User predictions for Apple Prosumer Silicon (M1X, M2 etc.)

[deconstruct60]

In a practical sense how wider can they get? They are already 8 wide when the closest competitor is 4 wide.

On general purpose, average code? Probably not much wider. Instruction level parallism isn't too much farther until hit a branch or the need for a previous interium result is require for the next computation. Can only skip ahead only so far in general. (If instruction level parallelism was super wide VLIW wouldn't have run into a 'wall'). On "dense" vector martrix code there is probably room. More execution units on seperate execution paths is also "wider" .

Things like the previously mentioned obj-c-dispatch predictor will help close more "stall and no-op" bubbles which has the side effect of increaing IPC count. ( what more so are doing is getting rid of time those wide units aren't sitting around doing nothing). So "smarter" predictors can increase IPC. ( there are some caps on just how smart those can be also. security wise and accuracy. )

Apple hasn't touched SMT. As long as the baseline design of the processors is more so driven by the mobile (and iPhone especially) they probably won't pick it up.

 

[crazy dave]

What? What is going on here, is MR suicidal??

I dunno, I can’t think what he could’ve done if he is actually suspended. I mean he’s not that kind of poster, obviously

 

[deconstruct60]

But the clock rate will probably be higher too (especially since they can get a little bit better cooling out of these entirely-new case designs).

The iMac 24" isn't a "higher clock rate" shift in case design. Putting MagSafe and HDMI probably isn't much of a shift for the MBP 16" . MBP 16" wasn't all that great thermal wise for what they were dealing with on Intel x86_64. Going down, from there, would be a better fit. Not so much better cooling but the higher headroom that was already there ( like the Mini. ). A M-series MBP 16" at max compute with effectively inaudible fan noise likely would get a big "touchdown dance" on stage during an Apple product reveal.

Leaks on the Mac Pro are about a "half sized" model. Again not keeping same system TDP coverage.

Apple slide back at WWDC 2020 was suggestive that they want to pull the desktops "back" to lower power zones the laptops have lived in. And to push the laptops up into the performance zones that the desktops covered (without going higher on power). All the build the systems they want to build. (.e.,g evolutionary thinner laptops over the last two decades. ) , Mac Pro 2013 , etc. Apple has a general track record; especially on laptops.

The GPU may be on a separate die, though. Not sure - it’s not really necessary given the size of the reticle, but if they want to offer a range of graphics capabilities it may be easier for them to do it that way.

For this MBP 16" targeted implementation. However, if they crank up the "big" cores to 20+ range and the GPU "cores" > 80 they could get close to the reticle if still committed to feeding that with custom , on package RAM and various fixed function media , other I/O , and large caches to feed the massive number of data consumers.

Apple may be trying to vary both CPU and GPU capabilities if shooting for something that scales up to fit the rest of the Mac Pro line they haven't covered systems that get the M1.

 

[deconstruct60]

So in a way it is @cmaier fault that AMD didn’t adopt IA-64 and kept x86 alive?

Errr, no. There are folks inside of Intel advocating for a 64-bit x86 also (was talk of this back on comp.arch rewsgroup back in the day) . if AMD hadn't done theirs Intel probably would have eventually done their own.

Itanium (IA-64) was a server market RISC chip killer . ( HP PA-RISC (and acquired DEC Alpha) , Digitial Sparc, MIPS ( in SGI ). IBM Power ). Can throw Cray and some smaller narrow HPC servers into that boat too. To a lessor extent a IBM mainframe market blunter also ( don't think Intel had delusions of "killing" but definiately boxing them in ).

HP killed off their RISC implementation when signed up with the project. So that was pragmatically two "dead" right there. SGI bolted and MIPS was left hanging. Sun/SPARC didn't bite but was a problem for them. IBtM moved POwer to overlap with mainframe (Z-seriees) so a partial collapse of competitors there.

The x86 faction in Intel partially sabotaged Itanium by saddling the initial version with the 32-bit x86 compatibility mode cruft that didn't make much sense. ( IBM put x86 "sidecars" on their mainframe offerings a bit later without pissing around with the main CPU die implementation. ). So part of Itanium's problem it was inside of a corporation where another part of the company wanted to kill it off. They had other issues on pedistrain "dick, jane , spot" code that was outside of the super hard core HPC zone. ( and major mismatch conflict with big budget system assigned to RDBMS workloads. )

Similar to how when Intel poured "x86 ketchup" on top of Larrabee. Just Why? ( besides internal corporate intrigue. ). backward compatibility with some generic 90's DOS code buys what in a GPU? Nothing.
But tilts the scales on the project enough to send it off into the weeds.


AMD64 was good in that it got AMD and Intel to cross license on new tech again. It opened up some flexibility on making progress in general . ( got virtualization support over time and some other additions on that same jointly more forward infrastructure.). Intel's one way walled x86 garden was bad in general to long term. Too much pragmatic monopoly power than leads to lazyness ( and "only paranoid survive" more about suppression than staying diligent and moving forward. )

 

[09872738]

I dunno, I can’t think what he could’ve done if he is actually suspended. I mean he’s not that kind of poster, obviously

I believe I got an idea. Did not find the suspected post though (removed?). Basically its about an answer to a, well, not particularly clever post in the „M1 CPU and GPU are disappointing“ - thread. He called it out, not holding back his opinion. While he critizised the post rather than the person posting, it may have been considered too harsh. I‘ve seen mods penalising similar things.
Anyway, just a guess

 

[crazy dave]

I have an idea. Did not find the suspected post though (removed?). Basically its about an answer to a, well, not particularly clever post in the „M1 CPU and GPU are disappointing“ - thread. He called it out, not holding back his opinion. While he critizised the post rather than the person posting it may have been considered too harsh. I‘ve seen mods penalising similar things.
Anyway, just a guess

Maybe ... I’m active on that thread and don’t see a missing post or remember one that could’ve gotten him suspended but entirely plausible I missed it. Although I guess there was a mod warning? But I thought that was awhile ago. I thought he actually was muted and reasonable with less snark that I was expecting given the quality and tenor of the posts with the weird and tiresome assertions being made therein. I generally try to be civil as well, but a couple of the posters there are really pushing it and I can’t say I didn’t respond a little harshly myself.

 

[09872738]

Maybe ... I’m active on that thread and don’t see a missing post or remember one that could’ve gotten him suspended but entirely plausible I missed it. Although I guess there was a mod warning? But I thought that was awhile ago. I thought he actually was muted and reasonable with less snark that I was expecting given the quality and tenor of the posts with the weird and tiresome assertions being made therein. I generally try to be civil as well, but a couple of the posters there are really pushing it and I can’t say I didn’t respond a little harshly myself.

If I am correct - and its of course entirely possible I am not - it was a reply (I do not find any more) to this: https://forums.macrumors.com/thread...d-is-very-disappointing.2293062/post-29970947

And yes, sometimes I am guilty too. Some posts just happen to provoke a stronger reply

 

[crazy dave]

If I am correct - and its of course entirely possible I am not - it was a reply (I do not find any more) to this: https://forums.macrumors.com/thread...d-is-very-disappointing.2293062/post-29970947

And yes, sometimes I am guilty too. Some posts just happen to provoke a stronger reply

You’re right. Now I remember it, that was probably it.

Edit: I understand why the rules are what they are, but that does force people to treat clearly bad faith posts as good faith posts no matter how in obviously bad faith they are. Which is itself not a good thing.

 

[leman]

You’re right. Now I remember it, that was probably it.

Edit: I understand why the rules are what they are, but that does force people to treat clearly bad faith posts as good faith posts no matter how in obviously bad faith they are. Which is itself not a good thing.

In my experience, replying to deliberately provocative anti-intellectual posts in a serous manner works. If you reply emotionally, you are just adding to the mess and it just goes downhill from there. Just firmly rebutting the poster as objectively as you can might not change their opinion or attitude, but it will do a lot to calm the thread down and to set a level for discussion for other posters. Best of course would be ignoring certain posts, but that unfortunately does not work.

 

[crazy dave]

In my experience, replying to deliberately provocative anti-intellectual posts in a serous manner works. If you reply emotionally, you are just adding to the mess and it just goes downhill from there. Just firmly rebutting the poster as objectively as you can might not change their opinion or attitude, but it will do a lot to calm the thread down and to set a level for discussion for other posters. Best of course would be ignoring certain posts, but that unfortunately does not work.

It depends. Nominally I agree with you and that’s the course I typically take if someone is actually engaging in conversation. But if a person is ignoring rebuttals completely and continuing to post repetitiously the same debunked tripe then they aren’t trying to have a conversation, they’re posting in bad faith. As you said, ignoring them doesn’t work because someone else will get pissed off and respond which is of course what they’re after and tolerating that makes the community worse not better. I mean cmaier is the one that got banned here. Obviously the mods disagree with me though I’ll grant you enforcement of bans against the described behavior is difficult as you can’t just do it off of a single post but rather posting behavior. So I’m not hard over on this, just noting their policy is a trade off and has downsides.

Edit: anyway sorry to derail your thread a bit, I just noticed it and wondered what happened. Tomorrow should be fun! (Hopefully)

 

[deconstruct60]

First of all, by the time these macs ship it will have been a year since firestorm/ice storm. How long do you expect them not to innovate on those?

This doesn't match up if line up that there are likely multiple instances of the M series.

M1 -- out last Fall ( been "born" and aging. )
M-bigger-die ( what suffix Apple wants to slap on it is misdirection). ( has not shipped so hasn't been "born" yet).

So the notion that M-bigger-die is now to be put on Social Security and sent to a retirement home is a bit odd. It hasn't even shipped, so how could it be "old". Meanwhile the small die has shipped and has "aged".


Apple shipped the iPad Pro after they got the iPhone going on all cylinders. There they primarily put the A__X series on a new every process shrink available. 16nm (A9X) -> 10nm (A10X) -> 7nm (A12X) . (A12Z was just place holder for time. Same die as the A12X will a GPU core turned on. Probably was there all along turned off to very slightly goose yields. .) Their iPad Pro Mac "bigger die" showed on the next shrink.

It has been on somewhat of a 18-24 month cycle. If Apple is ignoring the smaller M1 to do cycle tweak then could be that they are setting these two up to play leap-frog. smaller M1 class product even years and bigger die odd years. They could pipleline development on the different uncore die aspects and just pick up iPhone cores in the even odd years. that would allow them to limit Mac design resources and keep the iPhone (and lower end iPad) on the yearly SoC update cycle.

Apple's pretty long track record is that they do bigger dies slower. That might speed up for the M1 class sized die since the volume of the iPad Pro + MBA + MBP 13 + lower Mini + smaller iMac is probalbly sufficiently large to keep them interested in yearly updates. ( As Apple said more than half of the Macs sold this year have the M-series already in them. The major volume is not left in the still unconverted segment of the line up).

However, the lower the volume the slower that M-series sized dis is likely to come. ( based on their established track record). If Apple is in a "rush" to get onto the bigger core then pretty good chance that is primarily so they can back burner it again.

 

[AutisticGuy]

Assumptions/beliefs for predictions

• I believe the M1 was basically the new A14x for the new iPad that was rebranded so it could be placed in early laptops and the 24 inch iMac. This would allow Apple to place these chips in the existing MacBook Pro and MacBook Air chassis in order to contrast performance with existing Intel processors. I believe Apple wanted to ESPECIALLY contrast battery life with superior performance without changing the existing laptops.
• The M1 accomplished two things for Apple. First, it allowed Apple to begin the transition to Apple Silicon earlier then they otherwise would have. Second, it allowed Apple to market the new Apple Silicon's performance/efficiency advantage over x86 while eliminating all other factors. Just throw an M1 into the existing laptops with minimal changes to said laptops so that critics could not claim the re-engineering of said laptops contributed to superior performance and/or battery life. For instance, if Apple created a new 14 inch model and placed the M1 in it, it could have been claimed that perhaps the new screen was more energy efficient and therefore was a large factor in battery life.
• The above assumptions are important because it's my belief that the new chips that will be placed into the upcoming MacBook Pros are really the first Apple Silicon that will be placed exclusively in laptops. In other words, there isn't a possibility these chips will be placed in a future iPad Pro, for instance. These are the real laptop chips we've been waiting for, and the M1 was a marketing chip/placeholder.
• Evidence for Assumption: First, the rumors chip in the MacBook Pros likely released tomorrow is supposed to have 8 performance and 2 efficiency cores. Do we really believe Apple would keep identical-performing efficiency cores in higher end MacBook Pros in other to basically cut in half the efficiency performance of these higher end machines? Even there will be only 2 efficiency cores, they will be designed for a laptop and therefore will be faster. Second, Apple did release a Mac Mini with an A12Z processor in it for developers, so there is precedence for what I just said. Third, the M1 had extremely limited peripheral support. The fact that it could only work with one monitor was quite surprising. It's almost like the M1 was a mobile processor closer to an iPhone/iPad that was placed in laptops as an initial offering/placeholder.
• My last assumption/belief is that Apple is going to be the only manufacturer of laptops that can offer workstation level cpu performance, incredible battery efficiency, and high end graphics performance all in one laptop. Therefore, they are going to charge a premium for it (to an extent). I believe they'll do that by offering multiple GPU core options which allow them to 1) upcharge for higher graphics performance and 2) bin processors that didn't meet higher CPU/GPU standards like they did with the MacBook Air 7 GPU processor.

Predictions

• The new MacBook Pros will come in 14 and 16 inch displays that will not be mini-led due to supply constraints.
• Peripheral support will be significantly more robust. Support of minimum of 3 4k displays but more likely up to 3 5k displays. They will also come with an SD card slot and 4 USB C slots. Mini-display support will also be present. All of this will occur because these new MacBook Pros will offer workstation level performance.
• For things like web browsing, YouTube, and other low-end tasks, the 14 inch will get close to 25 hours and the 16 inch will get close to 30 hours. For extremely high end tasks, both machines will likely still get 8 hours or more performance. These will be battery life beasts!
• GPU cores will be more varied then just 16 and 32 GPUs for binning purposely. You'll see more what I mean by that when I provide pricing predictions.
• GPU core performance would scale linearly with a 32 core GPU quadrupling the performance of the 8 core GPU if the IPC and clock speed of the CPU stayed the same. However, that won't happen! The current rumor is that the 16 core GPU would somewhat exceed the 5500M and the 32 core would equal the performance of the 3070. I believe we need to add 15 to 25 percent on top of that!
• Since these will be the first true Apple Silicon designed exclusively for laptops, they will offer higher clock speeds and a higher IPC. Single core performance will be 25 percent better then the M1 and that will scale with 8 cores for multicore performance that rivals high end Mac Pros.
• It doesn't matter whether these chips are called M1X or M2, they will NOT be architecturally identical to the M1. However, I predict they will be called M2s but there will be a M2X and only the M1 will stand alone for future Apple Silicon releases.

Pricing Predictions

14 Inch
M2 with 10 core CPU and 14 core GPU (binned), 16 core Neural Engine
16 GB unified memory
512GB SSD
$1799

14 Inch
M2 with 10 core CPU and 16 core GPU, 16 core Neural Engine
16 GB unified memory
1TB SSD
$2000

14 Inch
M2 with 10 core CPU and 24 core GPU (this will be a binned 32 GPU processor with core purposely disable). I believe that thermal restraints will limit these machine to the 24 cores but still allow over 3 times the GPU performance of the M1
16 GB Unified Memory
1 TB SSD
$2400

16 inch
M2 with 10 core CPU and 16 core GPU and 16 core Neural Engine
16 GB unified memory
512 SSD
$2400

16 inch
M2 with 10 core CPU and 26 core GPU (binned 32 core with some cores disabled for marketing/pricing reasons) and 16 core Neural Engine
16 GB unified memory
1TB SSD
$2800

16 inch
M2 with 10 core CPU and 32 core GPU and 16 core Neural Engine
16 GB unified memory
1TB SSD
$3200

Edit: Just to clarify, I believe the 14 inch will be 32 GB max and the 16 inch will be 64 GB max. However, from a pricing perspective, I do not believe Apple will sell higher then 16 GB was a standard config. I also believe memory and SSD pricing for custom config will remain unchanged.

 

[altaic]

16 inch
M2 with 10 core CPU and 32 core GPU and 16 core Neural Engine
16 GB unified memory
1TB SSD
$3200

Interesting predictions. The only thing I disagree with is something you didn’t explain: 16 GB unified memory max. There is a thread that is a deep dive into the memory on the M1, and it seems to me that there would be a 24 or 32 GB offering.

 

[Joelist]

Interesting predictions. The only thing I disagree with is something you didn’t explain: 16 GB unified memory max. There is a thread that is a deep dive into the memory on the M1, and it seems to me that there would be a 24 or 32 GB offering.

Yep! The new SOCs will have 32 and 64 GB configurations also per rumors.

 

[AutisticGuy]

Interesting predictions. The only thing I disagree with is something you didn’t explain: 16 GB unified memory max. There is a thread that is a deep dive into the memory on the M1, and it seems to me that there would be a 24 or 32 GB offering.

Actually, I forgot to mention it. I predict 64 GB max, but the pricing I quoted if for standard configs you can pick up in the store. I don't believe Apple will offer a 32 gig model you can pick up in the store.

 

[Fomalhaut]

In my experience, replying to deliberately provocative anti-intellectual posts in a serous manner works. If you reply emotionally, you are just adding to the mess and it just goes downhill from there. Just firmly rebutting the poster as objectively as you can might not change their opinion or attitude, but it will do a lot to calm the thread down and to set a level for discussion for other posters. Best of course would be ignoring certain posts, but that unfortunately does not work.

I've read that the same approach is most productive for trying to change a conspiracy-theorist's beliefs....provide referenceable data from validated sources, and hope that the cognitive dissonance resolves itself naturally.

 

[crazy dave]

I've read that the same approach is most productive for trying to change a conspiracy-theorists beliefs....provide referenceable data from validated sources, and hope that the cognitive dissonance resolves itself naturally.

If only. For lightly held beliefs that may be enough, but unfortunately for deeply held beliefs that have become a part of a person’s identity, for most people that produces the opposite reaction - the relaying of facts no matter how dispassionately actually cause the conspiracy theory to dig deeper because it is easier for the brain to simply reject facts (even, perhaps especially from authoritative sources) than admit a mistake. The real solution, almost impossible online, is to emotionally connect with someone and become a trusted (note different from authoritative!) source. It’s basically deprogramming which is less about facts and logic than it is about emotion and trust.

 

[deconstruct60]

Evidence for Assumption: First, the rumors chip in the MacBook Pros likely released tomorrow is supposed to have 8 performance and 2 efficiency cores. Do we really believe Apple would keep identical-performing efficiency cores in higher end MacBook Pros in other to basically cut in half the efficiency performance of these higher end machines? Even there will be only 2 efficiency cores, they will be designed for a laptop and therefore will be faster. Second, Apple did release a Mac Mini with an A12Z processor in it for developers, so there is precedence for what I just said. Third, the M1 had extremely limited peripheral support. The fact that it could only work with one monitor was quite surprising. It's almost like the M1 was a mobile processor closer to an iPhone/iPad that was placed in laptops as an initial offering/placeholder.

The E cores are likely the same. More powerful (bigger ) E cores is a bit of an oxymoron. All four E cores and their L1/L2 cache basically add up to the same of one P core. It isn't like they are making some great space tradeoff by dropping 2 E cores doesn't buy another P. It is only "buying" half a P core or maybe one of the GPU "cores" (really a collection of compute units). The substantively crank up the number of P and/or GPU cores they just have to grow the die larger. Dropping the two E could help withg boosting the System Cache a bit. That larger collection of P/GPU cores will simply just burn more power ( when most or all are in use). Probably going to get "good enough" as opposed to completely maximized battery life.

Dropping to two makes far more sense if they are going to start taking this 10-core block ( 8 P , 2 E) and building with it. Two blocks gets 16P 4 E ( as many E as the M1 has. and A12X - A12Z ) and a four blocks get 32P and 8E ( which is more. ). Those two biger constructions probably won't show up in the MBP 16", but the scale down to one block factor in.

if they kept 4 E's in the baseline then scale would take it to 8 and 32.... the last probably not being all that useful.

It very similar to using the building block approach you are later doing with the GPU to "go bigger".

Predictions

• The new MacBook Pros will come in 14 and 16 inch displays that will not be mini-led due to supply constraints.
• Peripheral support will be significantly more robust. Support of minimum of 3 4k displays but more likely up to 3 5k displays. They will also come with an SD card slot and 4 USB C slots. Mini-display support will also be present. All of this will occur because these new MacBook Pros will offer workstation level performance.

4 USB C slots isn't all that likely. Relatively consistent rumor that magsafe is coming back. Dropping the need for 4 sockets because power supply was hogging one up. Similarly if there is a HDMI ( more mini-HDMI ) port. A connical video out port so don't need to use the USB-C for that.

Since their are USB 4 ports, there are new USB 4 hubs that can crank up the USB 4 Type-C count if needed.

What may get new on this iteration is Thunderbolt 4 certification on one port (on the side with HDMI ) [ The DisplayPort stream feeding the HDMI switched so that could send two out of the TB4 port. ]

Three externals + the built-in ( 4 total ) wouldn't be a surprising cap.


Apple top the Mac Pro with 8 cores and the 580X basic entry configuration. Possible. ( would fit Apple playbook of pointing at an old model (especially in the GPU aspect) to claim victory). Get most folks with workstation to massively dump their workstations ... probably not.

• For things like web browsing, YouTube, and other low-end tasks, the 14 inch will get close to 25 hours and the 16 inch will get close to 30 hours. For extremely high end tasks, both machines will likely still get 8 hours or more

If 14" has more cores (and less E cores) than the MBP 13" with M1 then the battery life probably isn't going up. the case size is likely similar. Battery size likely similar and if consume more power..... there is probably not a big win.
[ The M1 and Intel version currently there in that class have the same battery. If the 14" is the replacement for the 13" that is left ... likely going to be coupled to the exact same battery as the other in the line up. Cheaper for Apple.]

The smaller MBP 13" and MBA are probably on track to merge. Pretty good chance Apple is going to shoot for max battery life there. And that the 14" will be lower life with higher top end performance.

• performance. These will be battery life beasts!

Loosing the 80-100W dGPU and and a fraction of the 90W ( when loaded) CPU from the Intel model will be the battery gains.

• GPU core performance would scale linearly with a 32 core GPU quadrupling the performance of the 8 core GPU if the IPC and clock speed of the CPU stayed the same. However, that won't happen! The current rumor is that the 16 core GPU would somewhat exceed the 5500M and the 32 core would equal the performance of the 3070. I believe we need to add 15 to 25 percent on top of that!

Cranking the GPU core clocks higher with the same LPDDR4 memory probably won't help much. Especially after have dramatically up the overall core count ( data stream demand by the collective set of cores. )

 

[leman]

More powerful (bigger ) E cores is a bit of an oxymoron.

Why is it an oxymoron? Over the last couple of years, Apple has dramatically improved the performance of their E-cores without increasing the power consumption.

 

[Colstan]

What? What is going on here, is MR suicidal??

Suspending @cmaier from MacRumors is like banning the Pope from the Vatican. Yes, I exaggerate, but hopefully MR isn't foolish enough to make this permanent. I can understand why he gets frustrated with some of the dunderheads that post here, I personally choose not to engage, the effort doesn't seem worth it. Hopefully, once the suspension is over with, he will decide to return to posting.

On the slight chance that a mod is reading this, did you suspend him for what you consider repeated violations, or just take swift punitive action without giving him a warning first? If it is the latter, then that is an entirely disproportionate reaction and giving too much leeway to trollish posters who don't deserve autonomy on the forum, let alone the satisfaction of suspending one of the forum's most respected contributors. At least, that's how I see the situation currently, without further clarification.

 

[nothingtoseehere]

Suspending @cmaier from MacRumors is like banning the Pope from the Vatican. Yes, I exaggerate, but hopefully MR isn't foolish enough to make this permanent. I can understand why he gets frustrated with some of the dunderheads that post here, I personally choose not to engage, the effort doesn't seem worth it. Hopefully, once the suspension is over with, he will decide to return to posting.

On the slight chance that a mod is reading this, did you suspend him for what you consider repeated violations, or just take swift punitive action without giving him a warning first? If it is the latter, then that is an entirely disproportionate reaction and giving too much leeway to trollish posters who don't deserve autonomy on the forum, let alone the satisfaction of suspending one of the forum's most respected contributors. At least, that's how I see the situation currently, without further clarification.

^^This. I also appreciate the posts of @cmaier very much, learn a lot. Couldn't find any offensive things. Of course I do not check all his posts, but all I have seen were constructive and well informed. Maybe there might have been something somewhere but the bulk of his posts is IMO obviously ok with the rules.

 

[UBS28]

16" MBP will have a M2X (a modified A15) starting at 16 GB as the base and 256 GB SSD.

There is no way it will feature a M1X, because Apple could have released a M1X last year already (just slap on more cores and you are done).

 

[altaic]

^^This. I also appreciate the posts of @cmaier very much, learn a lot. Couldn't find any offensive things. Of course I do not check all his posts, but all I have seen were constructive and well informed. Maybe there might have been something somewhere but the bulk of his posts is IMO obviously ok with the rules.

I hope it’s just a 24 hour suspension. His reactions to the WWDC keynote would be valuable.

 

[09872738]

Suspending @cmaier from MacRumors is like banning the Pope from the Vatican.

Well said. Its exactly what I was trying to say

 

[Yebubbleman]

Since most leakers seem to agree that new Pro Macs will make an appearance at the WWDC, and given that there were a lot of recent speculations on what kind of chips these Macs will use, I though it would be a fun thing to collect all the different predictions folks here might come up with. After (and if) the products are announced, we can evaluate all these predictions and see which guesses came closest.

You can post your predictions in this thread, and I will do my best to keep this post up to date with all the different ideas you ladies and gentlemen care to share.

Ground rules

• Be realistic and reasonable
• Be as specific with your prediction as possible
• Include your reasoning
• One prediction per user (if you see multiple likely scenarios you have to pick one)
• Only sufficiently distinct predictions count as new (I will merge multiple predictions if they are very similar and if their authors agree that they can be merged)
• Once your prediction is registered, you cannot take it back or significantly change it! If new leaks come out that rule your guess out, you are out of luck. Minor adjustments are accepted
• Keep the bickering and ranting to the minimum (but do bicker a little bit so that this thread does not die)

Predictions

1. A completely new chip with a new name (tentatively P1) based on a new prosumer-oriented microarchitecture with significantly higher per-core performance compared to M1 and possibly some new features such as ARM vector extensions (SVE/SVE2), hardware ray tracing and possibly others (@leman )
2. The chips in the upcoming 2021 prosumer machines will be based on the next-generation Apple microarchitecture and will probably adopt a name variant based on M2, e.g. M2 pro, M2X etc. (@Fomalhaut , @senttoschool ,@cmaier , @quarkysg, @Andropov )
3. No special chips, just scaled up M-series with more cores (@ader42)
4. The performance cores are 4.35Ghz (420 KB per core (performance cores, 252 instructions + 256 data) 256 KB per core (efficient cores, 256 instructions + 128 data) with L2 (18 MB (performance cores) 8 MB (efficientcores) (copied from @Serban55's post)
5. M1 variant with the name of M1X (@Lemon Olive, @Jorbanead, @thingstoponder, @pshufd , @anshuvorty - detailed post)
6. New microarchitecture at TSMC 5nm+ or 4nm. 8 performance + 2 efficiency cores. Faster clocks (3.5 GHz?). 16 GPU cores (maybe up to 16 GPU cores so maybe some binning). Same number of Neural Engine cores as the M1. Name not guessed but not M1X. I'm guessing not M2 either. New memory controller for up to 64 GB. If LPDDR5 is available I'm guessing Apple goes with that. It looks like Micron has 128 Gb (16GB) LPDDR5. Both Intel Tiger Lake and Qualcomm Snapdragon support LPDDR5. Either 2 or 3 TB/USB4 controllers. PCIE4x4 SSD controller. Two or three 6K displays. That's a lot of cores and I/O so I'm guessing it is a pretty large SoC probably close to double the M1's 119 mm² (@jdb8167)
7. Apple will announce a line of P1 SOCs: one balanced compute/graphics, one mostly compute, and one mostly graphics. Macbook Pro will be balanced, Mac Mini Pro will be offered in all three variants. On the macOS side, we'll see some form of NUMA support. Provides eGPU and modular Mac Pro in one neat system. (@altaic)
8. M1P on same fab process as M1. Same basic P , E ( NPU and GPU ) cores as the M1. (@deconstruct60 #159)

Some background information/leaks

https://www.bloomberg.com/news/arti...ook-pro-macbook-air-revamps-with-faster-chips

Apple's follow-up to M1 chip goes into mass production for Mac

TSMC-made chipsets to replace Intel offerings in laptops set to launch in 2H

asia.nikkei.com

https://www.chinatimes.com/newspapers/20201026000140-260202?chdtv

So, here goes my prediction. I was thinking a while about it and decided to go with what I’d call an “optimistic prediction”, not because I think it’s necessarily the most likely thing for Apple to do, but because I think it would make the best product.

M1 is a great entry-level chip, but the more time passes, the less I am convinced that it will make a good prosumer platform. Apple likely seeks to dominate the prosumer performance segment, and I am not sure that Firestorm (M1 performance core) can reach the speeds to claim an undisputed lead over the x86 world, with Zen3 and Tiger Lake showing strong performance and Alder Lake looming over the horizon. And with A15 reported to enter production, it is clear that Apple has newer tech ready and it would make a lot of sense for them to include some of that tech in the prosumer Macs.

Specifically, my prediction is that the new prosumer Macs will be based on an updated microarchitecture (that might either share roots with A14 or A15), tweaked to better fit mid- and high-powered desktop computers. These new chips will have significantly higher single-core performance than M1 (at least 20-25%) and possibly support new hardware features such as ARM vector extensions (SVE/SVE2) and hardware ray tracing. They will obviously also have water memory interface (my guess is 256-bit LPDDR5 with ~200GB/s bandwidth) and more thunderbolt and display channels.

I also don’t think that Apple will retain the “M” moniker for this family of chips, instead going with something else to highlight their “pro”-nature. My bet is something like “Apple P1” (for pro/performance) or something completely different like “Apple X1”, although I could also see them settling on “M1 Pro” (as first-get Mac Pro Silicon).

If the Bloomberg report is accurate, this “P1” (codename Jade Die) will power the upcoming larger MacBook Pros and possibly the larger iMac, and will come with 8 high performance cores, 2 efficiency cores and options of 16-core of 32-core GPUs. Performance-wise, these should be considerably faster in single core than anything Intel or AMD can build at least until late 2022, and their multi-core performance should be at least 80% of that of the Zen3 16-core 5900X. The 16-core GPU will be roughly on par with the RTX 2060 (mobile) or the Pro 5600M.

Edit: to make this more clear — my speculated "P-series" is different from "M-series" significant ways. The crux of my guess is that the prosumer chips will use a different microarchitecture compared to the M-series, which will continue to power entry-level chips. In other words, my "P1" is not an "M2X" — it's a completely different core and design. In line with this speculation I'd expect the M-series to continue to share the microarchitecture with the iPhone chip, while the P-series to be something else.

I don't have a clear-cut answer as to what I predict (which I know goes against your ground rules). I think a lot depends on the timing of the launch. Prosser thinks we'll see a WWDC announcement. Kuo and Gurman don't.

So, given this, my thoughts are this:

- If we're seeing a WWDC announcement, then I'm going with the prediction that we'll see an M1X for the higher-end MacBook Pros. It seems like it's too early for M2, given Apple's penchant for annual release cycles on for its SoC families. If, for some reason, in this case, we do not see a redesign (in the case that isn't quite ready to go just yet) or an M2/X, I'd guess that we'd only see M1X in a 16-inch MacBook Pro and that the eventual M2/X MacBook Pros will give us our redesign and our 14-inch MacBook Pro to finally replace the 13-inch. But that's only because there may not be enough in an M1X over the standard M1 to justify an entire refresh to the M1 13" MacBook Pro (other than more ports and storage).

- If we're NOT seeing a WWDC announcement, then I'm going to go with the prediction that we'll likely see a late Summer or early fall release with the redesign and the M2, which will have improved upon the I/O and RAM limitations of M1 as well as faster/better cores (and not just a higher core count all around). I'm leaning this way anyway, but it seems a little early for M2 seeing as Apple seems to enjoy annual cycles for its SoC families.

(leman, if you need a tie-breaker for me, put me in the M2 camp.)

Apple has shown that they're completely okay with replacing pretty much every Intel Mac that isn't a high end 2018 Mac mini, a 16" MacBook Pro, a 27" iMac, or a Mac Pro with the M1. Hell, I'm still unsure as to what the reasoning for keeping the Ice Lake 4-port 13" MacBook Pro around was (other than Apple being afraid that those that need the small machine also need 32GB of RAM, 4TB of storage, and 2 extra ports).

I do think your notion of there being a separate SoC family for higher-end Macs is very possible, but that it will likely be reserved for the Mac Pro and the larger iMac and not any MacBook Pro family device...at least not initially. A lot also depends on whether or not Apple wants to bring the smaller sized MacBook Pro back in line with the larger one in terms of the performance disparity, or whether the smaller MacBook Pro is always going to be treated as its own prosumer device in and of itself. If they separate the 16-inch from the 13-inch/14-inch MacBook Pro as a separate device with its own demands (as was necessary during the Intel era), then I could see it also getting that higher-grade SoC and there being two distinctly different SoC families (one for 27-30" iMacs, 16" MacBook Pros, and Mac Pros, and one for everything else, give or take the iPad Pro). It's hard to say though.

The higher-end of this transition has always seemed to be the bigger mystery.