The argument wouldn't be that they needed to change the Max to acommodate the Ultra fusion. Rather, it would be that they had to wait until the Max design was sufficiently far along before they could finalize the fusion component. Of course, I suppose it's also possible that the Max and fusion were codeveloped, such that the 2x Max was finalized at the same time the Max was. But that's not the picture I have. Wonder if Apple will eventually talk about this.
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.
Could we see M3 before A17?
- Thread starter senttoschool
- Start date
- Sort by reaction score
According to https://www.semianalysis.com/p/tsmcs-3nm-conundrum-does-it-even , TSMC's N5 capacity in 2022 was 120,000 wafers/month. Based on this discusion, evem 45,000 wafers/month should have been more than enough capacity for all the iPhones and Macs produced in a year: https://forums.macrumors.com/thread...cs.2381637/page-7?post=31996254#post-31996254 Thus limitations in TSMC fab capacity wouldn't seem to explain the delayed release of the larger chips.
And for N3 in 2023, isn't TSMC over-capacity because of order cancellations by Intel and others?
I'm now so over the N3 anyways. Let's talk about the 2nm A18 Bionic. 😉
www.smartprix.com
Apple working on 2nm A18 Bionic Chipset: Expected to power iPhone 17 Ultra in 2025 - Smartprix
Tech giant Apple is all set to undergo major changes in terms of its chipset manufacturing unit. The company will slowly start replacing its current chip testing machines at the TSMC foundry so that they get ready for the 2nm production node in time for the iPhone 17 series in 2025. Some reports...
www.smartprix.com
I'm now so over the N3 anyways. Let's talk about the 2nm A18 Bionic. 😉
Apple working on 2nm A18 Bionic Chipset: Expected to power iPhone 17 Ultra in 2025 - Smartprix
Tech giant Apple is all set to undergo major changes in terms of its chipset manufacturing unit. The company will slowly start replacing its current chip testing machines at the TSMC foundry so that they get ready for the 2nm production node in time for the iPhone 17 series in 2025. Some reports...
The first A-series chip to use N2 would likely be the A20 in 2026.
My post wasn't meant seriously. That might have gotten lost in translation, since the "I'm so over the..." phrasing is what a teenage girl in the US would use. So the humor was supposed to be that, as a result of an overly-optimistic prediction that A18 will be on 2 nm, I was being fickle and suddently losing interest in N3. That's why I added the smiley face.
Last edited:
5nm production has risen from around 65-80k wafers/month (different stories there disagree on what the initial production capacity was) to over 140k per month as of the most recent report. With 3nm only at 45k wafers/month at present, that feeds into the notion that initial availability is somewhat constrained at this time
There do appear to be some overlaps (like, parts of the RT solution patented by Apple has some similarity to IMG patents), but Apple's GPU cores are different enough from IMG that they can't just use their solution directly.
Haha, fair enough
Regarding @Mago's leaks, I don't know, some of it sounds reasonable, but most of it just sounds too outlandish to me. So I choose to be sceptical. Note: Apple does have patents for multi-chip connections that describe bridges going over multiple dies. But that kind of bridge certainly won't be using PCIe...
I still want the first generation ASi Mac Pro to debut with M3 Ultra & M3 Extreme SoC options, on a daughtercard would be best for (limited) future upgradability...?
I also would like to see ASi (GP)GPU solutions rather than any third-party options, it seems like Apple would be reversing course after telling devs to optimize for pure ASi GPU cores for the past two+ years...
I expect we will all find out more come WWDC at the latest...?
All jokes are better when explained. Hence the McBain, “That’s the joke” vibe.
Great, now you’ve got me doing it.
Apple has a bunch of recently published patents (march 2023) that deal with work distribution across multiple GPU devices/clusters. The theme is that communication between clusters is slower than communication within clusters. This is somewhat similar to how Apple CPUs are organized.
Now, it’s not clear how this translates to concrete products. Maybe they are talking about a system with multiple physically distinct GPU boards, or maybe it’s about better utilization of multi-chip systems (Ultra etc.). It is also possible that the upcoming Apple GPUs get a completely new architecture and these patents are about that. Anyway, it is clear that Apple has been working on GPU scalability, so it’s reasonable to expect advances in this area.
Regarding Extreme: another very recent Apple patent describes a concrete multi-chip device architecture that uses a new on-chip network design and interrupt schema. It only mentions two chips.
- The M1 arrived just a month or so after the A14, they shared the same core generation and process node.
- The M2 was based off the A15 and they were both tapped out on the same process node. It was released about 9 months after he A15, 3 months before the A16.
- If the M3 is in fact being produced alongside the A17 on TSMC's latest node, then they will also have the same generation cores and we shouldn't really expect any device based off of it until Fall, late Summer at the earliest. As a reference...TSMC started producing in volume on their N5 node in December 2019. The iPhone 12 (A14) was released in Oct. '20 and the first ASi Macs (M1) in Nov. '20. However, one caveat could be initial demand. The A17 is only going in the iPhone Pro models, which more than likely would not have the same volume of sales. And the first M3-based Mac could be a low-volume system such as the Mac Pro or Mac Studio. So Apple would not need a massive stockpile before release, which could mean there's a possibility of an M3 (based on A17) Mac at WWDC.
- But, if the M3 is based on the A16 using the N4P node, then we could very well see those Macs much sooner.
Last edited:
I think history will not be kind to the M2 chip. The M2 Pro and M2 Max chips literally just came out however the M2 chip in the iPad Pro and MacBook Air left such a underwhelming taste in people's mouths that we are all collectively ready to put the M2 family of chips behind us and move on to M3.
That's also what I have been thinking. It would make sense to use the initial N3 production for a flagship product like the Mac Pro.
Especially since Apple needs to beef up the GPU performance for the ASi Mac Pro, 3nm should give a larger transistor budget, allowing more GPU cores...?
That's been my thinking as well—that the MP's low volume enables them to use components that might be in overly short supply for their other products. And this applies not only to N3 chips, but to other components as well, like RAM. Here's something I posted in Oct. 2022:
Let's suppose Apple does release an M3 Air on N3 sometime between now and WWDC. Rumors are that the 14"/16" MBP's won't get a chip update for about a year from when the M2's came out (in Jan). Wouldn't it ber a problem to have their flagship laptops linger so long without the newest tech? Might Apple thus do a mid-year refresh?
Apple may be able to compensate for this in an M2 Mac Pro by significantly increasing the CPU and GPU clocks (thus exceeding base M3 per-core performance). But for the M2 MBP's, the horse has left the barn.
Apple may be able to compensate for this in an M2 Mac Pro by significantly increasing the CPU and GPU clocks (thus exceeding base M3 per-core performance). But for the M2 MBP's, the horse has left the barn.
Last edited:
Could an M3 MBA be faster than an M2 Max MBP? What benchmark scores could the M3 get?
Could the M3 be faster than the M3 Pro/Max in 3D rendering if it includes hardware-accelerated ray tracing?
Note that I was referring to single-core CPU performance, so an M3 Air will certainly be faster by that metric than an M2 Max MBP, just as the M2 Air has higher SC performance than the M1 Max MBP.
I've no idea how much faster a 3 nm M3 would be. Based on https://www.anandtech.com/show/1735...n3e-in-2024-n2-in-2026-major-changes-incoming , the process improvement from N4P to N3B should give maybe 10% at equal power, but then there's also the improvement from the newer M3 architectiture, and I have no basis for estimating that. Also depends on how they clock it.
Afraid I have idea about your GPU question.
Last edited: