Apple’s ‘Unleashed’ October: Revving up the Arm transition
Article By : Brian Dipert
What PC-tailored SoC will Apple release after last year's M1? Two of them, actually, initially in high-end laptops with more systems likely to follow.
Those of you who were impressed by the lithography foundation, transistor count and other characteristics of Apple’s M1 SoC, conceptually unveiled at the mid-2020 Apple Worldwide Developers Conference and rolled out in initial-system form that same November (the new Mac mini and 13” MacBook Air and MacBook Pro, followed by a 24” iMac the following April), are going to have your minds blown by the successor SoCs launched this past Monday. But I’m getting ahead of myself, because the first part of this week’s “Unleashed” event was all about audio:
October Tunes
I’m sorry, but at least in my opinion, “new bold and expressive colors“ aren’t going to be enough to turn around what’s to date been an underwhelming HomePod dabble by Apple in the smart audio device market now (and long) dominated by Amazon (Echo) and Google (Nest):
(Re)hiring a new division head might help (fresh thinking, don’cha know) but it’ll take time, and it also depends on whether or not (and if so, for how long) he has the ear of Tim Cook and others above him in the org chart.
I’m similarly skeptical that a lower-priced, Siri voice controlled-only (and otherwise defeatured) Apple Music tier will put much of a dent in the comparative market shares of Spotify, Amazon Music, Google’s YouTube Music,Pandora-plus-Sirius XM, Tidal (for the highbrow) and others, but perhaps it’ll give folks a more price-palatable option for fiscal consideration once their free trials time out…
I’m more enthused about the third-generation AirPods:
These neatly upgrade their second-generation (and still sold, complete with a permanent price cut) predecessors while still preserving feature set differentiation versus their AirPods Pro cousins—refurb examples of both I personally own. They now have adaptive EQ (to automatically fine-tune their spectral response to each listener’s unique insertion and ear canal acoustics) and spatial audio support, along with longer claimed battery life, but they dispense with the “Pro” flexible tips and active noise cancellation capabilities, along with newly-added “conversation boost“ selective-volume-increase support. Both AirPods models also now offer wireless charging cases that include magnets for MagSafe support. And speaking of MagSafe…
Anyone Want Some Chips?
Back in November of last year, when the details of the M1 SoC became public, I wrote:
It’s a monster of a 5 nm-fabricated chip, containing 16 billion transistors, up from 11.8 billion on the A14 SoC that powers Apple’s various iPhone 12 flavors.
What I didn’t mention, because I don’t think it was yet public, was the chip’s die size: 120mm2. And since then, we’ve all suspected that a higher-end variant was en route, but nobody knew what Apple would call it (not that it matters): M1X, following in the footsteps of architecture half-steps that Apple’s done with iPad SoCs? M2 maybe? Now we have our answer…there are actually two new chips: M1 Pro and M1 Max (respectively the upper and lower images that follow, in both cases alongside DRAM in a system-on-package approach):
With both SoCs, Apple has migrated from the M1’s quad-”big” (which Apple refers to as “high-performance”)/quad-”little” (“high-efficiency”) CPU core arrangement to an eight (on die: not necessarily all functional) high-performance/two high-efficiency processor core cluster. Clearly, peak compute speed is more critical this time around. I suspect that Apple has stayed with the same “Firestorm”-plus-”Icestorm” CPU core architecture combination found in the original M1 (along with the A14), versus migrating to the successor “Avalanche”-plus-”Blizzard” cores included in the A15 SoC announced last month. But only time (and benchmarks, and developer leaks) will tell, since Apple as-usual didn’t even talk about clock speeds this week.
Another key area of advancement with the M1 Pro and Max versus the base M1 (as well as versus each other, come to think of it) is with respect to the graphics and system DRAM controller subsystems. Whereas the original M1 included eight graphics cores, only seven of which were enabled in certain system configurations, the M1 Pro includes up to sixteen, with the M1 Max doubling that to up to 32.
And, in addition to migrating from LPDDR4X DRAM support in the M1 to LPDDR5, Apple has also widened the total system memory bus width, translating into both peak bandwidth and maximum capacity benefits: from 128 bits on the M1 to 256 bits on the M1 Pro and 512 bits on the M1 Max. All of these enhancements are evident in the M1 Pro (top) and M1 Max (bottom) die shot images that Apple shared, which also allude to their SoC design commonality (if you’re thinking that the lower third of the M1 Max is a mirror image of its middle third, as well as a mirror of the M1 Pro’s lower half, you win a prize):
Apple executives were also happy to publicly pound their chests with respect to transistor count. All three M1 variants are, as far as I know, fabricated on the exact same 5 nm TSMC process, but whereas (as previously mentioned) the “base” M1 comprises 16 million transistors, the M1 Pro more than doubles that to 33.7 billion…which the M1 Max nearly doubles again, to 57 billion. Wow. What Apple didn’t publicly share (yet, at least) were the new chips’ die sizes. However, they did show all three chips’ dice side-by-side:
AnandTech quickly did some calculations and came up with estimates (again versus the “base” M1 at 120mm2):
- 245mm2 for the M1 Pro, and
- 432 mm2 for the M1 Max
Again, I say…wow. Even though TSMC’s 5 nm process is likely more mature (therefore yielding higher) than it was a year ago, these are some big slivers of silicon. That’s why, for yield maximization reasons (along with product tier differentiation reasons), depending on which system configuration you pick, you’ll find either six or all eight “high-performance” cores available in the M1 Pro. Similarly, various M1 Pro tiers offer either 14 or 16 GPU cores, while the M1 Max comes in both 24- and 32-core graphics configurations. Speaking of systems…
Send In the Macs
What about the initial systems based on the M1 Pro and M1 Max? They’re 14” and 16” MacBook Pro laptops, successors to Intel x86-based predecessors, thankfully now with system memory allocation options up to 64 Gbytes:
Between the fact that those predecessors’ shipping lead times had lengthened notably in recent weeks and all the leaks, their announcement wasn’t much of a surprise. Interestingly and ironically, they’re slightly thicker and heavier than those Intel-based predecessors, which (along with the claimed more energy-stingy CPUs driving them) likely factor into the longer operating life in the form of beefier onboard battery packs.
Robust graphics processors translate, among other things, into the ability to simultaneously drive a robust number of external displays along with the integrated LCDs. Those LCDs have thinner-than-before bezels, translating into the necessity of a “notch” for the now-1080p built-in webcams (a notch which system software can selectively “hide” by creating a virtual bezel up top). Pixel pitch is unchanged between the two models, although display sizes (obviously) vary between them; this means that the 14” model has a resolution of 3024×1964 pixels, while the 16” model’s display is 3456 x 2234. HDMI (“only” v2.0) has returned as an external display-tether option, accompanied by multiple Thunderbolt 4 ports, and a SD card slot and the long-missed MagSafe power connector are also back. The rarely leveraged Touch Bar, conversely, is fini.
Unsurprisingly, all of Apple’s “pro” software packages are being upgraded to support the new hardware:
And long-in-development MacOS 12 “Monterey” is also scheduled to “go gold” next Monday…I guess I’ll spend this weekend upgrading my existing systems to MacOS 10.14 “Catalina,” since Apple only supports the three most recent versions of its operating systems with bug-and-vulnerability patches and other updates. Speaking of my existing systems…
One Year (Supposedly) to Go
When Apple announced its planned computer product line conversion to “Apple Silicon” in mid-2020, it forecasted that the migration would be complete by the end of 2022. Roughly 1.5 years in, here’s where we stand with Arm-based successors to Intel-based system precursors:
What’s left, unsurprisingly, are mostly the high-end systems in Apple’s stable, along with the 21” iMac which Apple still sells (at least as I write these words). The 27” iMac, for example, remains Intel-only at the moment; recall that Apple EOL’d the even higher-end (and also Intel-based) iPad Pro earlier this year. And then there’s the professional-tailored Mac Pro desktop. How will Apple service them?
Well, as I’ve already noted, the die sizes on the new M1 Pro and M1 Max SoCs are already ridiculously big, and next-gen semiconductor processes aren’t arriving (at least in the volumes that Apple needs) any time soon. So, my best guess is that somewhere between now and the end of next year Apple will unveil “M2” SoCs roughly comparable from CPU and GPU core count standpoints with today’s offerings, but with upgraded A15-derived microarchitectures, and with multi-SoC interconnect hooks that multiply the core counts and system memory allotments at the system level…something like the “chiplet” architecture that AMD’s using on its latest CPU generations. Doing so would enable the company to leverage a common sliver of silicon across multiple system product variations. That said, the company could alternatively develop pro-tailored chips: a CPU core-only main processor IC more conventionally mated to a GPU core-only graphics processor IC. We’ll see…
And what about other rumored upcoming systems, such as a higher-end Mac mini? First off, let’s explore why such a product might exist in the first place. Here’s the back panel of the Intel-based late 2014 Mac mini, currently one of the two in-use systems in my Mac stable (along with the early 2015 13” MacBook Pro I recently mentioned):
Look at all those ports! Now here’s what the 2018-era successor, also Intel-based, looks like:
Again, it’s connector-rich. Finally, here’s the back panel of the new M1-based Mac mini:
Ugh. I’ve tried several times over the past year to talk myself into buying a new M1 Mac mini, but the comparative dearth of external connections coupled with the 16 Gbyte-max system memory (and ridiculous markup over the mainstream 8 Gbyte alternative) have always given me pause. But I’m admittedly atypical; today’s M1-based Mac mini is adequate for the masses, and for high-end users there’s always the Mac Pro.
Here’s the thing: the Mac mini was originally intended to be a low-priced “gateway drug” to the Apple ecosystem for PC users that already owned keyboards, mice and displays. Over time, however, the Mac mini got beefier and more expensive, blurring the distinction between it and the Mac Pro; the “Mini Gets Mighty and Pricey” catchphrase in the title of a 2018-generation review by The Verge says it all. So it wouldn’t surprise me if Apple leaves the Apple Silicon Mac mini alone going forward, and continues to sell the Intel-based high-end variant until the Apple Silicon-based Mac Pro is ready to go.
And what about the M1-based 13” MacBook Pro launched last November, now that its 14” and 16” bigger brethren are here? I suspect it’s not long for this world: the differentiation between it and the M1-based 13” MacBook Air was always dubious at best:
I’m nearing 2,000 words, so it’s probably time to wrap up. Those are my thoughts on this week’s Apple news; sound off with yours in the comments, please!
This article was originally published on EDN.
Brian Dipert is Editor-in-Chief of the Embedded Vision Alliance, and a Senior Analyst at BDTI and Editor-in-Chief of InsideDSP, the company’s online newsletter.
