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
Semantics.... the M1 MBA is an old model. The M2 and M3 are baseline notebook CPUs, the entire MBA line is Apple's "value" notebook line (in the context of Apple hardware).
Point being, 3nm manufacturing is both expensive and in short supply (so much so that Apple has booked out TSMC's entire capacity) - Apple aren't going to waste capacity on larger parts for some time (as smaller parts yield much better, particularly early on in a new process whilst it has a higher defect rate per square mm), especially not for a value product. It will go into mobile chips first as the smaller size gets better yield, and that's the segment the improved power efficiency is critical for anyhow.
I would not expect M3 in anything until at least very late 2023 through mid 2024 - and most certainly NOT before the iPhone gets 3nm parts.
Last edited:
You can also see it in the financials, which Apple does us no favors with meaningful detail. Sales are down yet profit % is historically very high.
Apple's vertical integration has been very good for profits.
5 out of 7 announcements for the Mac Pro were at WWDC. So, better than even odds that the next AS Mac Pro will debut at WWDC this year.
When Apple release the iPhone14 in 2022, only the Pro Max versions received the upgraded CPU (A16). The base iPhone14 uses the previous generation (A15). As part of their regulatory obligations to shareholders, TSMC has to provide updates regarding the capital expenditures and ROI. Digging into that information reveals important dates for when process nodes transition from development to volume production. These dates and the yield ramp are important to forecasting availability of supply.
Historically Apple has shown they prefer to update the hardware with the latest chips. They have historically been able to do so because their chip designs were manufactured using mature processes. When Apple acquired an interest to be an early adopter of new process nodes the date the process transitions to volume production and yield ramp now play into their product availability.
Thus a change in product strategy is mandated by the early adoption of the process nodes. The "new" product strategy enables Apple to release the product line at the earliest time. Otherwise, Apple would have to delay the release and/or significantly constrain supply.
Because of the availability of chip supply. Pro Max Ultra versions are used in lower volume high margin products. So, Apple can maintain a yearly update cadence and meet product demand by staggering product updates this way. This helps Apple maintain a performance edge over their competition.
If Apple where to put an Mn Pro/Max/Ultra/Extreme in a 13" Macbook Pro or Air - you'd have a point. But they don't do that. It's more complex. It does demonstrate Apple giving first preference to pro users.
Maybe they'll tweak their cadence and give the first 3nm die shrinks to the M2(x?) Ultra/Extreme...
So slightly less than 50% is also “WWDC is not for hardware announcements”? Really dude? And you had to make up an artificial argument too? You must really crave those internet points.
Two factors work against that. First, the initial runs of a new process always have lower yields of usable chips. Given that current production of 3nm wafers is already around 1/4 that of 5nm, there is a compounding effect regarding the number of SoCs available to put into machines. Second, that number is further reduced if they start with an Ultra SoC because its larger die size means less chips per wafer. To me, it would make more sense to start with a base M3 and stockpile the parts if necessary.
Apple has been booking the initial production runs of new TSMC processes for a while now, so the cost and supply factors aren't relevant to that practice. As these processes continue to get smaller, costs will continue to go up while yields will likely go down (at least in the initial production runs). I do think that it will be hard for Apple to procure enough 3nm parts for the iPhone 15 Pro/Pro Max without running into supply constraints like we saw with the 14 Pro and Pro Max last fall. Another option would be for Apple to drop those 3nm parts into the iPad Pro, especially if the rumors of a 14" model are true.
That's less than 50%. If you really argue, it's actually 3 hardware announcements since other 2 are just updates from previous generations. Now, does it sound like slightly less than 50%? Since WWDC WAS about hardware announcement until 2012, only 50% is already proves it's not for hardware announcements.
It's still expensive in terms of quantity of chips available out of it. If there are only X square MM of capacity at yield of y% then apple can either make a heap of mobile SOCs or a much smaller number of M3 and up.
The MacBook Air market is not cpu/gpu performance sensitive. The mobile and AR/VR market very much is. This is where 3nm will go first.
The only non-phone product I expect to see 3nm in this year is MAYBE the VR/AR headset, if we see it.
Too simplistic a view. That pertains to the definition:
- In a manner that simplifies a concept or issue so that its nuance and complexity are lost or important details are overlooked.
There has been lots of guess in the dark about this. Maybe WWDC will show up finally what Apple has kept completely out of sight?
Last edited:
So far, Mac Pro is rumored to use M2 Extreme or 4 chips of M2 Max so it's not really new. Cant really expect new features as M3 is expected to get ray tracing chips and other new chips. M2 series still disappointed a lot in terms of GPU performance as Nvidia out perform them even with RTX 30 series. Better to use M3 series.
This depends on the problem. There are benchmarks that show AS is faster than T4.
GitHub - bhattbhavesh91/apple-gpu-vs-nvidia-gpu-benchmarking: In this repository, we will do a benchmarking analysis by training a Tensorflow Deep Learning model on M2 MacBook Air and compare the training time with NVIDIA's Tesla T4 GPU on Google Col
In this repository, we will do a benchmarking analysis by training a Tensorflow Deep Learning model on M2 MacBook Air and compare the training time with NVIDIA's Tesla T4 GPU on Google Colab. -...
github.com
WWDC hardware announcements-
- 2011: MacBook Air, Thunderbolt Display
- 2012: MacBook Pro with Retina display
- 2013: Mac Pro, MacBook Air
- 2014: iMac with Retina display, MacBook Air
- 2015: 12" MacBook
- 2016: MacBook Pro with Touch Bar
- 2017: iMac Pro, HomePod, iPad Pro,
- 2018: MacBook Pro with Touch Bar
- 2019: Mac Pro, Pro Display XDR
- 2020: Macs with Apple Silicon
- 2021: AirPods Max, HomePod mini
- 2022: M2 MacBook Air and 13-inch MacBook Pro
- 2023: ?
Last edited:
That would be the Mac mini DTK (Developer Transition Kit); with an A12Z SoC, 16GB RAM, & 512GB SSD; loaded with a beta variant of macOS 11, aka "Big Sur"...
For some reason this never made it to the MR front page. No comments section either. 
www.macrumors.com
MacBook Pro: Just Updated! Features, Buying Advice, and More
The current 14- and 16-inch MacBook Pro models feature the M4, M4 Pro and M4 Max chips, Thunderbolt 5 on higher-end models, upgraded cameras, and...
www.macrumors.com
it really doesn’t have much to do with Tim Cook . There are lots of invented “but Apple blah blah blah “ reasons that get bandied around in these Mac/ apple focused forums . A more relevant drover though has been the Computek Conference that typically happpens a week earlier . That conference focuses on new tec( that drives the mainstream Windows PC market forward . So Intel and AMD and Nvidia typically line things up to reveal there .
So if more often than not, new Intel mobile processors came out and if appple had finished new laptop design to fit them …. then WWDC will happen to be right time to release.
I think your ‘ only 5‘ is about conspicuous , ‘dog and pony show’ introductions; not actual sysytem release tracking . iPhone gets its annual September event and there seems to be looks of over hype trying push other apple systrpems 8th WWDC so there can get some status in the Apple product line up be being at WWDC ( another fixed in stone in the calendar dog and pony show )
The Mac Pro 2012 ‘announcement‘ is a bit of over counting . Apple tried to spin that a “brand new “ big deal when it was mere a sped bump on processors that Intel had been shipping for a while . Meanwhile at same spline Dell / HP / Levovo were introducing next generation Xeon E5 v1 systems . Apple didn’t have anything for that whole generation . That was far closer an exercise is misdirection than it was a announce new system . In fact, backlash ( complaints with FTC , etc ) was so strong Apple took the ‘new’ tag off the system on the website because it really wasn’t new ( technical model number had not changed at all ) .
If start to throw in dates where MP 2019 got GPU updates the x out of y dates for mac Pro are going align even less with WWDC .
And announce is a bit of stretch when these asneak peaks that full configurartian details ( not all options listed ) , technical specs , and full prices for options .
Interesting bit of discussion around TSMC, Apple and 3nm at the third bookmark in this Broken Silicon podcast:
(Under "Is TSMC 3nm delaying Nvidia Blackwell")
(Under "Is TSMC 3nm delaying Nvidia Blackwell")
Last edited:
I wouldn't call Maxtech experts on the industry side of things. They just run tests (sometimes intentionally crafted to create a desired result) and compare one machine to another.
Agreed except this is not a delay but consequence of Apple strategy. Base M3 will come after M3 Pro Max Ultra. M3 on track for WWDC launch with Mac Pro.