Apple Silicon chips expected to be refreshed on an 18 month cycle
Apple reportedly intends to release updates to its Mac and iPad Pro Apple Silicon chips every 18 months, but it isn't clear why this isn't happening at the same speed as Apple's A-series processors.
Apple is in the middle of its two-year schedule to move its Mac lineup away from Intel processors to Apple Silicon. So far, it has brought out three chips under its M1 masthead, including the M1 Pro and M1 Max in its newest MacBook Pro updates, but more are certainly on the way.
In a report from Taiwan's Commercial Times, supply chain sources claim Apple is aiming for an update to its Apple Silicon range every 18 months. This is in stark contrast to the A-series chips, which get a generational jump each fall as part of annual changes to the iPhone.
The report says the "M2" lineup will start off In the first half of 2022, with a chip codenamed "Staten." The "M2 Pro" and "M2 Max" versions will apparently arrive as part of an M2X architecture codenamed "Rhodes" in the first half of 2023.
M2 will allegedly be produced using a 4-nanometer process, which is also allegedly rumored to be used in the "A16" in the 2022 "iPhone 14."
The following "M3" series would arrive 18 months after the M2, it is reckoned, and will apparently employ TSMC's 3-nanometer process.
A leaked roadmap from November detailed the next generation of Apple Silicon processors as having the codenames "Ibiza," "Lobos," and "Palma." Thought to be separate from Rhodes, the chips are said to each contain two dies, produced using a 5-nanometer process.
While "Ibiza" is anticipated to be the low-power version to be used in a MacBook Air and iPad, "Lobos" and "Palma" are expected to be used in the MacBook Pro and other Mac desktops.
Read on AppleInsider
Apple is in the middle of its two-year schedule to move its Mac lineup away from Intel processors to Apple Silicon. So far, it has brought out three chips under its M1 masthead, including the M1 Pro and M1 Max in its newest MacBook Pro updates, but more are certainly on the way.
In a report from Taiwan's Commercial Times, supply chain sources claim Apple is aiming for an update to its Apple Silicon range every 18 months. This is in stark contrast to the A-series chips, which get a generational jump each fall as part of annual changes to the iPhone.
The report says the "M2" lineup will start off In the first half of 2022, with a chip codenamed "Staten." The "M2 Pro" and "M2 Max" versions will apparently arrive as part of an M2X architecture codenamed "Rhodes" in the first half of 2023.
M2 will allegedly be produced using a 4-nanometer process, which is also allegedly rumored to be used in the "A16" in the 2022 "iPhone 14."
The following "M3" series would arrive 18 months after the M2, it is reckoned, and will apparently employ TSMC's 3-nanometer process.
A leaked roadmap from November detailed the next generation of Apple Silicon processors as having the codenames "Ibiza," "Lobos," and "Palma." Thought to be separate from Rhodes, the chips are said to each contain two dies, produced using a 5-nanometer process.
While "Ibiza" is anticipated to be the low-power version to be used in a MacBook Air and iPad, "Lobos" and "Palma" are expected to be used in the MacBook Pro and other Mac desktops.
Read on AppleInsider
Comments
Farewell, 14nm: Intel Launches Alder Lake - ExtremeTech
(edit: Grammar)
For now, the M1 Pro/Max with 2-die will do, with M2 improving on power efficiency so you'll see less throttle. I still want to see a 2-die configuration on the 16" MacBook Pro, seeing much cooling headroom suggesting it.
See the Overview section to compare codename and fabrication dates for desktop and mobile chips.
https://en.wikipedia.org/wiki/Intel_Core#Overview
it’ll be major update every 2 years (M1 to M2…) with the Pro/Max versions coming every other year. So new processor every year, new architecture every 2 years.
18 months is odd as it puts architecture changes out of step with iPhone processors.
All this is irrelevant as real world performance is dictated by software optimisation. Apple needs thousands of engineers forking open source projects for dedicated macOS framework optimised software to provide alternative libraries to x86. At the moment the closed politics of open source is working against them as Intel shuts down non-x86 projects (see Embree-ARM) ensuring the foundations for Cinebench & Blender underperform on ASi.
I do feel like Apple is in a bit of a predicament at the moment as they have to design both new A series chips, but those chips are also going to be the basis for the M series chips. Chicken or the egg situation, tit for tat.
For Macs, 18 months is arguably a better timetable for all three elements. For Apple Silicon, it takes time to ramp up from the base M-series to Pro/Max to multiple dies. For macOS, the annual pressure seems like it has become too much — Monterey had important features announced that ended up delayed, and an 18-month macOS cycle would ease that. Finally, an 18-month refresh cycle works well for marketing. People would know what to expect. They could upgrade every 18 months, or after three years, after four and a half years, or after six years. They could get AppleCare for any of those intervals.
For those reasons they'll likely go with an annual or a biennial cycle (for their notebook categories), with macOS still occurring every year free of charge. macOS get annual updates that works across all their other OSes. This helps gets switches and retain users. It's synergetic and makes all their device categories better than their sum parts.
It is amazing how the discourse went from "we are pleasantly surprised that Apple's CPUs are competitive with Intel's!", which was actually true, and "Apple's CPUs are clearly better than Intel's!", which was never true, and to the degree it was, it was only due to Apple's decision to use unified memory instead of RAM and being on a 5nm process instead of a 14nm one. On the former, general purpose CPU makers using unified memory is very stupid because unified memory removes flexibility and upgradability. On the latter, once Intel's 7nm chips arrive in 2023, while the Apple power-per-watt advantage will remain, it will significantly decrease to the point where no one is going to talk about it anymore. For example, you are going to see 7 inch Nintendo Switch-type devices and 12 inch Windows 11 tablets running 14th gen Intel Core i5 CPUs that won't require discrete GPUs or fans that will have very good battery life.
Even Apple claimed that they were never going to be able to outdo Intel (or AMD) in single core or multicore performance and their big advantage was going to be power per watt. The problem is that unless you run a data center or are someone whose job requires them to be constantly "on-the-go" (and the people in the latter group switched to smartphones and tablets as their primary devices ages ago) then power per watt isn't going to be something that you care about that much. People aren't going to start valuing that metric overnight just because Apple says that they should, and the people who are going to all of a sudden after all these years start claiming that power per watt is the most important thing are going to be loyal Mac customers already.
Add it all up and no, Apple updating their Mac CPUs less frequently than Intel, AMD and Qualcomm updates their PC counterparts won't give Apple a competitive advantage. Impossible to spin that otherwise. It won't be a disadvantage either, mind you.
You don’t know what actual advantage the power consumption can bring. If I can do the same thing with half of the power, what would happen if I target the same envelope? Modern fabs will allow you to drop more than a dozen cores with ease, so whether you want to decimate your opponent is just an option.