That is something I don’t find strange at all. The 13gen is the same as 12gen, just with higher clocks/power consumption and rebalanced core configuration. On desktop, these changes allow it to get some extra performance (mostly at the cost of higher power usage, but mobile is much more thermally constrained. It’s the same we saw with later Skylake iterations, where high end mobile CPUs barely showed any improvements since they were already running way outside any reasonable thermal range.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
M2 Pro and M2 Max will be disappointing?
- Thread starter Zest28
- Start date
- Sort by reaction score
That is something I don’t find strange at all. The 13gen is the same as 12gen, just with higher clocks/power consumption and rebalanced core configuration. On desktop, these changes allow it to get some extra performance (mostly at the cost of higher power usage, but mobile is much more thermally constrained. It’s the same we saw with later Skylake iterations, where high end mobile CPUs barely showed any improvements since they were already running way outside any reasonable thermal range.
Some of this is on Apple, for being secretive and simply not providing the early roadmaps like their competitors. There are competitive reasons for that, but it lets speculation get out of control. Also, I am concerned about the engineering behind Apple Silicon, the pace of advancement, but that's a different matter. Some of this is on leakers, who have bad sources and try to whip up Mac users for their own reasons.
However, I think most of this is on us. We tech nerds continue to dream up all sorts of exotic solutions, while Apple appears to be following a staid, tested, predictable, and yes, boring approach. The M2 is based upon the A15, the M2 Pro/Max/Ultra is based upon the M2, when the M2 "Extreme" for the Mac Pro is released, it will probably be a quad die M2 Max. They will all be based upon "Avalanche" and "Blizzard" without significant modification. Once the basic architecture is in-place for the A-series generation, it trickles up to all of the higher performance SoCs. Apple isn't going to step on the gas more than what gets released with the base M(x) SoC.
All of these fancy solutions that we keep fantasizing about don't match the pattern. Apple isn't going to separate their performance desktop and laptop lines, like you (and I) wish. When the Mac Pro gets released, there are going to be some very unhappy folks around here, because it's likely to follow that same pattern. It's the mediocrity principle applied to engineering. There's nothing wrong with that, it's just not something that we can get excited about. I've never had high expectations, because then I'd just be getting my hopes up for something that may not come.
We keep being disappointed because we want something more performant, more interesting, more imaginative than what Apple has thus far released. I'm open to being wrong, in fact I hope I am wrong because I want something different and delightful, but I've seen nothing but a wasteland of predictability, which isn't what we tech nerds desire in our bones.
Here's another one the same guy ("iro") posted a couple hours later. Not that we should believe this one either, but it doesn't have the issue with the MC crypto score. I check their GB posting history, and they did post what appear to be legit scores for the M1 Ultra on March 8, which is 10 days before it was released (https://browser.geekbench.com/user/iro and https://browser.geekbench.com/v5/cpu/13345054). But they also have some strange postings, like a 2019 Intel Mac Pro with an i9-13900KF (https://browser.geekbench.com/v5/cpu/18486658). Not sure what that means--that they can spoof postings, or that they also build Hackintoshes.
Mac14,6 - Geekbench
Benchmark results for a Mac14,6 with an Apple M2 Max processor.
browser.geekbench.com
It's not like our expectations are completely outlandish. It's about trying to understand the customers interests, industry capabilities and ultimately guess Apple's strategy. From a certain perspective, splitting the consumer and prosumer silicon makes sense. There is also the practical problem of powerful desktop, which is something Apple can choose (or not) to address. And of course, we have various modern capabilities like SVE or real-time raytracing where action is needed.
Apple only just began rolling out their hardware and I don't think we've seen enough to establish a long-term pattern. Not to mention all the challenges due to the pandemics, economy, political instability etc. We will see where all of this is going. One thing is clear though: they are playing the long game and avoid rushing into things. But step by step by step the technical superiority of their approach is becoming apparent.
Even if the M2 MAX would double the speed or be 4 times faster, it will be still very much slower.
Agreed, it really is striking just how poorly-coded Stockfish is for Apple Silicon, and thus how ill-conceived it would be to use it as a comparative benchmark. Kind of like assessing a Ferrari Daytona by driving it on a poorly-maintained dirt road.
That is what you meant, right?
Last edited:
No, but I think there's a lot of speculation which simply doesn't have any basis on the limited data that we do have. I didn't say that your ideas (or anyone else's) are wrong, just that I'm not seeing any indication of them. (And to be clear, @leman, I've always highly respected your opinions and thoughts.)
No, we don't, but working with "maybes" and "what ifs" isn't going to get us anywhere, beyond fun speculation. You said that you were disappointed that Apple hasn't met your expectations, and I sympathies with that, so I am being cautious with further expectations.
Right now, all we have to go on is the cadence of the M1 series, the leaks from Gurman about the M2 "Extreme", and our reliable leaker @Amethyst, who leaked the Mac Studio before it was announced, who also backs up Gurman. Now we have alleged Geekbench results which apparently show the M2-series continuing the same cadence as the M1. It's not yet a pattern, but definitely appears to be the beginning of one.
I don't give Apple a pass on this because their competitors are executing well under the circumstances. It's hard to look at AMD and say that they aren't performing remarkably. If a much smaller company like AMD can perform as they have, with far fewer resources, then there is no excuse for Apple. If the more advanced M2 chips are as pedestrian as they appear to be, then they should have already been out months ago.
Regardless, I've always thought that the M3 generation is going to be the most telling. If the P-cores haven't seen any improvement by the time it is released, then Apple is in trouble. If anyone reads my post history, then I'm clearly not a doomsayer, and I'm not ready to count out Johny Srouji and his team. I'm just being cautious about my expectations until proven otherwise.
I think even the least impressive 14-nm era generations had at least a ~2% SC improvement over the previous ones. But I was also under the impression that Alder Lake put out everything Intel had at the moment so I wasn't expecting another big improvement soon, so maybe it makes sense.
The A16 Bionic got a +8% single core performance bump over the A15 Bionic. A similar bump over the M2 would put the M3 at ~2050 single core Geekbench points, so Apple is in the clear for the M3. But the A16/M3 didn't even have a significant P-core redesign. Instead, it's the A17/M4 we should be looking for to see if Apple is still capable of delivering significant improvements to the P cores.
I'd expect the M3 to be on N3, which (inferring from Anandtech's tables*) should give it 5%–10% over N4P => 2150 – 2250 (1900*1.08*1.05 = 2150; 1900*1.08*1.1 = 2250). It's also possible that, for the desktops, they might allow the clocks to go a bit higher.
That's interesting about A16's P-cores. Perhaps the architectural focus for the M3 will be on the GPU (hardware RT).
*They say N5P (which I assume is what's now called N4P) should be 5% higher in perf. than N5, and that N3 should be 10%–15% higher than N5 => 5%–10% higher than N5P = N4P.
TSMC Roadmap Update: N3E in 2024, N2 in 2026, Major Changes Incoming
Last edited:
I know, which is why I am concerned. I suppose Apple could use the A17 for M3, but I doubt it, with their current cadence. The M3 is going to be competing with Zen 5, which does get an architectural overhaul. Meteor Lake seems less impressive.
My concern has always been the same as @leman, in that Apple hasn't yet shown the desire or ability to compete in the performance desktop category, which is the one I care about. I don't own a laptop, they are of no use to me, being the stationary sod that I am. The vast majority of Macs sold are laptops, I don't want us desktop users to be left behind, by default.
I'm not saying that's going to happen, just that in terms of raw performance, that is my primary concern. We still haven't seen the final desktop Mac line, with the high-end Mac mini and Mac Pro still stuck on Intel, but I don't see any reason that they won't simply scale the same as the Geekbench M2 leaks that we have.
The M1 Ultra is an impressive chip, but that ridiculous chart comparing the Ultra GPU to a 3090 was outright embarrassing. I don't know what Apple's PR team was thinking with that one. So, I don't have differing thoughts on the issue compared to everyone else here, I'm just not nearly as optimistic that Apple will shoot for the performance crown. I want faster GPU, P-cores, ray tracing, and strawberry pancakes, but I'm not sure Apple can deliver compared to the competition, with what limited information I have, thus far.
I'm with you. I also consider myself principally a desktop user (I use both, but it's typically a desktop I use for serious work, and thus it's desktop performance that's of most interest to me).
Consequently, as an Mac customer, I'm used to looking longingly over the fence at the higher performance capabilities of high-end PC desktops. Apple Silicon has turned the table on this for laptops—in that category, it's now PC's that trail in performance. You and I would love that to be the case for AS desktops as well, but it remains to be seen if that will happen, and I share your concern it won't, since Apple's focus is (understandably because of market share) more on the mobile sector.
I think the essential question is this: Now that Apple has some experience with the M-series chips, is it willing to undertake a separate design effort, for the desktops, to create performance cores optimized to run at higher clocks? Who knows—maybe that's what they've been working on for the Mac Pro and, if so, perhaps this will trickle down to the Studio.
Even as a Mac die hard, I have to admit that a Zen 4 + RDNA3 combo would be an excellent desktop machine, both in performance, relative power consumption, and comparative price point. I wouldn't say that Raptor Lake and Lovelace are bad, but I think AMD has found the correct balance between efficiency, price, and wattage.
I don't have the exact statistics, but the last time I researched it, which was admittedly a few years ago, roughly 85% of Macs sold were laptops. I may be off a bit, but probably not by much. My concern is that Apple will only be targetting that market segment from here forward. The M-series will be, first and foremost, designed for the MacBook Air, all other priorities rescinded.
Also, I realize that I failed to define what I meant by "exotic solutions" in my post above, which I should have clarified. I was speaking of the Mac Pro. We've seen a number of, shall we say, unique notions of what it will be, despite Gurman and @Amethyst both essentially corroborating that it's going to be a quad M2 Max. Given the new Geekbench leaks, that suggests that the M2 "Extreme" will have the same single-core performance as the base M2.
This is not good, and I'm not happy about it. I'm not in the market for an Apple Silicon Mac Pro, and never will be. However, I will eventually want either a high-end Mac mini or Mac Studio, which will likely share the same P/E/GPU cores as the Mac Pro, just fewer of them.
@leman's hope that Apple would add some juice to the higher-end M2 SoCs wasn't outlandish (I had unintentionally conflated that with the Mac Pro speculation) which is the part that I find unsettling, as someone who plans to eventually replace my Intel Mac with a performance Apple Silicon desktop. I want my next Mac to either meet or exceed the capabilities of high-end x86 machines. One of the big selling points of Apple making its own chips is that they could beat PCs across the board while requiring less wattage. That fanciful graph showing the M1 Ultra on the same level as the 3090 shows that Apple has been pondering it, or at least their PR team has.
My primary concern, at this point, is that the engineers don't share that mindset, and despite Jony Ive being no longer with the company, "thin and light" is still the mindset within our favorite fruit company. It's still early in the Apple Silicon saga, we're still waiting on the transition to finish, but I find everything I have seen thus far to be disquieting.
Even worse, IMO, was their decision to claim the XDR equaled or bettered a Trimaster, since it can't do the basic thing the Trimaster is designed for (serve as an HDR Mastering Monitor). I.e., it wasn't merely wrong in degree, it was wrong in kind.
It's like marketing a camera to scuba divers and saying it's superior to the Nikonos, when it can't even be used underwater.
This seemed especially boneheaded, because they worked so hard to gain back credibility with the pro video community, and then applied egg directly to their face. They were roasted for it, and rightly so, on some of the pro video forums.
For more details, see:
Something else to consider:
It seems the only reason AMD and Intel can beat Apple in SC desktop speeds is because they offer a much larger percentage "turbo boost" over their base clocks than the M-series chips do—93% for the i9-13900K and 27% for the Ryzen 9 7950X, compared with 7% for the M1 (based on https://www.anandtech.com/show/17024/apple-m1-max-performance-review )
Assuming the M2's turbo boost (max clock/base clock for P-cores) is the same 7% as the M1's, here's what the top chips from the big three would look like if they all had the same 7% boost as the M2:
SC GB scores (assuming linear relationship between SC score and clock speed)
i9-13900K: 1,230 @ 3.2 GHz
AMD Ryzen 9 7950X: 1,730 @ 4.8 GHz
M2: 1,900 @ 3.5 GHz
Here's how the M2 would compare to the actual Intel and AMD chips if we allowed it a 27% boost
i9-13900K: 2,227 @ 5.8 GHz
AMD Ryzen 9 7950X: 2,192 @ 5.7 GHz
M2: 2,250 @ 4.2 GHz
So why couldn't Apple implement a 27% boost over their base clock, like AMD does? Are their cores not designed to handle the needed increase in voltage? And, if so, could they be?
Assuming that power is quadratic with clock speed, this would increase power consumption for the turboed core by ~40% over what's currently used. I don't know what the max watts per core is for the M2's P-cores, but if it's, say, 5 W, then that would only be another 4 W to allow two P-cores to be boosted to 4.2 GHz, which seems insignificant for a desktop. If it's cubic, it's an additional 8 W for two cores. Granted, it could be exponential or follow some other functional form..
We can do the same calculation for the M3 on N3. The clock speed increased by 9.4% from A15 to A16, so I'll use the same % increase for M2 to M3. Then if we add a 7.5% increase in performance for going from N4P->N3, we get:
M3: 2,650
@ 4.6 GHz
It seems the only reason AMD and Intel can beat Apple in SC desktop speeds is because they offer a much larger percentage "turbo boost" over their base clocks than the M-series chips do—93% for the i9-13900K and 27% for the Ryzen 9 7950X, compared with 7% for the M1 (based on https://www.anandtech.com/show/17024/apple-m1-max-performance-review )
Assuming the M2's turbo boost (max clock/base clock for P-cores) is the same 7% as the M1's, here's what the top chips from the big three would look like if they all had the same 7% boost as the M2:
SC GB scores (assuming linear relationship between SC score and clock speed)
i9-13900K: 1,230 @ 3.2 GHz
AMD Ryzen 9 7950X: 1,730 @ 4.8 GHz
M2: 1,900 @ 3.5 GHz
Here's how the M2 would compare to the actual Intel and AMD chips if we allowed it a 27% boost
i9-13900K: 2,227 @ 5.8 GHz
AMD Ryzen 9 7950X: 2,192 @ 5.7 GHz
M2: 2,250 @ 4.2 GHz
So why couldn't Apple implement a 27% boost over their base clock, like AMD does? Are their cores not designed to handle the needed increase in voltage? And, if so, could they be?
Assuming that power is quadratic with clock speed, this would increase power consumption for the turboed core by ~40% over what's currently used. I don't know what the max watts per core is for the M2's P-cores, but if it's, say, 5 W, then that would only be another 4 W to allow two P-cores to be boosted to 4.2 GHz, which seems insignificant for a desktop. If it's cubic, it's an additional 8 W for two cores. Granted, it could be exponential or follow some other functional form..
We can do the same calculation for the M3 on N3. The clock speed increased by 9.4% from A15 to A16, so I'll use the same % increase for M2 to M3. Then if we add a 7.5% increase in performance for going from N4P->N3, we get:
M3: 2,650
@ 4.6 GHz
Last edited:
Exactly this. We don’t know how these scale with frequency or power, however. I agree there likely is headroom tho for desktop applications with better cooling to increase clocks on the same architecture.Something else to consider:
It seems the only reason AMD and Intel can beat Apple in SC desktop speeds is because they offer Turbo Boost:
i9-13990K: 2227 SC @ 5.8 GHz Turbo => 1,150 SC @ 3.0 GHz base clock*
AMD Ryzen 9 7950X: 2192 SC @ 5.7 GHz Turbo => 1,730 SC @ 4.5 GHz base clock*
*Assuming linearity
So why couldn't Apple do something like this for their desktops?
current M2: 1899 SC @ 3.49 GHz base clock => 2,290 SC @ 4.2 GHz Turbo (45% increase in SC power consumption for Turboed core, assuming it's quadratic with clock speed)
M3 on N3: 2200 SC @ 3.8 GHz base clock* => 2,600 SC @ 4.5 GHz Turbo
*clock speed increased by 9.4% from A15 to A16, so I used the same % increase for M2 to M3.
I mean TSMC N5 is hitting almost 6Ghz with Zen 4. Intel 7 (10nm++) is hitting 6Ghz with the 13900KS….
Why can’t M2 super duper ultimate hit 6Ghz with workstation class cooling?
That is the big question. There has been speculation that Apple has, thus far, optimized for a specific clock frequency and/or voltage, thus limiting headroom. Exactly the reasons for this, and if it's even true, are unknown. It makes me wonder if the A16 was designed with higher clocks in mind, it's just unnecessary with an iPhone, because battery life is king. This would be similar to what AMD has done with Zen 4, which is mostly a juiced version of the previous architecture.
Even without new P-cores, I think the M3 will be where Apple either shows us what they can do, or loses the plot. They'll have the first two generations under their belt, hopefully have learned lessons from those experiences, and made any corrections necessary.
That is, assuming Apple thinks they even need to compete with performance PC desktops. It may not be that they lack the talent to do so, but simply don't have an interest in participating in that segment. That's what concerns me the most.
Apple really has not sold the notion that they were building the broadest line up of SoC that beat everything. Cherry picking one single benchmark to throw at the 3090 is not a declaration of war to sell everything to everybody. Just small scale playing the same game that Nvidia/AMD always play when they do releases. Instead Apple picked out one corner where unified memory has at least as much leverage on the problem as computational grunt. What was really being shown there was that the Ultra Perf/power curve was steeper (better perf/watt) than 3090 longer, flatter pref/power curve. Not that Apple was aiming for total replacement.
"There is some hefty performance in the Ultra's GPU if properly access it" is the messaging there. Not the "king is dead, long live the king". That is the hype-train that tech porn press at that graph. Yes, Apple laid that out there as a fanboy trap to latch onto for the "Apple take my money fast' crowd.
Apple chops off their SoCs before they get to the diminishing returns zone. I think the bigger disconnect is the folks expectating Apple sot push their SoC way deep off the edge of the curve that the underlying fab process allows for. I doubt Apple is going to go there. There is very little upside (other than chasing bragging rights) for them to go down that rabbit hole. It doesn't help the entire rest of the line up.
Really doesn't help the high end Mac desktop market much either. In the general PC market there are dozens of vendors offering various cooling alternatives that try to 'undo' what going far out on the diminishing returns curve does ( 'undo' the tradeoff). Previous Mac Pro's didn't generally have 3rd party coolers.
Johny Srouji has gotten up and preached a pref/watt sermon at every single M-series product introduction. He is in charge of all the chip development. You think he is going to radically change religions. During one of those intros Apple said something to the effect of " they wanted to the leading iGPU implementation". So they want to be very competitive with dGPUs but mainly in changing the range of iGPU. Removing the tech press implied implication that iGPU means 'slow'.
The Ultra GPUs complex doesn't cover everthing that the 3090 covers but a "double Ultra" would cover much higher percentage. That high perf/watt curve means there is a straightfoward path to making something bigger.
[ similar with ray tracing. Later to market with a maximum perf/watt ray tracing is probably a path Apple would be on . That would work for their AR and/or VR tether-less google/glasses products also. ]
Similar with the P core complex. It doesn't run so hot that have problems scaling it up to 40-50 core ranges without lots of hassles with about zero loss in single thread performance. ( go look at top end Eypc , Xeon SP and typically see a single thread performance hit. The M2 and M2 Extreme both having the same single thread performance is actually hard to do at reasonable price points, if you have a small-midrange target too far down the dinimishing returns zone for 'maximum turbo'. ) . Apple is largely making a single user workation SoC where still have to have 'snappy' GUI stuff on individual, primary single focus programs.
What is Apple is woefully missing "desktop SoC" wise is relatively very high end general I/O aggregate bandwidth ; not GPU or GPU core hot rod features. The Mac Studio conveniently covers up that problem.
A major factor to Apple having a leadership position in the Phone SoC is because they were not tryin to make every SoC for every phone. Qualcomm (and others ) blew off Arm64 because phones didn't need 64 pointers to large RAM footprints. Apple mainly looked at it as a inflection point to clean up the instruction set and move on to better SoC for their relatively low RAM requirment iPhones. (wasn't about rapid expansion of RAM max capacity at all.)
The mindset of Apple is coming to crush the classic HEDT market metrics from 3-8 years ago. Probably not. It is still going to be probably driven by making better Macs than in making matches to mainstream PC products (and that exact breadth of offerings) . Apple is probalby going to take incremental single threaded 'wins' where they fall out of the overall improvements being made. But it doesn't seem likely that they will throw other aspect growth (e.g., effective 'bang for buck' accelerators, new GPU throughput wins , etc) 'under the bus' just to one relatively narrow area (e.g., single threaded).
The indications thus far are as you describe--that Apple doesn't care about achieving super-high ST perf. But I'd argue against the claim that this is a "narrow area"; indeed, I'd say it's quite the opposite: For most users, ST CPU perf is the single most important metric when it comes to how fast a computer feels.
E.g., take the M1 Pro MBP, and consider these four potential mods:
1) double the ST CPU speed (while ensuring RAM speed is increased sufficiently to keep up)
2) double the GPU speed
3) keep ST CPU speed the same, but double MT speed by doubling the number of cores
4) double the SSD speed
Which of the four would be most likely to bring a smile to the face of the typical user? Which would you buy? I think for most, overwhelmingly, it would be #1. In fact, if you want, you can already get #2 by buying an M1 Max, and #3 by buying an M1 Ultra. As for #4, other than for specific operations, I don't think the difference would be that noticeable to most users.
Last edited:
Pretty sure you typo'd this - starting from a baseline of a 10-core M1 Pro, you'd achieve #3 by buying M1 Max and #2 by buying M1 Ultra.
Ah, yes, I did indeed reverse them. Thanks for catching that. I'll correct my post.
Performance wise I am satisfied, I just hope Apple can put two more Efficient Cores in M2 Pro so that 14-inch MacBook Pro will have better battery life.
Rumors say the M2 Pro & M2 Max SoCs should have 12-core CPUs, eight performance & four efficiency...
Even with 2 efficiency cores, the battery life under lightweight casual usage should be better as the efficiency cores is where most of the improvements are with M2 relative to M1. But I hope people don't buy the M2 Pro and M2 Max chips for lightweight casual usage as that is not what it is intended for.
Under heavy usage (which is the use case of the M2 Pro and M2 Max chips), the battery life might be worse as the M2 is a power hungry chip relative to the M1 chip. So the same might apply for the M2 Pro and M2 Max also.
The problem is Apple's pricing. With similar price you can get much more from 14' MacBook Pro, compare to the new MacBook Air. So I guess many people are willing to pay a bit more, even they don't actually need such a powerful system.
The 14' MacBook Pro lasts up to 11 hours wireless web surfing, when 13' Intel-based MacBooks can last 10 hours. Whereas the new MacBook Air and 16' MacBook Pro can last 14-15 hours in wireless web surfing, which is obviously superior. The difference is huge.