If they were going to do that wouldn’t it have been just easier not to say the M1 line was complete? Again, Apple’s plans haven’t changed since March.
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.
iPhone 14 Pro and Redesigned MacBook Air Reportedly Stuck Using Technology Behind A15 Chip
- Thread starter MacRumors
- Start date
- Sort by reaction score
Because they can. Regardless of the feature set of the next iPhone, they’ll sell over 200 million of them before the next iPhones are up for sale (unless they just can’t make enough). The number of people that care deeply about USB-C aren’t enough to affect that number.
3nm is definitely happening. It seems likely that TSMC’s plans for 2 nm are also going to happen. Intel seems confident enough in sub 1 nm that they are planning to use Ångströms as a node designation. We have a ways to go before shrinks stop.
It’s not. As we learned a few years ago, whenever you read a report that talks about Apple having a decision to make NOW, that decision was actually made 12-24 months ago and it’s just getting out. That’s how much better the security has gotten over the companies that make Apple’s devices. Of course, those providing the information won’t say how old it is because that would highlight how NOT in the loop they are.
I have a few Ultras on order but even my Apple rep said he has no idea when I'll get them.
Multicore uplift is huge with Ultra....but single core perf is the same as the $600 base model Mac mini.
M2 is still going to have more advanced A15-class performance cores, efficiency cores and graphics cores versus the A14-class cores in M1. M1 is also fabbed on N5 whereas M2 will be fabbed on the improved N5P process which offers better performance and efficiency. M2 will also have 9/10 GPU cores vs. the 7/8 of M1. Taken all together, M2 is going to have real performance benefits over M1.
A16 is a bit more nebulous, as I don't believe there have been any leaks as to what performance, graphics and efficiency cores it will use. I have seen articles with the belief the graphics cores, at least, will be improved so even if the A16 sticks with the same performance (Avalanche) and efficiency (Blizzard) cores as A15, Apple will likely clock them higher so A16 will still benchmark better in then A15 (especially in graphics).
A16 is a bit more nebulous, as I don't believe there have been any leaks as to what performance, graphics and efficiency cores it will use. I have seen articles with the belief the graphics cores, at least, will be improved so even if the A16 sticks with the same performance (Avalanche) and efficiency (Blizzard) cores as A15, Apple will likely clock them higher so A16 will still benchmark better in then A15 (especially in graphics).
What are you talking about? 96% of mainstream applications are already Apple Silicon ready: Microsoft Office, most of Adobe Creative Cloud. Honestly, whatever parochial app you are talking about already runs fine using Rosetta. There is nothing really to 'unleash' or 'exploit' that hasn't already. Even Creatives who tests the Mac Studio have admitted, there isn't much to gain choosing a M1 Ultra over an M1 Max. You have even proven my point more, what's already released is good enough to not warrant an M2 now, 6 months from now or even 2 years from now. The rumors are saying an M2 is not gonna be dramatic like the M1 was and we'll probably just see a 6 to 9% performance boost. Thats not gonna get many on M1 Macs, especially those who already spent a pretty penny on a Mac Studio. Whoever Apple is building the Mac Pro for is extremely niche. So, unless you are ready to blow 10 grand on one and there are not many of you out there that gonna do that (especially in this economic climate), then more power to you. But lets remember, Apple stop caring about niche when the iPhone became mainstream.
Almost nobody upgrades every cycle and few upgrade even every other cycle. It was the same under Intel, PowerPC and Motorola (68K) as it will be under Apple Silicon.
However, there are plenty of people who are still using Intel Macs (or have PCs and are now interested in Macs) and those are the people who will buy M2 models.
Very niche even those who are on Intel Macs. Mac users are notorious for keeping their Macs very long. Go take a look in the Mac Studio forums, there are so many upgrading from 10 years Mac Pro's or iMacs. But there will be people who do upgrade based on whenever they had bought one. Which is why even the iPhone is safe. There will always be a cycle of iPhone users. I'm sure iPhone 6s and iPhone 7 are looking at getting new iPhone this fall.
Has anybody here ever heard of IPC gains and architectural improvements?
What a dumb title “stuck using technology behind A15” when referencing just the fabbing node.
Macrumors article stuck using the same formatting and font of the previous Macrumors article, no use in reading it.
What a dumb title “stuck using technology behind A15” when referencing just the fabbing node.
Macrumors article stuck using the same formatting and font of the previous Macrumors article, no use in reading it.
I agree. We could get a 5nm M2 next week (as you stated), and then the M2 Pro / M2 Max / M2 Ultra could be released next year on a 3nm process, in time for updated MacBook Pros and Max Studio.
“Reputable” sources have been talking about M2 for a while now, and it’s always “coming next keynote”. I want to see more dedicated software and games not going through Rosetta. What’s the use of all the power, without the software. They’re ahead enough where they can play to their strengths of their integration, and not need a power leap.
Apple is so far ahead of the competition that its not really a competition from the CPU standpoint.. Now release a dedicated graphics card for certain machines... And that will be really leave the playing field in the dust
3nm is. 2nm is dubious. They are still working on some serious leakage problems with gate-all-round FETs as the FinFET architecture won't go that small. The yield of gate-all-round FETs is far far far lower at this point than it was on FinFETs at this phase of development. As things get smaller, this is going to be a cost barrier.
We need to shift our thinking to parallelism, communication and offloading workloads to places where it's cheaper and more energy efficient to run them.
A10 was the largest chip ever in an iPhone at 125mm2. This year will be third year Apple is using 5nm to make their latest chip. The problem is Apple doesn't want to make such a large chip. A15 is already 108mm2 and if it receives the same transistor increase as previous years, A16 could be well over 130mm2.
Since 16nm, wafer and capital cost per wafer has gone up 4X. Anybody can increase performance by adding transistors, but it quickly becomes uneconomical to do so if die size also goes up.
Every one of those IPC gains and uarch improvements are the result of adding transistors. Why do you think A7 is 1 billion transistors while A14 is 11.8 billion transistors?
Those gains don't come out of thin air. It all comes from node improvements.
If litho is stalling, it means adding transistors adds to die size. The problem is 5nm wafers are extremely expensive.
Intel really didn’t delay new CPUs. They just released new ones that were about the same performance as the previous generation—sometimes slower and sometimes with weird limitations that made them not appropriate for Macs.
Apple is following TSMC’s schedule now but with their own designs that will always be appropriate for new Macs.
Well Samsung and Intel are planning on gate-all-around for their next nodes so I guess we will see if it works before TSMC’s 2 nm. Apple can absorb unbelievably large capital costs for the iPhone. I’m not worried about cost yet. Maybe next decade.
Since the debut of Apple Silicon Macs, the Mac has been a bigger revenue generator than the iPad. Also there is no way Apple is selling 25 million iPads in a quarter. That doesn't add up. If you look at Neil Cybart's analysis on Apple, iPad represents around 40 to 45 million units per year.
