Apple Silicon Roadmap Based on 18-Month Upgrade Cycle, Claims Report

[Bug-Creator]

Someone's math is off here....

If the M2 launches "in the 2nd half of 2022" it is more like 2 years (which makes sense since skipping odd numbered Axx was what Apple was doing for the iPadPro chip for quite some time).

M2ProMaxGiga in 1st half 2023 would be 18 months, but that would put the bigger M3s launching at the same time as the A18 in late 2024.

One has to ask wether launching the M1ProMax this fall was what was planned from the start or if there was some meat to the "M1x" rumours in late 2020/early 2021 (for non redesigned MBP and the 2fan 23.5" iMac).

My guess:
- New baseline Mx chip 1 month after every even numbered A chip (MacBook(Air) base MacMini)
- more Mx HW and Pro/Max variants sprinkled through the next year

 

[Red Oak]

Given the design cost of a 5nm chip is $500M (minimum), this news should be no surprise. At 4nm, it's at least a billion dollars. People who thought M-series processors would follow the same annual cadence as iPhones are simply out to lunch.

Where are you getting those estimates?

Apple already has the engineers. TSMC is the one doing the capital investment. Where is the additional cost for Apple coming from?

 

[huge_apple_fangirl]

Makes sense. A*X line was around 18 months, so I would expect the M line to be similar. If they update Macs every year and a half with significant performance upgrades that would be a huge benefit over Intel. Especially the Mac Pro.

 

[LavenderCaptain]

18 months look a bit dubious to me because that would mean that in a three-year cycle they would alternate between launching new M chips coinciding with the new iPhones and then in the off-season the next time.

That would cause a production bottleneck every 3 years. Why would they do something like that?

This looks like garbled information. It would seem more plausible to me if they ran a 2-year upgrade cycle, offset from the iPhone SoC upgrade cycle by 6 Months or something like that, but still keeping the product lines synchronized that way.

I really hope I'm wrong, but it feels like we're going back to the PowerPC days... A excellent and strong start (the M1x and friends), but what's the future and what lessons can we learn from the past? At some point in time, the reality distortion field might not be able to convince people that their machines are faster/better than the competition anymore, especially if they stay still too long... After all, it's not like Intel and AMD are sitting still, and the reasons why the M1x are faster (performance "enhancements" like putting RAM and media processing ASICs) are well understood by everyone...

 

[Whathappened]

Nobody upgrades just because something new is out (unless they are dumb oc)

Your very, very intelligent assumption shows you obviously don’t know about different people with different needs.

 

[Shirasaki]

That can be good news since the M2 can be armv9...
If the M2 would had arrived in April of next year, it would be impossible to be armv9

Not necessarily. ARMv9 may not be relevant when the chip was being designed. M3 for ARMv9 could be possible. Besides, I would not be surprised to see Apple going RISC-V really soon after ARM transition is over.

 

[CWallace]

So now we have a claim that there will be an M2 PRO and MAX ("Rhodes"). William Ma's report stated their would be an "M2 Duo", but did not have a code name. He did say there would be a complete family of M3 SoCs and provided the codenames (M3 - Ibiza | M3 PRO - Lobos | M3 MAX - Palma).

As to the "18 months for M2, yet 24 months for M2 Macs" issue, it could just be due to supply-chain issues pushing back the model launch. M1 PRO and M1 MAX were ready at WWDC, but Apple didn't have the parts to ship complete MacBook Pro 14/16 until October so it launched then.

 

[phos07]

thats interesting,
I thought they would alternate between a base M chip and pro/max chips yearly, so we would get something every year with newer generations every two years

 

[boss.king]

So new MacBook Air in spring and M2 in de second half of the year. How does that match up with the rumors that the new MacBook Air have M2 processors?

Why wouldn't the M2 be in the new Air? 18 months from initial release would land on May 2022. I don't know why Apple would release a second M1 air if they have a new chip right around the corner.

 

[boss.king]

People maxing our their MacBooks aren’t going to be annual upgraders.

Or if they are, it's likely a business expense and they'd be happy to pay for every inch of increased performance they can get.

 

[_Spinn_]

This makes sense. Most people don't upgrade their computers every year.

 

[siddavis]

I'm waiting for the M2 Pro Max.
Monterey is so buggy and memory leaks and missing features.
Is this gonna be the new Normal????

And in Mac mini form.
No cheesy breakable keyboard on the MacBook Pro for me or Battery or screen issues.

Buying A Mac mini you escape all these hassles

and if they force you to only install APPS from the APPLE APP store like the iPads and iPhones thats when I jump ship and move back to Windows 10 or 11.

Ugh, I hate when I come across comments like this. I'm still on Mojave, but getting ready to update in anticipation of getting the iMac Pro when it is released. I want to get up to speed with the latest OS (and get the wife used to it) before buying the new iMac. However, I am afraid the bugs will drive me nuts when I hear others still having issues at the .1 release.

 

[evertjr]

I really hope I'm wrong, but it feels like we're going back to the PowerPC days... A excellent and strong start (the M1x and friends), but what's the future and what lessons can we learn from the past? At some point in time, the reality distortion field might not be able to convince people that their machines are faster/better than the competition anymore, especially if they stay still too long... After all, it's not like Intel and AMD are sitting still, and the reasons why the M1x are faster (performance "enhancements" like putting RAM and media processing ASICs) are well understood by everyone...

lol you know M1 is based on A series chips that are basically competing with itself every year since everyone else is so far behind. Apple now have full control of their lineup, they will not "stay still too long" for no reason.

 

[IamTimCook]

Exactly. Everybody updates at a different cycle. Updating Macs every 18 months means that everybody can get a state of the art machine every time instead of having to wait a long time for the next update to release.

Really good point. I remember wanting to upgrade an Intel MBP that was already 4 years old. My options were the next version but it was already a few years old, itself.
I kept asking myself, “do I wait for the next model? Is it coming any time soon?”

This new update cycle will give user who stayed with there machine 4 to 5 years a definite idea of when they can purchase again.

 

[GuilleA]

18 month for a new generation of chip huh. Not quite yearly as crazy as A-series chip but still seems to be quite aggressive. Upgrading Mac every 18 month seems a bit much, especially for those going for $6000 maxed out options.

With that being said though, how much of that performance gain could be for each generation?

I you spent $6k on your Mac, you most definitely paid it off within a year of work (maybe even less), so upgrading every two years seems like a reasonable investment (pay off the initial purchase the first year, save up for the upgrade the following year, wait 6 months for fixes for any potential manufacturing defects, purchase new model).

 

[ElectricPotato]

Don't read too much into a proposed iteration tempo. It is not an immutable schedule or mandate of change purely for the sake of change.

 

[rickster2kuk]

That wouldn't be 18 months though, would it...?

Yes agreed sorry my mistake.

Perhaps they will have minor bump in summer next year using the M1 Pro, not really much they can add apart from FaceID and updated HDMI and SD card slot.

Or perhaps they will want until the following year for M2.

 

[anthonymoody]

Wait for the outrage when the AS Mac Pro ships with an M1 branded chip and M2 is released 3 months later or at the same time.

No one expects the M2 to be faster than an M1 Pro/Max, much less an MCM M1 Max. The customers buying Mac Pro don’t rely on marketing names to determine actual performance.

I don't think anyone thinks that a plain vanilla M2 will be faster than an M1 Pro/Max (though it could be for single core measurements). I think the potential for outrage lies in the fact that the basic, underlying building block in the "top of the line" Mac, whether Mac Pro or iMac Pro, would be supplanted so soon after release.

I understand that the Pro line of desktops aren't the volume machines, so I get why Apple released things in the order they have. But it still makes the order of release feel backwards because of this particular aspect.

 

[jdb8167]

Why do people fixate on marketing names? All that matters is if the next gen for each category is faster with more capabilities. M2 Pro vs. M3 Pro? Apple already knows what the plan is and a naming convention isn’t going to change their progress. 18 months per generation for each category seems like it is designed to match TSMC node changes.

 

[ajfahey]

The M1 was innovative. It was architecture innovation compounded by ARM and semiconductor manufacturing process evolution. The M1 Pro and Max are incremental extensions of the M1 architecture.

 

[mabhatter]

18 month for a new generation of chip huh. Not quite yearly as crazy as A-series chip but still seems to be quite aggressive. Upgrading Mac every 18 month seems a bit much, especially for those going for $6000 maxed out options.

With that being said though, how much of that performance gain could be for each generation?

Upgrading a Mac every 3 years (36 months) would be ably normal though. That's the typical AppleCare span.

I would have figured an even 24 months. Then it's a "tic-tock" of consumer machines and pro machines every other year.

 

[constructor]

I really hope I'm wrong, but it feels like we're going back to the PowerPC days... A excellent and strong start (the M1x and friends), but what's the future and what lessons can we learn from the past? At some point in time, the reality distortion field might not be able to convince people that their machines are faster/better than the competition anymore, especially if they stay still too long...

WTF? There is exactly zero evidence for a development like that!

The PowerPC stagnation was due to Apple being a relatively insignificant customer even for their CPU manufacturers and the whole platform not justifying major investments from their points of view (plus a chunk of incompetence on top).

None of that applies any more and Apple is actually by very far the most consistently progressing processor manufacturer worldwide now, with practically bottomless pockets to finance whatever they want from their wildly successful product lines and with the developmental prowess to make it happen, too.

After all, it's not like Intel and AMD are sitting still, and the reasons why the M1x are faster (performance "enhancements" like putting RAM and media processing ASICs) are well understood by everyone...

Yes, by comparison AMD and Intel are way back in Apple's rear view mirror and both are hampered by their inability to get away from the decrepit and increasingly limiting x86 platform. They have no real chance for overcoming that performance and efficiency gap because it's inherently baked into their very platform.

 

[Realityck]

According to sources in the supply chain industry, Apple Silicon will be updated every 18 months in the future. In the second half of 2022, Apple will first launch the M2 processor code-named Staten, and in the first half of 2023, it will launch the new M2X processor architecture code-named Rhodes, and release two processors such as M2 Pro and M2 Max according to the different graphics cores. Apple's M2 series processors all use the 4-nanometer process, and will be updated to the M3 series processors after an 18-month cycle. It is expected that they will be mass-produced using TSMC's 3-nanometer process.

One can’t help noticing that the way this is rumored that it could easily be by 2 years rather then 18 months. For example the M1 arrived Nov 2020, and they guess the M2 will arrive second half of 2022. But M2X going by the same logic won’t happen first half of 2023. They should have guessed it would occur second half of 2023.

 

[constructor]

Besides, I would not be surprised to see Apple going RISC-V really soon after ARM transition is over.

That wouldn't make the slightest sense.

For Apple there is nothing of value in RISC-V that could even remotely justify the massive expense of yet another binary-incompatible transition.

Advancing from the decrepit and metastasis-ridden x86 platform to the freshly designed and legacy-free ARM64 platform comes with major performance and efficiency gains, but RISC-V has no such advantage over ARM64 and it isn't even a really viable platform at all yet. It would make no sense at all to move to it.

 

[CarlJ]

I was hoping for every 24 months as I want stable machines and stable macOS. I hate bugs and if they have 24 months to fix everything (almost) behind the scenes I'm happy.

I believe they're talking about only the Apple Silicon SoC's here, not the Macs themselves.

I suspect the Macs will still stay on a 12 month cycle, simply because buying goes on a 12 month cycle (that is, everyone expects new Macs - and everything else - in the fall, ahead of the holiday season - not suggesting that anyone should be buying new Macs every 12 months). There are plenty of tweaks they can make during that 18 months, starting with simply increasing the clock rates, or shifting focus to the chips with extra GPU cores that were previously the high-end build-to-order model, as yields improve.

For the record, I too, want stable Macs and stable macOS. I don't recall hearing of any bugs in the M-series chips themselves.