The future of Apple Silicon Macs

[MandiMac]

I was just on Ars Technica where ikjadoon and their comment (5 hours ago) really stood out to me. I'll copy-paste it in here: What do you think?

1) iOS / iPad gaming making a day-one transition to macOS 11.0. Add in gamepad compatibility (i.e., macOS Big Sur ARM-native drivers) and Apple will have a potent, high-quality gaming library.

2) Apple going wild with system-wide integration, year after year: without Intel, Apple creates its SoC future: custom silicon budgets, custom die packaging, custom power gating, custom modem integrations, etc.

The list of integrations...looks endless.

• Apple could integrate a low-power GPS chip (a la LoJack) or a in-device AirTag (if those ever ship) with U1 hardware.
• Apple could integrate Package-on-Package unified VRAM + system RAM and/or experiment with alternative RAM / caching technologies.
• Apple could copy-paste the iPhone front-facing cameras & their signal processing each year onto that year's Macs and easily outclass even "high-end" Windows' integrated webcams, e.g., Microsoft's Surface line.
• Apple could offer far more PCIe lanes than Intel's restrictive 4x lanes in the thin-and-light space and create some wild Apple USB4 docks.
• Apple could, with on-die silicon, drastically decrease the cost barrier of entry to high-quality video editing. If you could record 4K streams from an integrated webcam, perhaps 4K editing might not remain a workstation-only path.
• As a first-party CPU vendor (and likewise its only customer), Apple now gets to influence the industry's codec choices, instruction & low-level code choices, RAM or PCIe standard transition timelines etc.
• Apple could create integrate its newly-bought modem IP from Intel to offer performant WiFi 6/6E on the SoC (modem integration on macOS before iOS!) without a secondary wireless chip. Maybe we'd finally see OFDMA in Wi-Fi 6's 2.4 GHz band (a joke. this will never happen). Or perhaps fully implementing WPA3's optional protections.
• Apple could dedicate more on-die GPU buffers for optimized, variable-refresh-rate 120 Hz laptop panels (i.e., panel self-refresh).
• Apple could add high-bandwidth interfaces to Apple Silicon to one day allow iOS or iPad OS devices to act as co-processors to macOS systems to accelerate computationally expensive tasks.
• Apple could use the increase in freed motherboard space to offer waterproofing and dust resistance (though more fanless devices feel more likely).
• Apple could offer a high-quality LTE modem with every single Mac device for anywhere, anytime connectivity with much lower chip-based power losses. Likewise, Apple could properly combine WiFi & LTE connectivity into a ultra-high-throughput internet connection.
• Apple could integrate more on-die, hardware-based CPU exploit prevention and protection, i.e., avoiding future Spectre-like vulnerabilities.
• Apple could add screen-sharing silicon to allow iOS & iPad OS to act as seamless secondary, low-latency screens for any MacOS device, i.e., beaming AirPlay device to device.
• Apple could use the increase in freed motherboard space and increased power budget to offer Project Soli / Leap Motion-like motion tracking sensors for low-cost education-focused laptops (with their own ML silicon to boot).

Of course, these are all fanciful ideas, but anyone who cannot see the massive vertical integration stack here is simply telling us they've found a nice spot to rest their head in the sand for the next ten years of computing.

Custom SoCs + modem IP -> phones -> tablets -> wearables -> TV boxes -> thin-and-light laptops -> desktop-replacement laptops -> SFF desktops -> all-in-one desktops -> desktop workstations.

In the end, the iPhone analogy seems closer and closer. When the iPhone released, it was easy to find criticisms uninterested to see past version one:

"No copy paste? DOA."

"Your fingers cover up the screen! How will you see anything?"

"Frankly, everything you can do on an iPhone, I can do a Palm Treo 650 already. With style. Nothing revolutionary here."

"Nobody is going to 'change the world' with a phone: the call quality isn't any better, you see."

 

[leman]

You know, as much as I love standards (especially open ones), there is also some merit in doing your own thing... provided you do it well. I also expect apple to go the full integration route. We will see some very new designs in the desktop space: unified memory, tightly packaged chips, radically decreased communication latency, all that kind of stuff.
By doing "their own thing" they can potentially deliver superior experience for both the user and the developer. One specific case: GPU and graphics programming. Apple GPUs are radically different from the traditional forward renderers from Nvidia and AMD and the programmers have basically console-level access to their internal architecture. This is just not something that one can do with an open standard like Vulkan, since the standard has to support the common hardware denominator. The question is whether the developers will bite, since the very obvious big downside is making cross-platform development more tedious.

One thing is sure: by going full custom Apple has dropped all the restraints that come with supporting third-party components. They don't have to adhere to common specs for electronic components. They can do whatever they want, up to developing their own RAM interfaces, using custom ISA in their ARM CPUs (which they are already doing btw. for ML acceleration) and doing other kind of crazy stuff. I am quite exited about this. While Apple's software quality was a bit wonky lately (especially on the desktop side), their custom chips are above anything else currently on the market.

 

[Boil]

All this custom silicon, reminds me of Silicon Graphics "back in the day". Makes me kinda excited to see what future arm64 Mac Pros might become. I could see a new Cube, a modern day SGI O2...

 

[Hexley]

I was just on Ars Technica where ikjadoon and their comment (5 hours ago) really stood out to me. I'll copy-paste it in here: What do you think?

1) iOS / iPad gaming making a day-one transition to macOS 11.0. Add in gamepad compatibility (i.e., macOS Big Sur ARM-native drivers) and Apple will have a potent, high-quality gaming library.

2) Apple going wild with system-wide integration, year after year: without Intel, Apple creates its SoC future: custom silicon budgets, custom die packaging, custom power gating, custom modem integrations, etc.

The list of integrations...looks endless.

• Apple could integrate a low-power GPS chip (a la LoJack) or a in-device AirTag (if those ever ship) with U1 hardware.
• Apple could integrate Package-on-Package unified VRAM + system RAM and/or experiment with alternative RAM / caching technologies.
• Apple could copy-paste the iPhone front-facing cameras & their signal processing each year onto that year's Macs and easily outclass even "high-end" Windows' integrated webcams, e.g., Microsoft's Surface line.
• Apple could offer far more PCIe lanes than Intel's restrictive 4x lanes in the thin-and-light space and create some wild Apple USB4 docks.
• Apple could, with on-die silicon, drastically decrease the cost barrier of entry to high-quality video editing. If you could record 4K streams from an integrated webcam, perhaps 4K editing might not remain a workstation-only path.
• As a first-party CPU vendor (and likewise its only customer), Apple now gets to influence the industry's codec choices, instruction & low-level code choices, RAM or PCIe standard transition timelines etc.
• Apple could create integrate its newly-bought modem IP from Intel to offer performant WiFi 6/6E on the SoC (modem integration on macOS before iOS!) without a secondary wireless chip. Maybe we'd finally see OFDMA in Wi-Fi 6's 2.4 GHz band (a joke. this will never happen). Or perhaps fully implementing WPA3's optional protections.
• Apple could dedicate more on-die GPU buffers for optimized, variable-refresh-rate 120 Hz laptop panels (i.e., panel self-refresh).
• Apple could add high-bandwidth interfaces to Apple Silicon to one day allow iOS or iPad OS devices to act as co-processors to macOS systems to accelerate computationally expensive tasks.
• Apple could use the increase in freed motherboard space to offer waterproofing and dust resistance (though more fanless devices feel more likely).
• Apple could offer a high-quality LTE modem with every single Mac device for anywhere, anytime connectivity with much lower chip-based power losses. Likewise, Apple could properly combine WiFi & LTE connectivity into a ultra-high-throughput internet connection.
• Apple could integrate more on-die, hardware-based CPU exploit prevention and protection, i.e., avoiding future Spectre-like vulnerabilities.
• Apple could add screen-sharing silicon to allow iOS & iPad OS to act as seamless secondary, low-latency screens for any MacOS device, i.e., beaming AirPlay device to device.
• Apple could use the increase in freed motherboard space and increased power budget to offer Project Soli / Leap Motion-like motion tracking sensors for low-cost education-focused laptops (with their own ML silicon to boot).

Of course, these are all fanciful ideas, but anyone who cannot see the massive vertical integration stack here is simply telling us they've found a nice spot to rest their head in the sand for the next ten years of computing.

Custom SoCs + modem IP -> phones -> tablets -> wearables -> TV boxes -> thin-and-light laptops -> desktop-replacement laptops -> SFF desktops -> all-in-one desktops -> desktop workstations.

In the end, the iPhone analogy seems closer and closer. When the iPhone released, it was easy to find criticisms uninterested to see past version one:

"No copy paste? DOA."

"Your fingers cover up the screen! How will you see anything?"

"Frankly, everything you can do on an iPhone, I can do a Palm Treo 650 already. With style. Nothing revolutionary here."

"Nobody is going to 'change the world' with a phone: the call quality isn't any better, you see."

TL;DR: Apple has the volume and margins to do whatever custom tech they want and make their products so distinctly superior that the Apple Tax is becomes justifiable to non-Apple users who have the ability to buy at Apple price points.