sustained loads. Otherwise you end up with a loud machine, a lap burner, or both. Which is unacceptable IMO. When a device is labeled "Pro" it should be able to run full whack, all day long, and have managed thermals without throttling. The M1 Pro satifies this. So does my old rMBP. Some intel touch bar era machines manage this. Others do not.

My most important point was the definition. If a processor slows down from its programmed continuous speed due to thermal constraints, it thermal throttled. Words must have consistent meanings, not adjust to the whims of the marketing department.

Besides, the "baseline" that Intel sets is a laughably arbitrary number, since clock speed is adaptive, and the real baseline, according to the definition of the word, is much lower.

 

[leman]

I don't exclusively think that throttling should be avoided. It is obviously inevitable in passively cooled devices under sustained loads. And a bit of it is reasonable enough in an actively cooled consumer device where sustained loads are uncommon. I do think that a device marketed as PRO should not hit 100C under sustained loads. Otherwise you end up with a loud machine, a lap burner, or both. Which is unacceptable IMO. When a device is labeled "Pro" it should be able to run full whack, all day long, and have managed thermals without throttling.

What you ask for is impossible. You have to limit the performance somehow. After all, you are using a chip that will happily draw 150 watts of you let it to (I’m talking about Intel here). If you want to make a laptop with that technology, you have to throttle it somehow.

And no, running a CPU at 100C doesn’t make the laptop hotter or louder. You are confusing heat and temperature. If you are willing to let your CPU run at 50 watts, you can either let it run at 100 C with minimal necessary cooling (like Apple does) or you can let it run at 70C with more aggressive cooling (like Dell does). The heat generated by both solutions will be the same (50 watts). The second system will be louder or bigger because it needs to run fans at higher power. That’s all. Macs run hot inside, but they tend to be cooler and quieter that other laptops with similar performance. And they dint blow hot air into your crotch.

Another point: if Apple were to design their cooling system to dissipate, say, 70 watts instead of 50 watts, their CPUs will still run at 100C - but at higher frequencies. In fact, that’s what Apple did with the 16” model. They have increase the cooling capacity by 15watts. The CPU still runs at 100C, but it’s faster.

You see the 100C figure and you get scared, because it’s a high number. But it’s not. It’s a perfectly safe way to run silicon. Intel themselves says so. Apple does this on purpose. Stop focusing on one isolated metric. Look at the whole picture. Again, Apple has been doing it with Intel Macs for the last 10 years.

P.S. I agree with you that a pro machine should be able to run all day without breaking a sweat. Well, complain to Intel. They lost the ability to make faster processors without increasing power consumption couple of years ago. Which is the primary reason why Apple is dumping them. Apple processors are faster and need less power. But they still run at 100 C. Because it’s a smart way to manage power...

 

[Argon_]

What you ask for is impossible. You have to limit the performance somehow. After all, you are using a chip that will happily draw 150 watts of you let it to (I’m talking about Intel here). If you want to make a laptop with that technology, you have to throttle it somehow.

And no, running a CPU at 100C doesn’t make the laptop hotter or louder. You are confusing heat and temperature. If you are willing to let your CPU run at 50 watts, you can either let it run at 100 C with minimal necessary cooling (like Apple does) or you can let it run at 70C with more aggressive cooling (like Dell does). The heat generated by both solutions will be the same (50 watts). The second system will be louder or bigger because it needs to run fans at higher power. That’s all. Macs run hot inside, but they tend to be cooler and quieter that other laptops with similar performance. And they dint blow hot air into your crotch.

Another point: if Apple were to design their cooling system to dissipate, say, 70 watts instead of 50 watts, their CPUs will still run at 100C - but at higher frequencies. In fact, that’s what Apple did with the 16” model. They have increase the cooling capacity by 15watts. The CPU still runs at 100C, but it’s faster.

You see the 100C figure and you get scared, because it’s a high number. But it’s not. It’s a perfectly safe way to run silicon. Intel themselves says so. Apple does this on purpose. Stop focusing on one isolated metric. Look at the whole picture. Again, Apple has been doing it with Intel Macs for the last 10 years.

P.S. I agree with you that a pro machine should be able to run all day without breaking a sweat. Well, complain to Intel. They lost the ability to make faster processors without increasing power consumption couple of years ago. Which is the primary reason why Apple is dumping them. Apple processors are faster and need less power. But they still run at 100 C. Because it’s a smart way to manage power...

The way you prevent it from drawing 150W is by programming a limit, or a stepped limit like the two port Intel Pros, rather than letting it get hot, then slow itself down.

Look. Here's a screenshot from the computer I typed this on.

In the case of my machine, the cooling system is adequately sized to the programmed power usage of the chip. That's all I'm asking.

Macs absolutely get hot, and when the cooling system is undersized, more of that heat ends up dissipating through the chassis, rather than exhausted air from the cooling system. Hot air is fine. A hot slab of aluminum much less so.

 

[Ethosik]

That seems like a CYA definition the industry cooked up to diffuse customer anger. "Oh, it's not thermal throttling, it's just reducing it's power consumption closer to baseline because of thermal constraints."

CYA if I've ever seen it.

Here's the definition according to Merriam Webster for 'Throttle.'
View attachment 1705645
According to that, one of these MBPs throttled because it reduced it's speed. Why did it throttle? Temperature. Seems like it... thermal throttled. The other machine just followed it's programmed clock speed curve without reaching 100C. It did not throttle. View attachment 1705647

Second image source: MaxTech

Intel cannot guarantee peak Turbo speeds unless you use liquid cooling and play around with some settings. Turbo is supposed to be short term high speed. Not constant speeds. Otherwise, these processors would be listed as 5Ghz not 4.2 with 5 turbo.

I have a i7-10700K in a custom built computer with a Noctua heatsink and two CPU fans. Base frequency is 3.8 and turbo is 5.1. Take a wild guess how long my CPU can sustain the 5.1 turbo? Sustained I can only achieve 4.3 Ghz.

Intel® Core™ i7-10700K Processor (16M Cache, up to 5.10 GHz) Product Specifications

Intel® Core™ i7-10700K Processor (16M Cache, up to 5.10 GHz) quick reference guide including specifications, features, pricing, compatibility, design documentation, ordering codes, spec codes and more.

ark.intel.com

If a Noctua cannot do this, you need liquid cooling which Apple will not use. Maintaining base or being slightly above is not the same as thermal throttling, otherwise every computer in existence thermal throttles that is not water cooled.

This is the CPU fan I have that cannot maintain turbo frequency - essentially "thermal throttling" according to some here.

Noctua NH-D15 chromax.Black, Dual-Tower CPU Cooler (140mm, Black)

The NH-D15 chromax.black is an all-black version of Noctua’s award-winning flagship model NH-D15 premium-quality quiet CPU cooler. Thanks to the same proven dual-tower heatsink design and NF-A15 PWM fans, it stays true to the NH-D15’s successful formula of rivalling the performance of all-in-one ...

www.amazon.com

Do people expect Laptops/iMacs to use this cooler which STILL "thermal throttles"?

 

[leman]

Look. Here's a screenshot from the computer I typed this on.

In the case of my machine, the cooling system is adequately sized to the programmed power usage of the chip. That's all I'm asking.

The way you prevent it from drawing 150W is by programming a limit, or a stepped limit like the two port Intel Pros, rather than letting it get hot, then slow itself down.

I am a bit confused, since your screenshot definitively shows a temperature-throttled machine. The cooling system is carefully engineered to keep the chip at it's maximal safe operating temperature when the chip is drawing 30 watts in sustained operation. It can run in this mode for basically forever, showing that the cooling system is adequate. This is exactly how any 15" and 16" Intel Mac operates as well. I though you were arguing agains this type of power management?

A power-throttled machine looks different: you would most likely see lower CPU temperatures while the machine bounces of the PL-levels instead of the temperature.

By the way, what your screenshot doesn't show is the initial frequency spike before the system has settled. Try setting the Power Gadget sampling rate a bit lower, and start the test again. You will see the chip immediately draw more power (probably closer to 50 watts) and run on higher frequencies, bouncing off 100C after a couple of seconds and then settling at around 95C/28W. That's how Apple thermal management works on Intel Macs.

Macs absolutely get hot, and when the cooling system is undersized, more of that heat ends up dissipating through the chassis, rather than exhausted air from the cooling system. Hot air is fine. A hot slab of aluminum much less so.

And yet Macs have consistently cooler chassis than most other laptops out there. And they are generally quieter. Look up notebookcheck reviews, they do measurements. It's not uncommon for a Windows laptop with a 45W class CPU to reach underside temperatures of over 50C — because they push the air out through the bottom chassis. Mac bottom chassis is on average around 10C cooler.

 

[the8thark]

Nobody outside Apple's management knows what Apple's top priority is.

Sure we do. It's making money (ie keeping their margins on products at certain levels) and others things that can only be talked about in the PRSI forum.

 

[the8thark]

Please just stop with the hypocrisy, preaching and posturing. You keep doing precisely what you’re accusing me of. Did it ever occur to you that you can scroll past something you disagree without having to push your agenda? You’re wasting your time.

I know the facts. I’ve been an “Apple person” for 30 years. Just because I prefer their platform doesn’t mean I need to stand up for or justify their questionable practices.

I think they make great products, but I’m also impartial enough to call it what it is, greed. I have no skin in the game other than my own money spent, so I am not a blind “cult-like” member either. I’ll leave that for the investors or for those who just have some twisted addiction/obsession with Apple. Those types are immediately identifiable by their continued defensive and passive aggressive attacks.

I know I have an unpopular opinion here in the “Apple lions den”, but you might wanna take a look over at your shoulder and remove your own gigantic “chip”. Like I said, the hypocrisy is pathetic.

Both of you are allowed to have an opinion without being called a hypocriate or accused of agenda pushing.
Neither of you have a chip, just differing opinions. Different opinions is a good thing, as it helps stir up good debate, assuming both sides are willing to listen to eath other.

Agree to disagree. Not even close to the markup on RAM that Apple charges. Besides that, most of these companies you name allow you to install your own RAM and HDs in their portables. They aren’t soldered on and are easily user replaceable.
[automerge]1593449945[/automerge]
Listen, my biggest beef with Apple despite how great their ecosystem and products are, is their shameless planned obsolescence. They solder RAM and HDs because they know that without a doubt, sooner rather than later we will buy new machines. So instead of milking a nice long life out of a laptop by upgrading the RAM and HD, we have to simply buy an entire new machine. It’s very simple. I’m sure the stockholders are giddy over this tactic.

The planned obselence is actually a good thing. Otherwise Apple would be stuck supporting legacy hardware/software from decades ago. Secondly why should the planned obselence affect you. I'm literally typing on a 2011 iMac now. It's still a great daily driver. Just buy the most ram and best CPU option you can afford at the time and your Mac will last you for a long time. Even if you can't use the latest OS, you should be fine for a long time.

Things like onboard soldiered ram expecially with the M series SoCs is one reason why the M series (and A Series also) are so good. When attacking something for being bad, please do mention the other good side as well.
I am surprised you are not milking the life out of your Mac products. Onboard soldiered RAM and SSD doesn't at all change their milkability.

 

[mode11]

Apple designs their entire cooling system so that the CPU will hit exactly close to 100C (it's maximal safe operating temperature) while drawing its nominal TDP and running at it's nominal frequency. Why do they do that? Your guess is as good as mine

Presumably so they can make their laptops as thin and as quiet as possible, long-term consequences be damned.

This entire "oh my CPU is hot, it's going to damage computer" myth is long overdue to be put in a coffin, where it belongs.

Unfortunately though, Apple laptops have a long history of GPU failures. Usually because of solder joints cracking under their BGA packages, due to the PCB flexing in response to large swings in temperature.

Apple laptops are not throttling — they are performing as well as they should. It's just that people don't understand the technical details.

This is a distinction without a difference. So what if the thermal protection type of throttling (not the regular, continuous power management) is not kicking in? Any competently designed machine will not allow this to happen. The argument is over whether Apple's cooling solutions were under-sized - and therefore relied on running components too hot and / or leaving performance on the table.

The extreme power efficiency of the M1 avoids this trade off for the moment, but it could potentially resurface in future. Apple may need to increase AS clock speeds to compete with new generations of x86 or ARM processors, or if they want to make very thin Pro machines with higher-end variants.

And they dint blow hot air into your crotch.

You've said this a few times. I'm not aware of any laptop that expels hot air towards the user - it's usually to the side or back. And despite the nomenclature, no-one using a laptop in anger literally has it on their lap anyway - it will be on a desk. I do agree that bottom mounted vents and a high underside temperature are awkward if e.g. surfing the web on a sofa or in bed.

 

[leman]

Presumably so they can make their laptops as thin and as quiet as possible, long-term consequences be damned.

[...]

Unfortunately though, Apple laptops have a long history of GPU failures. Usually because of solder joints cracking under their BGA packages, due to the PCB flexing in response to large swings in temperature.

I am not aware of any long-term consequences of this approach to power management. You mention GPU failures, but can you provide any proof that these failures have anything to do with high internal CPU temperatures? Not to mention that the most recent MBP generations — where the argument of "undersized cooling system" first started — are actually problem-free in this regard.

This is a distinction without a difference. So what if the thermal protection type of throttling (not the regular, continuous power management) is not kicking in? Any competently designed machine will not allow this to happen.

Temperature-driven throttling is not the same as thermal protection throttling. If your chip truly overheats, it will simply shut off.

The argument is over whether Apple's cooling solutions were under-sized - and therefore relied on running components too hot and / or leaving performance on the table.

I think I have provided a number of arguments that challenge this view. To summarize them:

- Apple laptops are not slower than other multimedia laptops with comparable form factors, and they quite often do better in keeping their sustained performance
- There is no evidence that running a CPU at temperatures close to 100C will result in long-term damage. The CPU manufacturer themselves say that it's perfectly safe
- There are no known CPU-related reliability issues in Mac laptops
- Mac laptops are systematically quieter and cooler on the outside than many comparable form factor laptops with the same hardware

In the end, Apple carefully matches the cooling system with the goal of providing minimal sufficient cooling. The entire thing is designed to maintain the CPU at close to 100C in sustained power draw (TDP) state. Their goal is reducing the noise, external temperatures and improving battery life. Many other manufacturers choose to overprovision and throttle by power consumption instead, because it's simpler.

The extreme power efficiency of the M1 avoids this trade off for the moment, but it could potentially resurface in future. Apple may need to increase AS clock speeds to compete with new generations of x86 or ARM processors, or if they want to make very thin Pro machines with higher-end variants.

Apple still uses temperature throttling to manage the M1, it's just more sophisticated. The power and temperature curves suggest some sort of complex ML-driven implementation that uses a sophisticated procedure to balance the resources out. It seems to me that they use a combination of temperature and power throttling, while also taking the processor utilization into account. I think this suggest that M1 can potentially run faster than what Apple allows so far.

I can't say I share your pessimism on this, given how power efficient Apple Silicon is. The heat issue only really surface when you want to run multiple cores at high clocks simultaneously, and Apple is years ahead of both Intel and AMD in this area. Eight high-performance M1 cores running at 3.0ghz would draw only around 40-45 watts — comparable to 8 Intel Tiger Lake cores at 2.0-2.4 ghz. And an M1 core is 50-70% faster per clock than an Intel core.

You've said this a few times. I'm not aware of any laptop that expels hot air towards the user - it's usually to the side or back. And despite the nomenclature, no-one using a laptop in anger literally has it on their lap anyway - it will be on a desk. I do agree that bottom mounted vents and a high underside temperature are awkward if e.g. surfing the web on a sofa or in bed.

Many laptops use direct air exchange, where the intake/exhaust grill is directly over the heatsink. Look at any modern Windows laptop — they all have bottom chassis riddled with holes. No wonder their bottom surface gets very hot. And it's not just about using a laptop on your lap — having the underside reach 50 degrees C or more has the potential to damage wooden desks.

 

[Apple Knowledge Navigator]

Why does Apple act like Intel is always holding them back from releasing current machines every year?

I actually don't understand your original question. What do you mean, "act like" Intel is holding them back? Has Apple criticised or questioned Intel's roadmap publicly?

If you're referring to their release strategy (which neither you nor I know has anything to do with Apple "acting" in response to Intel's products) then it all comes down to volume and demand.

Put simply, everything is relative when it comes to manufacturing; the cost of components, the order volume, the discount on the order per quantity, the final RRP of the product, the sales - all these factors create a final profit margin and risk.

So if we take the Mac mini as an example, bring together the facts.

- Pre-M1 model, Apple increased the RRP significantly. This would have affected sales based on perceived value and competition (which, arguably, was better value). Customers likely could not justify the increased price of the product in relation to performance.

- The thermal management was notorious for being poor under moderate load, which would have affected the customer's view of long-term usage.

- Furthermore, next generation CPUs of the Mac mini class didn't address lower core temperatures, which potentially wouldn't have been viable in the chassis's already poor thermal management (Ring any bells? Trash can Mac Pro...)

- The Mac mini is a low volume product segment for Apple, and on top of the fact that the CPU isn't shared with other Mac models, they would need to purchase fewer CPUs from Intel at a higher cost.

Those are the main detractors. If not enough customers purchased the product, then there is no incentive for Apple to figure out a way of selling off any unused CPUs and ordering the next generation, only for the same problem to occur. In other words, Apple shot themselves in the foot with that generation Mac mini.

 

[mode11]

I am not aware of any long-term consequences of this approach to power management. You mention GPU failures, but can you provide any proof that these failures have anything to do with high internal CPU temperatures? Not to mention that the most recent MBP generations — where the argument of "undersized cooling system" first started — are actually problem-free in this regard.

The CPU and GPU often share the same cooling system, so will be running at similar temperatures. I'm not saying the failures happen after 1 year, but the long term reliability of older MBPs has been questionable - even before the 'undersized cooling system' of recent generations.

Temperature-driven throttling is not the same as thermal protection throttling. If your chip truly overheats, it will simply shut off.

This sounds like splitting hairs. CPUs do clock down when they hit their thermal limit - they don't just switch off. Perhaps if someone pulled the heatsink off a running CPU that would happen.

I think I have provided a number of arguments that challenge this view. To summarize them:

- Apple laptops are not slower than other multimedia laptops with comparable form factors, and they quite often do better in keeping their sustained performance

Emphasis mine above. Sure, if someone copies the MBP design it will have the same characteristics - the physics is the same in both cases. The discussion is whether the form factor makes sense (other than aesthetically).

- There is no evidence that running a CPU at temperatures close to 100C will result in long-term damage. The CPU manufacturer themselves say that it's perfectly safe
- There are no known CPU-related reliability issues in Mac laptops

I agree with this. However, the cooling system may be the difference between e.g. hitting 100°C at 3GHz rather than 3.5GHz.

- Mac laptops are systematically quieter and cooler on the outside than many comparable form factor laptops with the same hardware

I agree they're quiet. Maybe effective fan design, use of copper heat pipes etc. helps here. If the case is cool, despite being made out of a highly heat-conductive metal, the heat must be being evacuated effectively, or heat would eventually soak through.

In the end, Apple carefully matches the cooling system with the goal of providing minimal sufficient cooling. The entire thing is designed to maintain the CPU at close to 100C in sustained power draw (TDP) state. Their goal is reducing the noise, external temperatures and improving battery life. Many other manufacturers choose to overprovision and throttle by power consumption instead, because it's simpler.

I still think Apple's goal is largely aesthetic: thinness + quietness. As you say, they achieve this by letting CPUs (and GPUs) run near their maximum temperatures, under careful control. They mostly get away with it. The thing about the Mac platform though is that the customer has no choice in this matter - if a MBP generation runs hot, has few ports or terrible keyboards, they'll still sell regardless, as the only alternative is moving to Windows. People grumble and put up with things because macOS is so nice.

I can't say I share your pessimism on this, given how power efficient Apple Silicon is. The heat issue only really surface when you want to run multiple cores at high clocks simultaneously, and Apple is years ahead of both Intel and AMD in this area. Eight high-performance M1 cores running at 3.0ghz would draw only around 40-45 watts — comparable to 8 Intel Tiger Lake cores at 2.0-2.4 ghz. And an M1 core is 50-70% faster per clock than an Intel core.

I'm not at all pessimistic about AS - it's amazing. I only mentioned AS to suggest how this discussion could be relevant to the future, as the Intel days are largely behind us now.

Many laptops use direct air exchange, where the intake/exhaust grill is directly over the heatsink. Look at any modern Windows laptop — they all have bottom chassis riddled with holes. No wonder their bottom surface gets very hot.

There's no obvious reason why having air intake vents on the underside would cause the bottom to get hot. Presumably the metal underside is being intentionally used as an additional surface to dissipate heat - otherwise, why not make / coat the underside with an insulator, rather than something highly thermally conductive?

And it's not just about using a laptop on your lap — having the underside reach 50 degrees C or more has the potential to damage wooden desks.

I'm sure most Windows laptops have feet on the bottom, if for no other reason than to let air into the holes you mention.