M2 Pro and M2 Max

[dada_dave]

I wonder what the connection looks like for the Max and what RE can tell us about the max Maxes that can be put into an SOC.

 

[Cmaier]

I wonder what the connection looks like for the Max and what RE can tell us about the max Maxes that can be put into an SOC.

Yeah, saw a full photo of an unpackaged Pro, but not a Max yet.

 

[Andropov]

Ah, I didn't even see the 30-core M2 Max, sorry. I was looking at the 16" model.

You can get it on the 16" too, if you upgrade on the the base models with M2 Pro. But it's kind of an obscure SKU. I only knew it was there because I saw a footnote somewhere stating that only the 38-core Max could upgrade to 96GB.

Btw, the M2 Pro no longer is a "chopped" M2 Max. The GPU cores are slightly rearranged in the die shots of the M2 Max. The upper half of the M2 Max has 20 GPU cores. EDIT: Scratch that.

EDIT: I can't pinpoint the Neural Engine either, but it looks like this time around the Max does not have two of them.

 

[Nycturne]

Fully loaded M2 Pro Mac Mini costs the same as M1 Max Mac Studio?

Advice to you since you are my friends: get the studio

Yeah, the overlap at the high end of the Mini is real.

That said, a M2 Pro Mac mini is still interesting as a home VM server. But I probably have to wait a bit longer before I can make the jump. Need to do some testing with Win 11 VMs on ARM to see how well it can handle hosting dedicated servers (Astroneer, Valhalla, etc) that are still x64. That's the main reason I still have an i7 Mac mini hosting VMs in my network closet.

 

[Andropov]

EDIT: I can't pinpoint the Neural Engine either, but it looks like this time around the Max does not have two of them.

I tried to annotate the die shot of the M2 Pro (it looks a lot like the M1 Pro though, except the GPU core arrangement).

 

[Cmaier]

This should surprise nobody (other than, perhaps, digitimes):

Kuo: 2024 MacBook Pro Models to Feature 3nm M3 Pro and M3 Max Chips

Though Apple just refreshed the 14-inch and 16-inch MacBook Pro models this morning, Apple analyst Ming-Chi Kuo is already sharing details about...

www.macrumors.com

 

[Jimmyjames]

What do we think these updates mean for the update cycle?

Iirc it's been just over a year since the M1 Pro and Max. Do they now keep to a yearly cadence? That would be great. I'm seeing some say M3 Pro/Max in Oct/Nov 24! That seems like a long time.

 

[Andropov]

What do we think these updates mean for the update cycle?

IMHO the M2 Pro/Max was due for release on Oct/Nov 2022 and something delayed it. I would expect the M3 Pro/Max to arrive late 2023 or early 2024. Certainly not Oct/Nov 2024.

 

[Cmaier]

What do we think these updates mean for the update cycle?

Iirc it's been just over a year since the M1 Pro and Max. Do they now keep to a yearly cadence? That would be great. I'm seeing some say M3 Pro/Max in Oct/Nov 24! That seems like a long time.

I think we are likely looking at a more or less annual cadence, at least to the extent that TSMC sticks to its roadmap and we don’t have any other major commercial disruptions. War, famine, plague, etc. can all throw everything into limbo again.

 

[Yoused]

Advice to you since you are my friends: get the studio

If you want high performance, the time to wait is now.

 

[Jimmyjames]

IMHO the M2 Pro/Max was due for release on Oct/Nov 2022 and something delayed it. I would expect the M3 Pro/Max to arrive late 2023 or early 2024. Certainly not Oct/Nov 2024.

That's what I was thinking

I think we are likely looking at a more or less annual cadence, at least to the extent that TSMC sticks to its roadmap and we don’t have any other major commercial disruptions. War, famine, plague, etc. can all throw everything into limbo again.

Makes sense.

I've also seen some claim Apple will skip the A16 for the M3. I would be surprised as this would add a longer time between updates. It seems (as clear as can be with only two iterations) that they will be quite methodical with Apple Silicon: Phone SoC in September, base M series chip to follow the following June - September, then Pro/Max Nov - Jan.

 

[Jimmyjames]

I feel a great disturbance in the force, as if a million Cinebench runs were being put forward as a means of discrediting Apple Silicon!

I"m braced.

 

[Cmaier]

That's what I was thinking

Makes sense.

I've also seen some claim Apple will skip the A16 for the M3. I would be surprised as this would add a longer time between updates. It seems (as clear as can be with only two iterations) that they will be quite methodical with Apple Silicon: Phone SoC in September, base M series chip to follow the following June - September, then Pro/Max Nov - Jan.

Hard to say. I tend to think that Apple is entering a phase where it isn’t particularly interested in increasing performance of iPhone anymore (at least not the general compute cores), and is more concerned with power consumption/battery life.

We may be coming to a new paradigm where the cores are not the same microarchitecture between M- and A-series chips. (or maybe M-series efficiency cores become A-series performance cores, or who knows!)

It will be very interesting to see how things progress.

 

[leman]

I've also seen some claim Apple will skip the A16 for the M3. I would be surprised as this would add a longer time between updates. It seems (as clear as can be with only two iterations) that they will be quite methodical with Apple Silicon: Phone SoC in September, base M series chip to follow the following June - September, then Pro/Max Nov - Jan.

Probably also depends when they make these decisions. I imagine that designing a chip takes a while, so if Apple indeed decides to skip A16 (which would make sense if they target N3, no?), that decision would need to be made a while ago.

Hard to say. I tend to think that Apple is entering a phase where it isn’t particularly interested in increasing performance of iPhone anymore (at least not the general compute cores), and is more concerned with power consumption/battery life.

We may be coming to a new paradigm where the cores are not the same microarchitecture between M- and A-series chips. (or maybe M-series efficiency cores become A-series performance cores, or who knows!)

Or maybe this means that A17 will be designed with high-perf in mind and then severely under-clocked to be more efficient? Of course, the big question is which decisions Apple takes to increase performance (if any).

 

[Cmaier]

Probably also depends when they make these decisions. I imagine that designing a chip takes a while, so if Apple indeed decides to skip A16 (which would make sense if they target N3, no?), that decision would need to be made a while ago.



Or maybe this means that A17 will be designed with high-perf in mind and then severely under-clocked to be more efficient? Of course, the big question is which decisions Apple takes to increase performance (if any).

Underclocking a high-performance design isn’t a great way to design a power-efficient part, but who knows.

 

[Andropov]

Hard to say. I tend to think that Apple is entering a phase where it isn’t particularly interested in increasing performance of iPhone anymore (at least not the general compute cores), and is more concerned with power consumption/battery life.

We may be coming to a new paradigm where the cores are not the same microarchitecture between M- and A-series chips. (or maybe M-series efficiency cores become A-series performance cores, or who knows!)

It will be very interesting to see how things progress.

Do you think it would make sense for Apple to go for a 3-tier CPU core desing? The E-cores used to be the Apple Watch main cores. These last two years Apple has put significant effort in making the E cores faster, but I can't help but notice that the Apple Watch SoC hasn't got any upgrades in these two years. I wonder if the E-cores may be too power-hungry for the Watch SoC now and whether it would make sense to have an even lower power core design.

Android phones kinda have slightly lower power cores, although much slower. Apples current E cores are closer with every iteration to the performance (but not power consumption) of Android phones middle cores.

 

[Cmaier]

Do you think it would make sense for Apple to go for a 3-tier CPU core desing? The E-cores used to be the Apple Watch main cores. These last two years Apple has put significant effort in making the E cores faster, but I can't help but notice that the Apple Watch SoC hasn't got any upgrades in these two years. I wonder if the E-cores may be too power-hungry for the Watch SoC now and whether it would make sense to have an even lower power core design.

Android phones kinda have slightly lower power cores, although much slower. Apples current E cores are closer with every iteration to the performance (but not power consumption) of Android phones middle cores.

It does feel to me like they need 3 tiers of core just to cover their entire product range. I don’t think they would need 3 types of core in a single chip, though (I.e I don’t think they need a middle tier on M-silicon) - to be sure, though, I would have to do a lot of analysis of actual software loads and I don’t want to do that

 

[Colstan]

Fully loaded M2 Pro Mac Mini costs the same as M1 Max Mac Studio?

Advice to you since you are my friends: get the studio

I had always had my eye on the base model Mac Studio featuring the Max SoC, while ticking the box for the upgraded GPU cores for an additional $200. Assuming this general cadence with Apple Silicon holds in the future, then I'll likely be getting an M3 or M4 generation of such a model. That's assuming that Apple regularly updates most of the Mac line with current generation chips. There's some chatter that the iMac won't be getting the M2, and jump straight to the M3. If Apple doesn't release an "Extreme" version for the upcoming Mac Pro, then I could see them keeping the Mac Studio on the M1 generation, so that marketing can concentrate on the M2 Ultra being only featured within the Apple Silicon Mac Pro.

The M2 Pro Mac mini is appealing as a concept, but the pricing pushes those on a budget toward the base model, while the Mac Studio is more attractive for performance desktop users.

My current 2018 Mac mini is now feeling alone, with the last of its brethren having finally met the executioner's axe. Now we wait for the Apple Silicon Mac Pro, while the old Intel model is left forlornly wondering when its turn at the gallows shall arrive. If Gurman is correct, then we can expect it in March, but I suppose it could arrive as late as WWDC. If nothing else, it will be good to finally have the transition complete, and most of the speculation will then become centered around the future of the M-series, rather than exotic solutions for the Mac Pro.

 

[tomO2013]

Has anybody checked on trade in allowances from Apple.....

Parking aside the sanity of even looking to trade in a 16" M1 Max with 64GB ram and 4TB storage......

This was a laptop that cost me over CAD $6300 in 2021.

Fast forward to today.... apple will offer me $630 for the same laptop.

I'm pretty sure when I traded my 16" i9 intel MacBook Pro for the M1 Max, they offered me a significantly higher trade in discount relative to the original purchase price.

I'm shocked - literally shocked at just how crappy the trade in allowance is from Apple.

 

[tomO2013]

On a side note I'm resisting to the urge to go and check on the feedback to M2 Pro and M2 Max over at 'the other place'.

I'm gonna leave it to thicker skinned folks like leman to try and bring some balance of reason to the Apple silicon forum over there....


"But it still doesn't run crysis as fast as my intel DinoLake chip with neon blue lights....

Anybody willing to have a game of fishmongers chess benchmark where my rtx4090 spanks the new M2 Max.

Intel never gonna give you up, never gonna let you down, never gonna run around (with AMD) and desert you"