Rumor: New 14-Inch and 16-Inch MacBook Pros Delayed Until Next Year

[PinkyMacGodess]

There will always be a later model coming out. The vast majority of users don't live in this forum, and have a different mindset.

Buy what you need, when you need it. There's good offers to be had now, 10%-20% off. I garantee you their will be a 10-20% price INCREASE when the m2 pros come out next year. at which point you're looking at up to 40% value proposition difference, plus 6 months use of an m1. they will not be that much faster, heck there's only a tiny portion of work loads that max out an m1 system right now.

I worked at Circuit City years ago, years before they died too.

One guy comes swaggering in and asks why he should buy anything now because it will only be out of date in a year. I said 'Or less'. So he asks me how to make sure what he buys will still be good in a year. I said 'Buy big, buy the top of the line. It should last longer'. But he asks again, 'Why should I buy anything now when it will only be out of date in a year?' I said 'By that logic you won't be buying anything ever, so why come in here looking. Everything here for sale is already out of date if you look closely at it, so don't buy anything!'. His wife punched his arm and said 'SEE! That's what I've been saying all along!' I'll bet she was near impossible to live with for a few months. They did indeed buy something that night.

For longevity, buy expensive, but even that isn't a solid buy against the future. All of the people that bought the days before a major (minor) hardware update came out likely missed out on minor incremental performance improvements. (I read that IBM even released a mainframe that was throttled down to not cannibalize sales of other models. They eventually removed the leash) At some point people need to realize they are trying to hit a moving target. If you need a system, buy it. Don't let yourself be sucked into the 'it'll be out of date too soon' morass. I mean, unless you definitely KNOW there is an update you need coming out. It's not like the M3 or M5 is likely going to be that much faster to make the agony (remember you needed a computer) of waiting worth it. Buy. Buy now. Buy and be happy...

 

[PowerButton]

There’s absolutely no reason for Apple to continue the yearly refresh cycles right now. I hope they shift to a 2 year one. It would benefit both the consumers as well as the environment.

That model doesn’t work if you’re the one designing the chips. If Apple was purchasing their chips, sure they could wait a cycle and update with the most current chips available.

 

[Carlson-online]

Basically, if you are the kind of person that can wait 6 months in the hope of a slight performance boost, i'd argue the *need* for the hardware isn't really a need, but a want.

Buy what you *need*, when you need it. If my laptop gave way on say march 1st, 2023, i would still go out and buy a 2021 macbook pro, because without it I haev no revenue source.

 

[Nords777]

In some 5 months a new model will come. so to buy the old for $5000 is no sense. in fact, the M1 is already an outgoing, but that’s not even the point, the new faster memory can much more improve the processing of complex tasks. Apple done the mistake of postponing the release of the macbook on the M2, I was already ready to pull out the money.
Now Apple will have to wait for my money for six months , now I have laptop to work for this time.

 

[Carlson-online]

In some 5 months a new model will come. so to buy the old for $5000 is no sense. in fact, the M1 is already an outgoing, but that’s not even the point, the new faster memory can much more improve the processing of complex tasks. Apple done the mistake of postponing the release of the macbook on the M2, I was already ready to pull out the money.
Now Apple will have to wait for my money for six months , now I have laptop to work for this time.

So it is a want, not a need then

You're free to use your money however you want, just don't make the assumption that everyone else will make the same decision as you

 

[chucker23n1]

the new faster memory can much more improve the processing of complex tasks.

How do you know the M2 Pro will have faster memory?

(The M1 Pro already has the speed the M2 got. They might go to LPDDR5X, but they also might not.

Don’t buy based on rumors and assumptions.

 

[hoodlum90]

Late 2023 3nm iPhone and Macbook Air
Mid 2024 3nm iPad Pro and Macbook Pro
You heard it here first folks.

TSMC had a large customer delay their orders for this year until the 2nd half of next year. That 50% drop in orders for this year must be the iPhone 15, which will likely stay on 4nm. I suspect Apple was not happy with the initial 3nm delay for the Macbook Pro, Etc and didn't want to risk a delay to the iphone 15.

TSMC Slashed Supply Chain Orders Up To 40% Or 50% - Industry News - News - PANASYS

TSMC Slashed Supply Chain Orders Up To 40% Or 50%,Industry News

www.panadisplay.com

It is reported in the industry that TSMC encountered a temporary cancellation of orders by a major customer for the 3-nanometer process. As a result, the monthly production capacity of the 3-nanometer process that will be mass-produced in the near future is significantly lower than the original plan, only about 10,000 pieces, and will have to wait until the second half of next year.

 

[aapl owner]

What do you mean by "die combining issues"? I know they combine two M1 Max's to get an M1 Ultra, but I've not heard of any issues with that.

Here's what I get with Google:
View attachment 2105921

The M1 Ultra, which combines 2 M1 Max dice, does not scale properly because of data bus issues (the die-to-die interface bus contains over 2000 connections that are made during the wafer processing). The complexity of the interface limited the ability of the die to combine properly from a performance standpoint and the M1 Ultra was the building block for the new Mac Pro which, rumor had it, was combining 4 M1 Max dice and had the same die combining issues. The M1 Pro and M1 Max did not have any issues. The M2 may also have similar interface problems which is most likely why the Mac Pro has yet to be released (speculation on my part). What Apple is trying to pull off with the massive number of connections within the die-to-die interface is impressive and very ambitious but also very complex to get working properly.

 

[theorist9]

The M1 Ultra, which combines 2 M1 Max dice, does not scale properly because of data bus issues (the die-to-die interface bus contains over 2000 connections that are made during the wafer processing). The complexity of the interface limited the ability of the die to combine properly from a performance standpoint and the M1 Ultra was the building block for the new Mac Pro which, rumor had it, was combining 4 M1 Max dice and had the same die combining issues. The M1 Pro and M1 Max did not have any issues. The M2 may also have similar interface problems which is most likely why the Mac Pro has yet to be released (speculation on my part). What Apple is trying to pull off with the massive number of connections within the die-to-die interface is impressive and very ambitious but also very complex to get working properly.

Do you have a credible reference to support this? Scaling issues are not uncommon with high-core-count CPUs, and often the reason is that the software scales poorly at high core counts. That would not be surprising to encounter with the Ultra, since AS is relative new, and highly parallelized software may not yet be sufficiently optimized to scale well beyond 10 cores.

Indeed, Craig A. Hunter used a computational fluid dynamics app (NASA USM3D CFD solver) to compare the scaling of the Ultra to various Mac Pro Xeons, and found the Ultra's scaling performance was exceptional. He tested it with up to 16 cores, which means it was using both "halves" of the Ultra (which has 20 cores):

2022 Mac Studio (20-core M1 Ultra) Review

 

[quarkysg]

Do you have a credible reference to support this? Scaling issues are not uncommon with high-core-count CPUs, and typically the reason is that the software scales poorly at high core counts.

My understanding is that in the scenario of the CFD benchmark performed, the bottleneck was mainly caused by the lack of memory bandwidth, resulting in the CPU cores being starved of data to process for the Intel CPUs, including the Xeons.

So it appears that Apple's UltraFusion interconnect is doing it's job as designed, with excellent results I might add. So Apple's engineers knew where the bottleneck was. I suspect that is the reason why the M1 SoCs are running at the 3GHz range as going higher will result in diminishing returns.

 

[quarkysg]

The M1 Ultra, which combines 2 M1 Max dice, does not scale properly because of data bus issues (the die-to-die interface bus contains over 2000 connections that are made during the wafer processing).

You are off by a factor 5. Apple said UltraFusion carries more than 10,000 signals between the two M1 Max dies.

 

[theorist9]

My understanding is that in the scenario of the CFD benchmark performed, the bottleneck was mainly caused by the lack of memory bandwidth, resulting in the CPU cores being starved of data to process for the Intel CPUs, including the Xeons.

So it appears that Apple's UltraFusion interconnect is doing it's job as designed, with excellent results I might add. So Apple's engineers knew where the bottleneck was. I suspect that is the reason why the M1 SoCs are running at the 3GHz range as going higher will result in diminishing returns.

Though with the majority of tasks being single-threaded, I am curious why they didn't offer something equivalent to Intel's Turbo Boost for cases in which you are only maxing out one or a few cores. With just a few cores running at high clocks, they'd certainly have the memory bandwidth, and the thermals, to handle this. It would reduce the efficiency, but I think desktop users, in particular, would welcome such a tradeoff.

You could offer it to laptop users as well, but make it a switchable "high performance" mode. Apple does have such a mode for the 16" MBP, but I've heard it doesn't do much. By contrast, no one is going to say that about a mode that allows one or two cores to max out at, say, 5 GHz.

 

[quarkysg]

Though with the majority of tasks being single-threaded, I am curious why they didn't offer something equivalent to Intel's Turbo Boost for cases in which you are only maxing out one or a few cores. With just a few cores running at high clocks, they'd certainly have the memory bandwidth, and the thermals, to handle this. It would reduce the efficiency, but I think desktop users, in particular, would welcome such a tradeoff.

Probably due to a trade-offs of work done vs power required. My very simplistic napkin maths tells me that, if the M1 CPU can retire 1 instruction per clock and each instruction can consume 64-bit of data, at 3.2GHz, the memory bandwidth required will be:

3.2 GHz * 64 bits / 8 bits = 25.6 GB/s

Since the CPU have many ALUs, they can probably retire many instructions per clock?

So going higher clocks requires more bandwidth.

Other processing cores in the SoC (like GPU, NPU, etc) also required data, so the CPU alone should not be dominating the memory bandwidth.

I may be completely off based here, so happy to be corrected.

 

[theorist9]

Probably due to a trade-offs of work done vs power required. My very simplistic napkin maths tells me that, if the M1 CPU can retire 1 instruction per clock and each instruction can consume 64-bit of data, at 3.2GHz, the memory bandwidth required will be:

3.2 GHz * 64 bits / 8 bits = 25.6 GB/s

Since the CPU have many ALUs, they can probably retire many instructions per clock?

So going higher clocks requires more bandwidth.

Other processing cores in the SoC (like GPU, NPU, etc) also required data, so the CPU alone should not be dominating the memory bandwidth.

I may be completely off based here, so happy to be corrected.

Maybe this is overly simplistic, but if the Max has enough bandwidth to run 10 cores at ~3 GHz (including attendant GPU operations), then why, in the case where it's only maxing out, say, 2 cores, why wouldn't it have the bandwidth to run those at 5 GHz, given the memory demand from the remaining eight would be much lower?

 

[quarkysg]

Maybe this is overly simplistic, but if the Max has enough bandwidth to run 10 cores at ~3 GHz (including attendant GPU operations), then why, in the case where it's only maxing out, say, 2 cores, why wouldn't it have the bandwidth to run those at 5 GHz, given the memory demand from the remaining eight would be much lower?

In theory Apple could push the CPU cores all the way to 5GHz if they want to I suspect, but that means you need to make sure the L1, L2 and SLC cache is similarly clocked sufficiently high to accommodate the CPU cores, and this will likely push the power envelope of the CPU core too high over Apple's engineer target.

Apple is unlikely trying to be top dog here. They likely are targeting to meet a performance profile and likely accept the tradeoff between top single core vs balanced overall SoC performance profile.

 

[Tagbert]

These are the ones that will see the return of the glowing Apple logo.

Apple is unlikely to make any design changes for a while. They will update the chips, perhaps more than once, before the next design change.

 

[theorist9]

In theory Apple could push the CPU cores all the way to 5GHz if they want to I suspect, but that means you need to make sure the L1, L2 and SLC cache is similarly clocked sufficiently high to accommodate the CPU cores, and this will likely push the power envelope of the CPU core too high over Apple's engineer target.

Apple is unlikely trying to be top dog here. They likely are targeting to meet a performance profile and likely accept the tradeoff between top single core vs balanced overall SoC performance profile.

That's along the lines of what I was thinking might be the explanation—in order to get the great efficiency they did, they needed to optimize it for ~3 GHz (3.2 for M1, 3.5 for M2), which meant it wouldn't work well with significantly higher clocks. Hence no turbo boost.

 

[Allen_Wentz]

Makes total sense, the M1 Pro/Max MBPs are still the king of the hill and a heck of a lot of firepower; why bother cannibalizing sales of those? Might as well as squeeze another half a year of sales out of them.

The reason is that with the year-newer M2 on the street for months now, sales of M1 (probably except for the still-new Studios) are likely dramatically shrinking. This is tech and many high-end MBP buyers about due to upgrade (like me) will wait for the latest update before upgrading to a box to use for the next 3-6 years.

 

[1129846]

Makes total sense, the M1 Pro/Max MBPs are still the king of the hill and a heck of a lot of firepower; why bother cannibalizing sales of those? Might as well as squeeze another half a year of sales out of them.

Because sales are dying already. A lot of people are like me who are sitting on buying a new Mac until the M2 14/16 pros come out.

T
As soon as they are released I plan on buying a 14in m2 pro with some upgrades.

 

[Allen_Wentz]

There will always be a later model coming out. The vast majority of users don't live in this forum, and have a different mindset.

Buy what you need, when you need it. There's good offers to be had now, 10%-20% off. I garantee you their will be a 10-20% price INCREASE when the m2 pros come out next year. at which point you're looking at up to 40% value proposition difference, plus 6 months use of an m1. they will not be that much faster, heck there's only a tiny portion of work loads that max out an m1 system right now.

Sorry that old "Buy what you need, when you need it" is so simplistic as to be wrong. Most buyers upgrade their computers infrequently, typically every 3-6 years. Normally there is not a precise "when you need it" time, but rather a slowly aging out box. As such, when upgrades are expected in the next few months it makes sense to wait and see the latest before making one's 3-6 year purchase decision.

 

[Allen_Wentz]

Basically, if you are the kind of person that can wait 6 months in the hope of a slight performance boost, i'd argue the *need* for the hardware isn't really a need, but a want.

Buy what you *need*, when you need it. If my laptop gave way on say march 1st, 2023, i would still go out and buy a 2021 macbook pro, because without it I haev no revenue source.

Wrong - - except for the rare event when one's "laptop gives way." More often is that a box lasts for 3-6 years and is slowly aging out for months. So normally there is plenty of time to analyze and wait for opportune upgrade purchase timing.

Buying at the low end like MBAs and Minis may be a bit different, since such boxes are often entry level. But this thread discusses the higher end like MBPs with Max/Pro SoC.

 

[Allen_Wentz]

So it is a want, not a need then

You're free to use your money however you want, just don't make the assumption that everyone else will make the same decision as you

Your definition of "want" versus "need" is skewed. Folks buy high end MBPs on a 3-6 year cycle, with typically a long period of the older box aging out for months. Your implication that folks must wait for the old box to flat die, and only then do they "need" a new box is lame thinking.

Smart buyers at the high end we are discussing choose their purchase timing. It is not some silly want versus need calculation.

 

[theorist9]

I garantee you their will be a 10-20% price INCREASE when the m2 pros come out next year.

Highly unlikely. Apple generally keeps pricing the the same, or nearly so, when it does a spec bump. Indeed, we can revisit this when the M2 MBP's are released.

So it is a want, not a need then

Just because someone can wait a few months doesn't change a need into a want. That's an overly simplistic view.

By your argument, if you're supposed to, say, get a colonoscopy once every 5 years, then unless you have to get it *exactly* every five years (as opposed to 5 years ± 6 months), it's a want rather than a need. Which is of course ridiculous.

Likewise, you can have an older computer that needs upgrading, and that need is not downgraded to a mere want just because you don't have to upgrade immediately.

 

[Tagbert]

The reason is that with the year-newer M2 on the street for months now, sales of M1 (probably except for the still-new Studios) are likely dramatically shrinking. This is tech and many high-end MBP buyers about due to upgrade (like me) will wait for the latest update before upgrading to a box to use for the next 3-6 years.

Most customers don’t care that much about the specific chip. M1 vs M2 is not going to make much difference in how most customers use their computers.

 

[chucker23n1]

Most customers don’t care that much about the specific chip.

This.

1. I’m due for a new Mac.
2. What’s the current one? Will I hate that for any reason?
3. Buy it.
4. Hey, neat upgrade.

“Oh, but if I had read the rumors, I’d know that a new thing may or may not come out between today and the heat death of the universe” is irrelevant.