Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Wando64

macrumors 68020
Original poster
Jul 11, 2013
2,349
3,129
I have just seen reports of the M2 MBP severely throttling due to inadequate ventilation and the M2 overheating.
How are they expecting this SOC to work on the new Air and, possibly, new iPads without crippling its performance down to M1 (or worse) levels?

 
Same as the Intel models which is basically nothing and position it as essentially a chrome book that can do SOME heavy lifting if needed. That said, folks do need to understand just because you can do it doesn't mean it's meant for it. Same for towing something, just because your small car can move a trailer doesn't mean it won't completely fall apart after a few miles.

Expectation management is key but like most tech people want it all for cheap. Just wait, you'll see people moan/posts about not being able to export 8K or whatever without the fanless laptop getting "warm". It will look something like this.

'My M2 MBA gets warm when editing Jurassic Park 12, is this ok?'
 
Other people are going to view it potentially differently but IMO M2's full potential won't be seen until it appears in the bigger laptops. In the 256GB machines they have partnered it with a single 256GB NAND chip instead of two 128GB chips. This throttles the read/write speed, which makes the real-life experience of using the computer slower than the equivalent M1 machine, even though the CPU and GPU are faster. That was a bad mistake. I don't know whether they chose to do it to rationalise their supply chain or whether 128GB chip shortages forced their hand.

From Max Tech tests there seems to be issues with excessive thermal throttling in the 13" M2 Pro because the cooling isn't up to the job, but hopefully the 14" and 16" should sort that. God alone knows how the fanless Air will cope, though.

Never a day goes by where I don't think I got overall the best deal by buying M1.
 
Seems weird that the image shows both 108ºC and a 1208 rpm fan speed. I'd wait to see if this behaviour is commonplace. The M2 shouldn't have a much different TDP than the M1 (other than the added GPU cores), it doesn't make any sense that the M1 is perfectly fine but the M2 is suddenly unusable due to unmanageable throttling.

Anyhow the whole idea that the M2 is somehow going to be slower than the M1 due to throttling is BS. The M2 is more efficient at the same wattage than the M1, so even if Apple needed to cap the power consumption at M1 levels, the M2 would be faster. The behaviour that youtuber has found for his benchmark seems indeed flawed, throttling shouldn't be as aggressive / work like that. But Apple can easily fix that on a firmware update.
 
Last edited:
Seems weird that the image shows both 108ºC and a 1208 rpm fan speed. I'd wait to see if this behaviour is commonplace. The M2 shouldn't have a much different TDP than the M1 (other than the added GPU cores), it doesn't make any sense that the M1 is perfectly fine but the M2 is suddenly unusable due to unmanageable throttling.

Right.
It’s hard to believe that there could be such a big difference. Alas…

What would be the point of using a severely throttled chip on fan-less machine such as the Air and/or the iPad?
You are not really going to move forward unless you can keep the thermals under control.

Let’s face it, unless Apple can do miracles, the ability of the Air to dissipate heat will be less than the mbp.
 
  • Like
Reactions: yitwail
The main value of M2 for performance is a significantly faster GPU subsystem (with lot of the gains coming from faster RAM and larger cache). It is entirely possible that the CPU performance under load on the iPad Pro M2 will stay comparable to iPad Pro M1, which won't matter a bit, since nobody is using iPads for sustained CPU processing.

Vadim's tweets do show evidence of throttling, which is not ok, but Apple will likely fix it with a power controller software update. There is probably a bug or other issue that results in the machine boosting higher/longer than it should.
 
It makes one wonder if they should have waited for the 3nm process node for M2, especially when it comes to iPads…

Rumor is the higher SKUs of M2 will be 3nm based.
 
  • Like
Reactions: MK500
Ok I’ll give them credit when it’s due: this is presentable evidence of thermal throttle. And the “sudden downclocks” is the worst imaginable solution to user experience.

Question is: Since they have publicly opened up and tampered around with their sample, possibly hit and shifted around the thermal solution, can this be reproduced by anyone else?

It makes one wonder if they should have waited for the 3nm process node for M2, especially when it comes to iPads…

Rumor is the higher SKUs of M2 will be 3nm based.
Keep in mind a key part of the M2 performance improvements was that they switched to DDR5. The M1 Pro/Max was already on DDR5 so the gains would be lesser if they don’t switch nodes. Thing is usually a node change meant cheaper chips but now on the level we are that’s not true anymore, so an M2 3nm this would likely have been a higher cost chip, especially when the node is new. For a MBP 14/16 you just charge the customer the extra cost and move on, but for a high volume chip like the M2 I get why they went with 5nm again.

Maybe there will be an altered 3nm M2 when it’s time for the iPads. But that’s a big maybe.
 
I'm fully convinced this thing was rushed to market. The outdated 13 inch Memebook Pro chassis that STILL has the touch bar, slower SSD speeds than the 2020 M1 model, and now this. What the hell is going on at the Mac team? The Mac was on fire with M1 but this year they've had so many misfires and bad choices.
 
Seems weird that the image shows both 108ºC and a 1208 rpm fan speed. I'd wait to see if this behaviour is commonplace. The M2 shouldn't have a much different TDP than the M1 (other than the added GPU cores), it doesn't make any sense that the M1 is perfectly fine but the M2 is suddenly unusable due to unmanageable throttling.

Anyhow the whole idea that the M2 is somehow going to be slower than the M1 due to throttling is BS. The M2 is more efficient at the same wattage than the M1, so even if Apple needed to cap the power consumption at M1 levels, the M2 would be faster. The behaviour that youtuber has found for his benchmark seems indeed flawed, throttling shouldn't be as aggressive / work like that. But Apple can easily fix that on a firmware update.
m1 is 15w,m2 is 19
 
  • Like
Reactions: yitwail
I'm fully convinced this thing was rushed to market. The outdated 13 inch Memebook Pro chassis that STILL has the touch bar, slower SSD speeds than the 2020 M1 model, and now this. What the hell is going on at the Mac team? The Mac was on fire with M1 but this year they've had so many misfires and bad choices.
…you think a chip that takes 3+ years of development was rushed?

The slower SSD is ONLY the base 256 model, because it’s one chip instead of two.

The 13” is simply an option, it wasn’t going to be redesigned because the 14” takes its actual place.

Look up “Gish Gallop” because that’s exactly what you’re doing here.
 
…you think a chip that takes 3+ years of development was rushed?

Doesn't matter how long it takes to develop something. It can be rushed when it's not ready. Just look at Cyberpunk 2077. Was in development for years and was still a broken mess.

The slower SSD is ONLY the base 256 model, because it’s one chip instead of two.

That doesn't justify this issue, especially when base spec is gonna be what most people will be getting.

The 13” is simply an option, it wasn’t going to be redesigned because the 14” takes its actual place.

Look up “Gish Gallop” because that’s exactly what you’re doing here.

Lmao?
 
m1 is 15w,m2 is 19

If true, then the cooling system in 13-inch MBP is 'rubbish' (perhaps indeed) 🤣

Rumours said M1 series was the last architectures overseen by GW III before he departed Apple to found Nuvia.

I would think M2 is still early to make a call on the effect from that event.

Seems to me what happened right now is TSMC caught a little cold, and Apple started sneezing. So, Apple should be doubling down its influence on TSMC to ensure mutual success. At the same time, Apple perhaps is already in talks with Intel and/or Samsung to mitigate risks in future.

I would also think it's a calculated move by Apple to let M2 as-is happen on 13-inch MBP and yet to be seen MBA. Few owners would have sustained workload on these machines. Most of their workloads are very bursty. So Apple can get away with any thermal issue. Good that Max Tech and other youtubers caught it, and can brag about it in their viewership.

14-inch & 16-inch MBP, and Mac Studio should have bigger thermal headroom to cope with M2 on N5P (Well, let's wait & see..)

I would also think the yet to be released M2 Pro Mac Mini with its re-design should have enough thermal headroom to cope with M2. 😅
 
They gave it a better GPU and more CPU clocks (which I'm against it, for me they should have improved efficiency) and still only one display per chip.

Doesn't make a lot of sense in my book.
The chip supports two displays: on MacBooks that means the built-in one and an external display. M1 minis natively support 2 external displays just fine.
 
  • Like
Reactions: KeithBN
Throttling on the Air is by design due to its position in the product stack. Apple made this clear when they removed the fan from M1 MBA.

If it didn’t throttle, it would perform as fast as the MBP 13.

Apple literally sells performance via thermal capacity. Base model iMac has 1 fan. Upgraded models have 2. MacBook Air has none, while MBP 13 has 1. MBP 14/16 has better dissipation capacity.
 
Throttling on the Air is by design due to its position in the product stack. Apple made this clear when they removed the fan from M1 MBA.

If it didn’t throttle, it would perform as fast as the MBP 13.
yea like that old MBA fan connected to nothing made a difference :))))
 
  • Haha
  • Like
Reactions: Lcgiv and macsforme
It makes one wonder if they should have waited for the 3nm process node for M2, especially when it comes to iPads…

Rumor is the higher SKUs of M2 will be 3nm based.

TSMC N3 (3nm) is a new node, and it will take some time before it becomes mature enough for big silicon (M1 Pro/Max equivalent).

Expect it first on small iPhone chips for a year or so, as the yield starts to improve with time.

And until then they have no option but to produce chips on N5P (M2) which is slightly better than N5 (M1).
 
Last edited:
  • Like
Reactions: kvic
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.