Alleged Apple Chip Plans Suggest 'A16' Will Stick With 5nm, 'M2' to Make Jump to 3nm Instead

[senttoschool]

Alleged Apple Chip Plans Suggest 'A16' Will Stick With 5nm, 'M2' to Make Jump to 3nm Instead

The "A16" chip for the iPhone will reportedly be manufactured with the same process as the iPhone 13's A15 Bionic, with Apple saving a...

www.macrumors.com

The leaker is "ShrimpApplePro" on Twitter.

First of all, 4nm is just an enhanced 5nm process. It's part of the 5nm family. So calling it 5nm enhanced or 4nm doesn't matter in the grand scheme of things. We've known for a long time now that A16 won't be on 3nm because 3nm is delayed.

This brings it to the second rumor which is that M2 will be on 3nm. Yea, I'm not buying this. It makes no sense. Why? Because all rumors point to second half launch for an M2 Macbook Air. 3nm isn't scheduled to go into high volume production until 2023. How can we get mass M2 devices when 3nm isn't ready until 2023?

Lastly, when asked about ray tracing support in the new SoCs, this leaker replied saying that he thought ray tracing was an Nvidia exclusive feature. This shows that the leaker knows nothing about chips.

https://twitter.com/i/web/status/1530109888633765888

I'm surprised Macrumors would even report on this.

 

[JPack]

Aligns with the credible rumour from Kuo the MacBook Air refresh this year is sticking with M1.

N3 is ramping later this year and TSMC expects to record revenue in early 2023, which means deliveries at that time.

Remember when some people dismissed Nikkei thinking it was silly? N3-based M2 is looking correct.

Apple and Intel become first to adopt TSMC's latest chip tech

Taiwanese company remains vital partner for American tech giants

asia.nikkei.com

 

[thenewperson]

Aligns with the credible rumour from Kuo the MacBook Air refresh this year is sticking with M1.

Huh? How does it align? It seems to align with the Nikkei report though.

First of all, 4nm is just an enhanced 5nm process. It's part of the 5nm family. So calling it 5nm enhanced or 4nm doesn't matter in the grand scheme of things. We've known for a long time now that A16 won't be on 3nm because 3nm is delayed.

That person specifically says it's not 4nm.

 

[JPack]

Huh? How does it align? It seems to align with the Nikkei report though.

It fully aligns. Some dismissed Kuo's prediction of M1 being used in 2022 MBA. If M2 uses N3 litho, which doesn't ship until early 2023, then it absolutely makes sense.

If A16 uses N5, it will be a big, big chip. So it makes even more sense to use A15 for mainstream iPhone 14 models.

 

[UBS28]

So no new Mac’s besides the Mac Pro then as 3nm is planned in 2023. Makes sense giving how Apple doesn’t have capacity to produce chips anymore for this year.

 

[BigPotatoLobbyist]

Alleged Apple Chip Plans Suggest 'A16' Will Stick With 5nm, 'M2' to Make Jump to 3nm Instead

The "A16" chip for the iPhone will reportedly be manufactured with the same process as the iPhone 13's A15 Bionic, with Apple saving a...

www.macrumors.com

The leaker is "ShrimpApplePro" on Twitter.

First of all, 4nm is just an enhanced 5nm process. It's part of the 5nm family. So calling it 5nm enhanced or 4nm doesn't matter in the grand scheme of things. We've known for a long time now that A16 won't be on 3nm because 3nm is delayed.

This brings it to the second rumor which is that M2 will be on 3nm. Yea, I'm not buying this. It makes no sense. Why? Because all rumors point to second half launch for an M2 Macbook Air. 3nm isn't scheduled to go into high volume production until 2023. How can we get mass M2 devices when 3nm isn't ready until 2023?

Lastly, when asked about ray tracing support in the new SoCs, this leaker replied saying that he thought ray tracing was an Nvidia exclusive feature. This shows that the leaker knows nothing about chips.

https://twitter.com/i/web/status/1530109888633765888

I'm surprised Macrumors would even report on this.

On a tangent unrelated to this comment of yours: I'm glad you've come around to the low likelihood of Apple adopting an annual cadence for laptop SoC refreshes.

But yes, this rumor is absurd other than the A16 staying on the N5 family - the A15 was N5P but we may see N4P with the A16 for example, but either way the jump there is isn't particularly notable. The majority of any performance gains we see will have to come from architectural improvements or more cache, at least on the SoC itself. I am hoping to see a real 5-10% ballpark IPC gain on the big cores, even 5% with some area hit would be great. Apple have the highest Performance Per Clock out there right now so that hypothetical 5-10%, provided it's over general workloads, would be no joke. Also hoping to see SVE2 and Arm V9 - though that seems almost certain.

I will add that we could see some A14 -> A15-esque efficiency improvements from the adoption of LPDDR5x, which is a low-hanging fruit of sorts for Apple's A16. Very underrated in tech commentary on the A16, though I don't think it will influence the GPU as much as people would like to believe at least with Apple's existing SLC cache and with mobile gaming loads. But efficiency? Easy win.

 

[Bodhitree]

Anything could happen with the capacity shortage in chip manufacturing.

 

[darngooddesign]

Aligns with the credible rumour from Kuo the MacBook Air refresh this year is sticking with M1.

Can't the Airs get a faster M1, not M1-Pro/Max but a faster version of the regular M1, this year without moving the 3nm?

 

[JPack]

Can't the Airs get a faster M1, not M1-Pro/Max but a faster version of the regular M1, this year without moving the 3nm?

Clocked higher and maybe all 8 GPU cores enabled. Either way, it won’t be substantial. To get performance increases, you need more transistors. Without a shrink, you can’t add more transistors without making the chip substantially more expensive.

 

[UBS28]

Can't the Airs get a faster M1, not M1-Pro/Max but a faster version of the regular M1, this year without moving the 3nm?

What is the point of releasing something this year? It will then most likely be a "paper" launch as Apple cannot produce enough chips at the moment.

I am even curious to see how Apple will launch their new iPhone this year, which will definitely get a much higher priority than Mac (as the iPhone is Apple their money maker).

 

[JPack]

Rationale from Kuo supporting the rumor.

https://twitter.com/i/web/status/1530849357154914304

https://twitter.com/i/web/status/1530849362158710784

https://twitter.com/i/web/status/1530849364092276736

 

[andrewstirling]

What is the point of releasing something this year? It will then most likely be a "paper" launch as Apple cannot produce enough chips at the moment.

I am even curious to see how Apple will launch their new iPhone this year, which will definitely get a much higher priority than Mac (as the iPhone is Apple their money maker).

I guess they could improve it in other ways. Better screen, more ports etc alongside a modest speed increase.

The air has never really been aimed at those chasing speed so they might get away with it.

 

[darngooddesign]

What is the point of releasing something this year? It will then most likely be a "paper" launch as Apple cannot produce enough chips at the moment.

I am even curious to see how Apple will launch their new iPhone this year, which will definitely get a much higher priority than Mac (as the iPhone is Apple their money maker).

Apple can not-meet-demand with either the M1s or M2s; the result is the same but the latter demonstrates that Apple Silicon has long-term momentum instead of the transition being a sporadic success.

I wonder if the chip shortage actually helps with the M2 because it means there aren't a lot of leftover M1s that Apple has to do something with.

 

[leman]

Looking at the past trends, Apple Silicon development operates on a two year cadence, with complementary upgrades being delivered each year. The pattern that emerges is that one year brings upgrades to the memory subsystem and the GPU, and then the next year brings CPU core redesign. This was the path Apple has used at least since A12

Even A16/M2 is stil "stuck" at N5P, it does not mean we cannot expect significant improvements. Apple did manage to double caches and add more GPU cores on the A15 for example, so limits of 5nm are not yet reached. Let's not forget that this is still the most cutting edge production node right now. Improved cache, redesigned circuitry, more execution pipelines — one does not need a smaller node to do these things, especially as 5nm is most likely more mature now and yields are better than two years ago.

 

[senttoschool]

On a tangent unrelated to this comment of yours: I'm glad you've come around to the low likelihood of Apple adopting an annual cadence for laptop SoC refreshes.

With the M2 being based on A15 instead of A16, we're more likely to get a yearly cadence.

 

[leman]

With the M2 being based on A15 instead of A16, we're more likely to get a yearly cadence.

Yep, that’s what it seems. All things considered, disappointing, but I assume there are still hiccups in the rollout process. Hopefully these releases will be smoother in the future. M2 is what, almost 10 months late?

 

[senttoschool]

Yep, that’s what it seems. All things considered, disappointing, but I assume there are still hiccups in the rollout process. Hopefully these releases will be smoother in the future. M2 is what, almost 10 months late?

All reports suggested that the M2 was ready by the Spring event. In fact, there were credible reports that the M2 will launch with the 13" MBP in Spring.

This suggests that M2's delayed launch had a lot to do with the new Air being late. Apple didn't want to introduce the M2 with an old design. They wanted to introduce M2 and the new Air together.

I was initially on the yearly-update train. Then jumped to the 2-year update train when rumors were suggesting that the M2 will launch in Fall of this year. Now I'm squarely back to the yearly-train.

Yearly updates to the M series make a lot of sense. This is because unlike the old A-X chips, which only went into iPads, the M series will go into far more products. It's now worth the effort to update it yearly.

A-X went into iPads only.
M series go into iPad Pros, iPad Air, Macbook Air, Mac Mini, Macbook Pro, iMac.

 

[leman]

All reports suggested that the M2 was ready by the Spring event. In fact, there were credible reports that the M2 will launch with the 13" MBP in Spring.

This suggests that M2's delayed launch had a lot to do with the new Air being late. Apple didn't want to introduce the M2 with an old design. They wanted to introduce M2 and the new Air together.

I was initially on the yearly-update train. Then jumped to the 2-year update train when rumors were suggesting that the M2 will launch in Fall of this year. Now I'm squarely back to the yearly-train.

Yeah, I think so too. Lockdowns in china and russian war crimes don’t make the best environment for timely product rollouts… I’m sure we’ll get yearly releases in the future: consumer stuff in spring, prosumer stuff in fall.

 

[senttoschool]

Yeah, I think so too. Lockdowns in china and russian war crimes don’t make the best environment for timely product rollouts… I’m sure we’ll get yearly releases in the future: consumer stuff in spring, prosumer stuff in fall.

If Apple can consistently launch a new M in the Summer, that's pretty exciting. This means Apple should be able to keep up with AMD and Intel's raw performance numbers while of course winning the perf/watt battle.

I was really hoping for yearly updates. Now we're more likely to get it than not. Two years seemed really long and Apple would risk falling behind in raw performance.

 

[thenewperson]

I wonder if the release dates from here on are more like today: A(n) chip in September of 202X, M(n) chip in summer 202X+1.

My hope is they slowly increase things a bit ahead of 1 year to align more with the 2020 release of A14 and M1. So maybe early 2023 we could get M3 on 3nm and M4 late 2024. But that seems more unlikely.

 

[senttoschool]

I wonder if the release dates from here on are more like today: A(n) chip in September of 202X, M(n) chip in summer 202X+1.

My hope is they slowly increase things a bit ahead of 1 year to align more with the 2020 release of A14 and M1. So maybe early 2023 we could get M3 on 3nm and M4 late 2024. But that seems more unlikely.

I think it's more likely to be A series in September, then M series in Spring the next year. All rumors pointed to the Spring event as the launch for M2.

Or perhaps Apple has always intended for the new A and M series to be released alongside each other and that the new Air design delay, chip shortage, inflation, supply chain issues, and lockdowns have messed this schedule up.

I do think that going forward, the A and M series will get closer and closer with their launch dates.

 

[thenewperson]

Yeah, Spring does sound more likely to me, though I really liked the much closer launch of 2020.

 

[jdb8167]

Yep, that’s what it seems. All things considered, disappointing, but I assume there are still hiccups in the rollout process. Hopefully these releases will be smoother in the future. M2 is what, almost 10 months late?

10 months? The M1 came out at the end of November 2020. So even if Apple meant to have a yearly cadence, that makes it 6 months late. And I doubt that Apple intends to have Mac SoCs updated yearly. Every 18 months seems more likely. That puts the M2 pretty squarely on schedule.

 

[BigPotatoLobbyist]

All reports suggested that the M2 was ready by the Spring event. In fact, there were credible reports that the M2 will launch with the 13" MBP in Spring.

This suggests that M2's delayed launch had a lot to do with the new Air being late. Apple didn't want to introduce the M2 with an old design. They wanted to introduce M2 and the new Air together.

I was initially on the yearly-update train. Then jumped to the 2-year update train when rumors were suggesting that the M2 will launch in Fall of this year. Now I'm squarely back to the yearly-train.

Yearly updates to the M series make a lot of sense. This is because unlike the old A-X chips, which only went into iPads, the M series will go into far more products. It's now worth the effort to update it yearly.

A-X went into iPads only.
M series go into iPad Pros, iPad Air, Macbook Air, Mac Mini, Macbook Pro, iMac.

This is copium. There will be not be yearly updates, it’s absurdly expensive to do so and Apple have more profitable uses of leading-edge wafer capacity.

Many here need to get this through their skulls: there is not going to be a Mega Pro Mx Max annual release. And as for the A15 point, that in no way contradicts my point about the cadence. They will follow the A-chip architectures as predicted, but that does not ipso facto suggest an annual cadence.

 

[BigPotatoLobbyist]

I think it's more likely to be A series in September, then M series in Spring the next year. All rumors pointed to the Spring event as the launch for M2.

Or perhaps Apple has always intended for the new A and M series to be released alongside each other and that the new Air design delay, chip shortage, inflation, supply chain issues, and lockdowns have messed this schedule up.

I do think that going forward, the A and M series will get closer and closer with their launch dates.

LOL, the A and M series are by the current trend, growing in distance. It was only 2 months apart in 2020. Now it’s almost a year, but still the A series is at the forefront, which makes sense.