I'd agree with this (and iirc there was a recent report that basically said this was the plan). They previously would skip ANX chips for years where there was no die shrink, so maybe that's just been moved to the Pro/Max variants.
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.
Will Apple's March Event be a big splash for M2?
- Thread starter DHagan4755
- Start date
- Sort by reaction score
I think we know the M2 gpu ...it will be around 40-50% improvement
The extra 2 gpu core, and the A15 gpu core > A14 one...so expect around 40-50%
Or maybe that's what they want us to think...
Seriously - we know we're discussing rumours here and there's an evens chance that March will see a refreshed iPad Air, some magical new Watch bands and squat all else. If it is a deliberate, strategic leak then it could be to damp down expectations.
There's absolutely nothing unusual about Apple updating an existing design with a new-generation chip. Happened all the time with Intel.
As for the touch bar - you'd have to change the case design slightly. Not a huge deal but it will have some extra cost associated. Whereas it is quite plausible that the M2 SoC might be a drop-in, pin-compatible replacement for the M1 (it would make an awful lot of sense for Apple to design it that way).
As others have pointed out - the M2 is supposed to be a replacement for the M1 and once the M2 manufacturing process is sorted out Apple are going to want to stop making M1s ASAP. If an M1 product isn't being dropped or replaced with a new design, the next thing to do is stick an M2 in it.
Because a lot of people seem to see the Mini as a starter to get working with macOS, if possible, they’d go all in in an all-in-one that could be used as a Mac Flati in clamshell mode or on a stand with an external mouse and keyboard. I know people that’d like a MBP but opt for a less expensive Mini with externals for the time being.
Imagine your Mini could be folded open and came with a great display, keyboard and trackpad installed.
I want an M2 Mac Mini in the old chassis compatible with the recently released Satechi usb-c base-dock thingy with integrated m.2 ssd.
The classic 2010 MacMini design needs to live on for all the accessories and rack-mounting hardware in colocation farms and stuff.
Who’s with me?
The classic 2010 MacMini design needs to live on for all the accessories and rack-mounting hardware in colocation farms and stuff.
Who’s with me?
Or maybe that's what they want us to think...
Seriously - we know we're discussing rumours here and there's an evens chance that March will see a refreshed iPad Air, some magical new Watch bands and squat all else. If it is a deliberate, strategic leak then it could be to damp down expectations.
There's absolutely nothing unusual about Apple updating an existing design with a new-generation chip. Happened all the time with Intel.
As for the touch bar - you'd have to change the case design slightly. Not a huge deal but it will have some extra cost associated. Whereas it is quite plausible that the M2 SoC might be a drop-in, pin-compatible replacement for the M1 (it would make an awful lot of sense for Apple to design it that way).
As others have pointed out - the M2 is supposed to be a replacement for the M1 and once the M2 manufacturing process is sorted out Apple are going to want to stop making M1s ASAP. If an M1 product isn't being dropped or replaced with a new design, the next thing to do is stick an M2 in it.
It would be much cheaper to change the top case to remove the Touch Bar than continue to manufacture and ship the Touch Bar.
Not necessarily. One of those can only be achieved by some level of re-tooling/re-design which needs up front cash-y money, the other is just cranking the handle on an existing process - not to mention the question of how many touch bars (or components thereof) they had in stock and/or had already contracted for - bearing in mind that they're a failed idea and Plan A was probably to keep using them across MacBook Pros.
That's --- a different product entirely.
If one has no need/desire for it to ever be a laptop, that's money wasted really.
Last edited:
Or they can just reuse the original design for the 2 port 13" MBP which didn't have a Touch Bar.
If Apple built all Macs to order, you'd have a point. But they don't. That's evidenced by them still selling 2018 iPad Pros (among many other products) via their refurb/clearance store.
Having an 'ambidextrous' M1 helps Apple. They're pretty good chips, and if Apple has loads at little carrying cost, they can use them in alternative devices as they see fit.
Not if the constraint is available design and engineering talent. Sticking with an existing design with the margins already established doesn’t cost any additional resources.
Only if the parts are identical. That’s not likely.
Last edited:
Aren’t Apple refurbs returns and trade-ins? How does having a refurb store have anything to do with having component stock for new manufacturing?
Just to play devil's advocate here... what if the 13" MacBook Pro still has the M1? Hear me out. They sack the Touch Bar but add MagSafe 3 & a 1080p webcam. On the desktop front, the high-end Mac mini rolls with M1 Pro chip, and we get the iMac 27" in colors. At or just before WWDC we get the 27" iMac Pro M1 Pro/Max with mini-LED & Mac Pro M1 Ultra (or whatever it will be branded), along with the standalone less expensive Apple mini-LED display. Thoughts on this idea?
Another option, and quite possible! I‘m only 50/50 on whether we’ll see the M2 next month, and a partially redesigned MBP13 with M1 is a possibility, although I suspect it might be short-lived. I’m hoping for a Mini Pro with M1 Pro/Max and some news on a larger iMac
It will be N4P or N5P, all of which are just extensions or ehancements of N5. N4, for what it's worth, is unlikely to be tangibly superior off the bat to N5 (and certainly not N5P) on power and frequency characteristics, for reasons I'll get to below.
I've seen the wishcasting caucus suggesting MediaTek's Dimensity 9000 SoC (on N4) has a process node advantage to the A15 (on N5P), for the most part this simply isn't the case any reasonable reading of TSMC's own documentation. N4 and the associated processes are an extension of N5 with similar design rules and N4 itself claims "possibly enhanced power via BEOL enhancement" and a 6% density improvement over N5, but 0 real power figures inherent to the transistors are disclosed compared to N5.
On the other hand, where N5P is concerned relative to N5: N5P is stated to generally exhibit a 7-10% frequency @ iso-power gain on N5, and a 15% power @ iso-performance gain on N5 (and plausibly more for Apple since they pushed the A15 so hard - e.g. at 2.7-3GHz on Avalanche, I imagine the A15's N5P really shines on power efficiency).
So N4 would likely prove to be a process downgrade from the A15's (likely) N5P process. No way it's N4, and then on top of that, Apple actually still have a process lead on MediaTek's Dimensity 9000 and/Or Qualcomm's Snapdragon 8 Gen 1+, which is said to be fabricated on N4. And N4P's characteristics? A good 6%+ frequency boost @ iso-power upon *N5P*. (Update: the figures are actually mixed on this note, but either way I'm confident N4P and N5P are probably superior to N4 on immediate power and frequency characteristics)
By the time the A16 comes out with N4P this fall — comparing it to said Qualcomm Snapdragon 8 Gen 1+ (plus variant on TSMC N4) or the Dimensity 9000 (N4) will by extension be incorporating a good *two* "enhanced" nodes (since even N5P is superior to N4) into the Apple performance figures which is going to be wildly annoying since everyone will miss that N4P is a successor to N5P which was already superior to N4, but whatever.
TSMC Expands Advanced Technology Leadership with N4P Process|Taiwan Semiconductor Manufacturing Company Limited
Hsinchu, Taiwan, R.O.C., Oct. 26, 2021 - TSMC (TWSE: 2330, NYSE: TSM) today introduced its N4P process, a performance-focused enhancement of the 5-nanometer technology platform.
Attachments
Last edited:
Yup. Take the arch upgrade from the A14 to the A15 - which is probably somewhat buoyed by N5P at that depending on how leakage plays in but whatever - and you've got a good 20-25% off the bat, throw in 2 more GPU cores and you've got a good 40-60% improvement in performance @ iso-power depending on the workload and the baseline measurement (as a fraction, the extra two GPU cores and new GPU arch will be a bigger boon to the baseline 9-core M2 model in comparison to the 7-core M1 model, but we're talking single digit artifacts of relative proportions, nbd)
But yeah it's funny, I've been musing on this too and think it's wildly underrated how much better the GPU will be. M1's GPU currently realizes 2.8 FP32 TFLOPS @ 10-10.5W of GPU power. We should see a good 4-4.5 TFLOPS in a similar (maybe slightly higher for peak figures albeit still a win if you comptrolled for the wattage) ballpark now. It's going to basically dismember the RDNA2 circleje- uh, hype cycles.
Tech companies don’t release new desktop/laptop processors so that those who purchased the previous generation can upgrade. They release them mainly for users who are looking to upgrade their much older machines.
The M series is only really significantly ahead by one metric - power efficiency. Apples newest pro chips are already behind Intel on pure performance. Apple holds no lead at all in the desktop space, where battery life is of no concern.
Last edited:
Will the average user care though? The only mainstream use for much faster graphics would be gaming, no? Far as I know gaming is nonexistent on the Mac. Is this a play to change that?
I find this emphasis on pure performance to be overrated as well.
For example, Apple's M1 pro chips are capable of sustaining their performance under load, and when not plugged into a power supply. Intel's processors narrowly beat Apple's chips, while consuming a lot more power, and generating a lot more heat in the process. I think it's clear what matters more to professionals.
In the desktop space, this will continue to be of importance in iMacs, which tend to be more thermally constrained as well. If there are indeed people for whom that little extra bit of performance is so important, and they don't mind all the associated drawbacks, then let them buy Intel chips. There's no point in trying to cater to this market demographic, and we still haven't seen what Apple has planned for the iMac Pro and Mac Pro yet.
I think it matters more to early adopters and tech enthusiasts than to professionals. Actual professionals tend to pick tools they are familiar with, because getting used to new tools can be a significant expense. Sometimes the tools are really old by consumer standards, because professionals rarely upgrade without a good reason. And when they do upgrade, they tend to favor something similar to their old tools over something better but different.
I know they’re both N5 technologies. N4 and N5P have very similar power/performance characteristics. Combining multiple sources of information they’re both around 5% performance improvement over N5:It will be N4P or N5P, all of which are just extensions or ehancements of N5. N4, for what it's worth, is unlikely to be tangibly superior off the bat to N5 (and certainly not N5P) on power and frequency characteristics, for reasons I'll get to below.
I've seen the wishcasting caucus suggesting MediaTek's Dimensity 9000 SoC (on N4) has a process node advantage to the A15 (on N5P), for the most part this simply isn't the case any reasonable reading of TSMC's own documentation. N4 and the associated processes are an extension of N5 with similar design rules and N4 itself claims "possibly enhanced power via BEOL enhancement" and a 6% density improvement over N5, but 0 real power figures inherent to the transistors are disclosed compared to N5.
On the other hand, where N5P is concerned relative to N5: N5P is stated to generally exhibit a 7-10% frequency @ iso-power gain on N5, and a 15% power @ iso-performance gain on N5 (and plausibly more for Apple since they pushed the A15 so hard - e.g. at 2.7-3GHz on Avalanche, I imagine the A15's N5P really shines on power efficiency).
So N4 would likely prove to be a process downgrade from the A15's (likely) N5P process. No way it's N4, and then on top of that, Apple actually still have a process lead on MediaTek's Dimensity 9000 and/Or Qualcomm's Snapdragon 8 Gen 1+, which is said to be fabricated on N4. And N4P's characteristics? A good 6%+ frequency boost @ iso-power upon *N5P*. (Update: the figures are actually mixed on this note, but either way I'm confident N4P and N5P are probably superior to N4 on immediate power and frequency characteristics)
By the time the A16 comes out with N4P this fall — comparing it to said Qualcomm Snapdragon 8 Gen 1+ (plus variant on TSMC N4) or the Dimensity 9000 (N4) will by extension be incorporating a good *two* "enhanced" nodes (since even N5P is superior to N4) into the Apple performance figures which is going to be wildly annoying since everyone will miss that N4P is a successor to N5P which was already superior to N4, but whatever.
TSMC Expands Advanced Technology Leadership with N4P Process|Taiwan Semiconductor Manufacturing Company Limited
Hsinchu, Taiwan, R.O.C., Oct. 26, 2021 - TSMC (TWSE: 2330, NYSE: TSM) today introduced its N4P process, a performance-focused enhancement of the 5-nanometer technology platform.pr.tsmc.com
N4P is 11/22% improvement on N5, 6/(11?)% on N4. N5P is 5/10%.
Thus N4 is basically N5P with a slightly higher density and simpler/cheaper manufacturing process. It is also available now if M2 is to be produced now. But Apple could stick with N5P since it is also available and already what the cores are being produced on. It’s just a pity that N3 slipped enough that the next iPhone cores (probably) won’t be made on it.
Last edited: