M2 is really just m1.1 (Update, I was wrong).

[teh_hunterer]

Imagine all those people getting MBAs with M2, they'll be able to cook bacon.

That remark aside, it seems I was quite right when it comes to the power draw and heat generation.

I think it'll just throttle more aggressively than the M1 rather than the chassis itself being hotter. Looking at the A14 vs A15, particularly the 5 core GPU version, it throttles more aggressively than the A14, but the level of performance it throttles to is still higher than where the A14 throttles to.

I don't watch Max Tech anymore, but my guess is they're making a huge deal out of the fact it throttles as if any throttling is automatically embarrassing to Apple and bad. What they're probably not talking about is where performance settles once it throttles.

 

[ArkSingularity]

I think it'll just throttle more aggressively than the M1 rather than the chassis itself being hotter. Looking at the A14 vs A15, particularly the 5 core GPU version, it throttles more aggressively than the A14, but the level of performance it throttles to is still higher than where the A14 throttles to.

I don't watch Max Tech anymore, but my guess is they're making a huge deal out of the fact it throttles as if any throttling is automatically embarrassing to Apple and bad. What they're probably not talking about is where performance settles once it throttles.

I'd like to see what sensors he is using. Something just doesn't seem right about the readings in his video. My M1 13" MPB rarely ever breaks 50C even under the heaviest loads, so I have a hard time believing he really got the chip up to 104C in Cinebench without doing something weird with the measurements (or setting wrong fan curves on purpose).

He doesn't exactly have the best reputation when it comes to this sort of thing. He might be right, but I'll need to see more measurements before I believe it.

 

[jav6454]

I think it'll just throttle more aggressively than the M1 rather than the chassis itself being hotter. Looking at the A14 vs A15, particularly the 5 core GPU version, it throttles more aggressively than the A14, but the level of performance it throttles to is still higher than where the A14 throttles to.

I don't watch Max Tech anymore, but my guess is they're making a huge deal out of the fact it throttles as if any throttling is automatically embarrassing to Apple and bad. What they're probably not talking about is where performance settles once it throttles.

You are right in that sense, but it means the MBA won't be the power monster people think it is. In fact, people will be burned (no pun intended) by how little the performance gains will be due to the inherent throttling.

 

[jav6454]

I'd like to see what sensors he is using. Something just doesn't seem right about the readings in his video. My M1 13" MPB rarely ever breaks 50C even under the heaviest loads, so I have a hard time believing he really got the chip up to 104C in Cinebench without doing something weird with the measurements (or setting wrong fan curves on purpose).

He doesn't exactly have the best reputation when it comes to this sort of thing. He might be right, but I'll need to see more measurements before I believe it.

He tested the M2, the M1 showed what we all know, it didn't even break 50.

 

[leman]

Intel has improved much faster these last few years. The i7 has improved from 10th to 12th generation as much as from 4th to 10th generation.

Not really suspiring given that Sunny Cove/Golden Cove are Intel's first real microarchitecture updates since 2014... and it's not like Skylake was a huge update over Sandy Bridge in terms of microarchitecture either...

I’m pretty sure this has been discussed here before. I think @leman had some information? In any case iirc it doesn’t kill silicon in any meaningful way.

From the observations, Apple tends to be as lazy with ramping up the fans as possible. This is a reasonable strategy, especially on Apple Silicon since many demanding workloads are rather short-lived and will be over before the thermal capacity of the system is saturated. You can quieter operation, better battery, and no detrimental effects to speak of.

Well, it does. I can't remember which MacBook Pro was it, but it suffered from issues from the heat generated which detached the GPU's BGAs.

Heat doesn't "kill" it silicone chips, it just allows them to desolder.

Solders have much higher melting temperature than 105C these chips maximally run at. The GPU solder issue was because of systematic manufacturing/quality issues and hardly because of chip running too hot.

 

[MayaUser]

You are right in that sense, but it means the MBA won't be the power monster people think it is. In fact, people will be burned (no pun intended) by how little the performance gains will be due to the inherent throttling.

Just wait and see how the high end M1 Mba will performed compared to the high end M2 Mba
If the cpu is over 10% increase and the gpu over 30% then its what people expect

 

[MayaUser]

Well, it does. I can't remember which MacBook Pro was it, but it suffered from issues from the heat generated which detached the GPU's BGAs.

Heat doesn't "kill" it silicone chips, it just allows them to desolder.

....material used in todays chip are for higher temps than 120C. Check TSMC materials used and check the chemistry
And im sure at 107C the system is shutting down

 

[MayaUser]

Not quite, as was mentioned, the MBP throttles as the fan curves seem to be broken. As soon as the reviewer fixed that, the MBP performed as expected. However, an MBA will not have a fan to help once the entire thermal system soaks in heat.

What reviewer fixed what? the cores runs at around 3.2- 3.3hz no matter how cold the SoC it is
So the Mba will run the cores also around 3.2 Ghz at 104C, so the improvements will come from larger cache , more bandwidth and more cores

 

[Carlson-online]

tick tock

 

[macguy360]

The new M2 chip is a Lemon.

Lets cook some eggs.

Even Max tech says the m2 is really just a m1.5.

 

[thenewperson]

Even Max tech says the m2 is really just a m1.5.

So?

 

[ericwn]

Even Max tech says the m2 is really just a m1.5.

That hot air bubble says a lot if the day is long and not enough views occur naturally.

 

[macguy360]

So?

This is a forum it’s a place to talk about apple products. Things we like, things we don’t like, thoughts etc. I’m not sure if you were aware but this particular forum post is about how the m2 isn’t much of an upgrade and is really more like an m1.1.

 

[thenewperson]

This is a forum it’s a place to talk about apple products. Things we like, things we don’t like, thoughts etc. I’m not sure if you were aware but this particular forum post is about how the m2 isn’t much of an upgrade and is really more like an m1.1.

And I asked you why it matters that Max Tech said it's an M1.5 (especially considering you claimed it was an M1.1). The reasoning Vadim gives for his assertion is silly. Is yours?

 

[ahurst]

Catch up with alder lake? So funny. Alder lake sucks power to achieve what it does, no offense but Intel groupies need a reality shot. No doubt alder lake is good for a high power chip, it’s small gains over m1 were cool, but m1 was 2020 and still much lower power than alder lake. Now m2 is right in line, no doubt next version of high power Intel chips will probably surpass m2, but so?

I think you misread my post, I was saying that Alder Lake was Intel’s panic to catch up with the rest of the industry after nearly a decade of underwhelming single-core generational improvements. Like I said, they achieved this partially by throwing a ton of power/heat at the problem, meaning I highly doubt they’ll be able to achieve similar generational improvements in the years to come.

 

[Juraj22]

This is a forum it’s a place to talk about apple products. Things we like, things we don’t like, thoughts etc. I’m not sure if you were aware but this particular forum post is about how the m2 isn’t much of an upgrade and is really more like an m1.1.

generally true, M2 is modest update to M1. We need to wait for armv9 architecture change that is already in progress. But that requires some time. M2 is buying that time.

 

[jav6454]

generally true, M2 is modest update to M1. We need to wait for armv9 architecture change that is already in progress. But that requires some time. M2 is buying that time.

You mean for M3?

 

[Juraj22]

You mean for M3?

M3 is marketing name. If it will have armv9 architecture, then yes.

 

[Slartibart]

Apple don't use ARM's microarchitecture.

Apple designs their own chips using ARM ISA (instruction set architecture). They design completely different chips separate from ARM.

 

[MrGunny94]

I'm a bit confused about those who were expecting a big jump in performance especially knowing that we would stay on the same. Regardless, this is targeted towards the people who have a Intel Mac / Windows computer.. The real upgrade path would be the M3 line.

I might get an iMac Mini M2 for Plex if they end up releasing it soon.

 

[jav6454]

M3 is marketing name. If it will have armv9 architecture, then yes.

Of course, just mentioning the likely marketing drive behind ARMv9

 

[Unregistered 4U]

Alright, I think you‘re dug in on the idea that it’s worth sacrificing desktop performance to be able to say it’s the same as laptop performance.

Not at all, just looking at reality, what just happened is that Apple made their best first effort at a processor that can run macOS code and, in so doing, produced a chip that with impressive single threaded performance. Performance that beat almost EVERY Mac that came before it, mobile or desktop. That’s remarkable. Perhaps what Apple should have done was to artificially hinder the single threaded performance of their mobile chip so that there would be more of a performance delta between the high end and the low end? I mean, there’s precedence for this… it’s what AMD and Intel have been doing for years.

I just don’t think so.

And frankly, not doing so will give x86 an ongoing edge in desktop computing. While AS will continue to be the killer laptop processor, Apple’s competitors will continue to be able to claim the highest raw performance.

Laptops have been outselling desktops for years. (Apple’s not the only company shipping FAR more laptops and mobile systems than desktops). And, the trend is that ultraportables part of the laptop market is on track to become the new top seller. Apple, being at the nexus of performance and efficiency in that market (where AMD and Intel will never be able to provide a performant AND efficient solution) puts Apple in a very good position… Not only with the Mac, but with the iPad, as well.

Multicore performance is another path to performance growth, but it’s a compromise. Not all workloads are adapted to multicore systems and when they are they rarely scale linear with core count. Improving the single core performance improves the multicore as a bonus.

I’m actually quite interested to see what the MacPro has in store. Do they keep their architecture sipping power, or do they open the throttle a bit?

The REAL compromise, in my mind, is mindlessly shoving massively more power and heat into the system. Architecting a balanced solution (software/hardware/OS) that pulls what it needs in order to provide the performance advertised is just good engineering. If it needs the power, so be it. My guess is that Apple won’t need the power for the M2 Ultra to perform considerably better than the M1 Ultra.

 

[quarkysg]

Perhaps what Apple should have done was to artificially hinder the single threaded performance of their mobile chip so that there would be more of a performance delta between the high end and the low end? I mean, there’s precedence for this… it’s what AMD and Intel have been doing for years.

I just don’t think so.

Yup ... Apple would want the same user experience when using a MacBook Air or a Mac Studio, i.e. to be just as snappy. Makes macOS planning and development a lot easier.

 

[Analog Kid]

Perhaps what Apple should have done was to artificially hinder the single threaded performance of their mobile chip so that there would be more of a performance delta between the high end and the low end?

Why do you keep repeating this silly straw man? Have I said anything that suggests I’m favor of this idea?

You do realize that your position is in favor of artificially hindering the single threaded performance of the desktop chip so there’s no performance delta between high end and low end, right?

I would like to proudly come out in favor of not artificially hindering anything.

Yup ... Apple would want the same user experience when using a MacBook Air or a Mac Studio, i.e. to be just as snappy. Makes macOS planning and development a lot easier.

Do you think events are still timed with for(; ; ) loops?

I seriously can’t get my head around the idea that someone thinks Apple should intentionally make their processors slower than naturally can be…. Maybe it’s time for me to just accept I’m being played…

 

[quarkysg]

Do you think events are still timed with for(; ; ) loops?

I seriously can’t get my head around the idea that someone thinks Apple should intentionally make their processors slower than naturally can be…. Maybe it’s time for me to just accept I’m being played…

Not sure what you're trying to bring across. My point is that Apple will unlikely go to the extend of intentionally crippling their CPU core performance for base model vs high end model processor. They will just have more of them.

Base model Mac with base Mx SoC will have the same or similar single core performance as high end model Macs with with high end Mx SoCs.

If all Macs have similar base performance for a given Mx family, IMHO, it makes it easier to tune macOS to work better with the SoC. Having YoY improvement is already enough of a variable to content with when it comes to software development. Having more variables across a single SoC family will make it worst, so I would think making it simpler for the software (both internal and external) developer is a core strategy for Apple as well.