How long can Apple stay ahead in CPU performance?

[BootsWalking]

Yes, Intel can beat it in Multicore.

Agree with everything you wrote but wanted to point out the Alder Lake i9 slightly beats all M1 variants in single-core as well.

 

[cnnyy20p]

The M1 series max out at 3.2GHz (even with more ipc) while other CPUs in the market have to clock up to 5GHz just to beat M1.

I don't agree with Apple stayed with Intel alongside Apple Silicon. It would end up like Windows on ARM where few apps are being ported natively. Even if Apple Silicon is superior.

 

[Mcckoe]

Really these chips are plenty fast to do what we need them to do… instantaneously(to humans) compute anything to the resolution of the given display…. The problem is people want more… they don’t just want a system that can play amazing native games, they want it to be able to play any game, at any resolution, anywhere. People don’t just want to take and edit video, they want to take and edit extreme resolutions at insane frame rates… and that’s great, because we have gotten to a point that you can do a lot of that… But, at some point you have to ask yourself… how much more powerful does my phone really need to be? Does my Tablet really need to be able to render 12K footage in real time? If the only advancements left: are in gaming and hi-resolution graphics works, why not just upgrade a system for that, and keep what we have for everything else?

 

[jav6454]

That chart isn’t a fair representation since it doesn’t quantify what “performance” is. Apple prides itself on comparing itself to Intel up to a certain wattage but after that it’s a tough call. The more watts you throw at an Intel chip the stronger it performs.

I could be wrong but I don’t think Apple makes anything that can compete with the top end x86 processors at Single or Multi core tasks. This is primarily due to the power draw.

I think that is crucial point of all this. Yes, Intel performs better, but at what cost? Portability as their chips are power hungry. Intel has good CPUs in the desktop space that still lag behind AMD, but in the mobile space they are getting butchered by Apple and AMD.

Now, as to the question, can Apple stay ahead? If leaderships listens to their engineers in terms of technologies and limitations and if Apple doesn't become complacent like Intel did, then yes.

 

[Realityck]

My pondering thought is when will Apple Mac users regret that Apple moved away from Intel? Maybe when Intel is beating Apple in performance per watt OR could it be that Intel when Intel has a 50% more performant CPUs. I know I would because is that came to fruition then I would be disappointed that if Apple would stayed with Intel I would had access to multiple OS with a wide range of applications just from Dual booting on a Mac.

You can't compare the what was to the what is without looking at how the MacOS is scaling to match the abilities of these new AS SoC's. Looking at this strictly processor versus processor omits the fact that very few vendors actually design complete solutions around hardware/processors and operating systems/software.

Whose really pushing aheads in technology? If Intels improves their CPU's how is that maxed out with existing designs and someone else's OS's?

 

[Kpjoslee]

Apple Silicon will be transitioning to 3nm node in 2023 and will stay on that node until 2026 (since TSMC isn't planning on mass producing 2nm until 2H 2025, which means that the SoCs won't be ready for product til 2026). It means Apple Silicon will no longer be able to enjoy pretty big node advantage like it had with M1 based AS. I expect Apple to lead in terms of Perf/watt for a while, but the lead won't be as big as M1 based SoCs vs AMD and Intel's current lineups.

 

[navaira]

Maybe when Intel is beating Apple in performance per watt OR could it be that Intel when Intel has a 50% more performant CPUs.

Awww, bless their overheated, slowwwwww, energy-gulping, 10nm hearts <3 <3 <3 Poor, doomed Apple will beg Intel for mercy soon.

– sent from my room-temperature-while-fanless, 10hr battery life, 8/8 core M1 Macbook that only ever got warm because Brave Browser's hardware acceleration needed to be switched off

 

[ian87w]

How long? As long as Apple can keep that Srouji guy? ?

Anyway, there are a lot more reasons for Apple to go Apple Silicon than better CPU performance

1. Vertical integration and economies of scale. Apple now have a chip that can be used for macs and iPads. That improves economies of scale and reduce cost, which translates better profits. Cannot do that with intel.
2. Much better integrated GPU performance with media encoders.
3. The fight now imo is no longer in raw CPU performance. It’s now in ML. The neural engine, ISP, etc will be the key parts bringing new features. I don’t even know where intel is in this regards.

Being faster at higher power is meaningless for most consumers today as they are more likely to own a laptop than a desktop. Even then, with inflation affecting everyone, less electronic bill will probably be desirable in the end.

 

[Tyler O'Bannon]

This is a bit of a loaded question. Just like the super car benchmark became power to weight ratio, this is going to start to be power per watt. M chips are awesome. But anybody can beat them if you throw a million watts at it with liquid cooling.

Apple’s chips are fantastic, and their power to battery life and lack of heat in laptops is amazing. And they are going to drive hard that direction.

Looking forward to what their high end desktop silicon looks like over the next several generations (looking at you Mac Pro).

 

[Andropov]

That chart isn’t a fair representation since it doesn’t quantify what “performance” is. Apple prides itself on comparing itself to Intel up to a certain wattage but after that it’s a tough call. The more watts you throw at an Intel chip the stronger it performs.

I could be wrong but I don’t think Apple makes anything that can compete with the top end x86 processors at Single or Multi core tasks. This is primarily due to the power draw.

The performance the chart refers to is the maximum SPECint2006 score of each CPU.

One thing that I think Apple has to improve on, is the GPU and throwing cores into the SOC isn't the answer. They need to be able to compete toe to toe with Nividia and AMD, if they want to grow marketshare beyond content creators and students. Its not that the graphic processing power is poor, but they need to up their game in the up and coming generations.

For the GPU performance, throwing more cores into the SoC actually *is* the answer

 

[Mcckoe]

This is a bit of a loaded question. Just like the super car benchmark became power to weight ratio, this is going to start to be power per watt. M chips are awesome. But anybody can beat them if you throw a million watts at it with liquid cooling.

Apple’s chips are fantastic, and their power to battery life and lack of heat in laptops is amazing. And they are going to drive hard that direction.

Looking forward to what their high end desktop silicon looks like over the next several generations (looking at you Mac Pro).

Not really, the graphs apple is showing is really highlighting the law of diminishing returns… meaning there is no reason the average person needs to ever go beyond a certain wattage point on Apple Silicon for most tasks, as most tasks will be completed before the processor needs to max Out performance.

For labor intensive tasks, that require constant processings: like gaming, video editing, and calculation analysis; these systems while still quite efficient, can’t overcome raw wattage advantage.

Neither system is bad, and advancements in either; generally leads to future advancements in both.

We need to stop pitting these things against each other… unless your a video editor, game developer, or gamer… none of this really matters, and if you do need that power; you already know what you need to complete your Computational goals.

Apple Custom Silicon, is an amazing achievement, and will eventually lead to a completely unified product line for apple. But, that doesn’t mean it will be a one size fits all band-aid for the entire industry…

 

[Bodhitree]

I think it will really depend on what we will see from the M2 and M3 over the next few years. I think Apple has really stretched their chip design facilities by making all these different components for iPhone, Apple Watch, Airtags and so on, and then adding on top making the M1, M1 Pro, M1 Max and M1 Ultra. It was a major design effort.

You can tell by the smaller jump forward for the A15, which had a modest battery life gain but only about 10% speed gain compared to A14 (geekbench 5 of 1700 compared to 1570). The best current Intel cpus rate about 1950 single core, so about a 10% speed difference.

The thing is, will Apple be able to hit their stride and continue delivering yearly or bi-yearly gains?

 

[BigMcGuire]

I had a 15' MBP with an i7 in it and a dedicated GPU and that thing was constantly screaming fans - I burned my legs trying to use it for a few minutes on my lap one time.

I have an i7 32GB 2TB 13' MBP (bought last year). That thing is always burning hot to the touch. I use parallels with vs 2022 and do some light to medium work with it (32GB of ram is almost yellow). CPU fans are always going off and on - I had to put an eGPU on it to keep the fan noise down (sonnet eGPU) because it got so annoying. I would like to think that I fairly easily hit the "limits" of that processor due to fan noise.

This 16' MBP M1 Max ALWAYS has the fans off. I mean... I've never heard them once - not even when gaming. According to iStat Menus I'm not even using 20% of the capability of this CPU at my heaviest of usage times which is very embarrassing. Everything feels instant and is night and day faster - even when using W11 Arm in Parallels. I am constantly astounded. All the while the case is cool to the touch!

From my non-scientific perspective, there's a MASSIVE difference between the two processors. Night and day difference for what I do.

 

[mi7chy]

That chart isn’t a fair representation since it doesn’t quantify what “performance” is. Apple prides itself on comparing itself to Intel up to a certain wattage but after that it’s a tough call. The more watts you throw at an Intel chip the stronger it performs.

I could be wrong but I don’t think Apple makes anything that can compete with the top end x86 processors at Single or Multi core tasks. This is primarily due to the power draw.

Fanboys like that chart since specint easily fit within CPU cache so tend to show overly optimistic performance compared to real world workloads. And, that's only one aspect of the total package so doesn't factor in things like slow storage I/O, broken trackpad palm rejection, etc.

 

[HobeSoundDarryl]

If we're going to use charts created by Apple Marketing to represent Apple vs. Wintel options (raw power), latest & greatest from Apple is ALWAYS going to be "better" than anything else. So if we use that to make such proclamations, Apple will always win such contests.

Else, the easy thing to do is to shift classic arguments about "who has the most power" to "who has the most power PER WATT"... a migration already rapidly taking hold around here so that our favorite company can pretty much always win vs. Intel comparisons. And be sure to throw in doses of questioning any classic benchmark measuring tools unless those plot Apple #1 in select measures... which is the only time they are right and programmed correctly.

Else, if you want to apply objective measures, Wintel platforms are probably going to be more powerful as measured in traditional ways. If you search objective measures now and read/watch them with objective eyes, Wintel has not ceded "most powerful" as measured in traditional ways vs. M1 either... unless we apply "per watt" (as Apple Marketing likes to do too) and/or via only cherry picked measures. For example, one of the best ones that probably applies here is our classic: "but who has the most profitable..." which- I'm sure- Apple will win every time with Mac platforms now too. So just sling that when in a debate if you need Apple to win.

However, all that offered, it is reasonable to consider the "whole" as the proposition now is NOT (raw) powerful chips vs. (raw) powerful chips but chips that run macOS vs. chips running Windows or something else. Sling around digs about burning laps, cooking eggs, and needing nuclear power plants, etc to prop up the rationalizations and then buy the platform that does whatever is best for each buyer in need of a computer.

In my case, I bought Studio Ultra to get the most powerful Mac. Not for one second do I believe Apple (Marketing) performance charts plotting it as much more powerful than "latest & greatest" from Wintel, but I do the bulk of what I do on Macs, so this is the most powerful new Mac I can own right now.

If some of us needs to thoroughly win the "I own the most (raw) powerful computer" debates, we probably need to buy ourselves loaded Wintels and adopt or switch back to Windows. And that's probably NOT going to change with M2 or M3, etc.. unless Apple decides to erode the current "per watt" game by cranking up the processing speeds... as piling in more and more cores can only go so far... and many key measures of "power" is single core vs. single core anyway.

The good news: by the time Apple might crank up the power, some of them nuclear power plants may be available in the refurb store.

My advice: if you like macOS, buy yourself whatever level of power Mac you need when you need a computer. If you like Windows, buy yourself whatever level of power PC you need when you need a computer. This latter group will probably- OBJECTIVELY- have the most powerful hardware as measured in traditional ways. But Mac people will have macOS and a more power efficient computer... AND the most profitable ones for their maker.

Bonus advice: don't worry about this issue. Worrying about what other people think and/or "keep up with the Joneses" is a game you can almost never win... or not for very long when it comes to tech. Buy the computer you need/want now and enjoy it. When it doesn't feel powerful enough for your future needs, buy a new one then. If you have friends who want to brag their super Windows "latest & greatest" is more powerful than your Mac, let 'em do it... knowing they are having to deal with Windows "fun", a much more robust virus battlefield, shorter battery life or much heavier boxes with bigger batteries, etc. Then go back to your not-quite-as-powerful Mac and get your work done with macOS software, enjoy unique benefits within the ecosystem, save that extra $3/month on your electric bill and not glow in the dark from any radiation leaks from the power plant.

 

[exoticSpice]

The performance the chart refers to is the maximum SPECint2006 score of each CPU.

Don't we need to use real world applications to measure CPU perf. Synthetic bench marks hold no weight in my book.

 

[exoticSpice]

I had a 15' MBP with an i7 in it and a dedicated GPU and that thing was constantly screaming fans - I burned my legs trying to use it for a few minutes on my lap one time.

I have an i7 32GB 2TB 13' MBP (bought last year). That thing is always burning hot to the touch. I use parallels with vs 2022 and do some light to medium work with it (32GB of ram is almost yellow). CPU fans are always going off and on - I had to put an eGPU on it to keep the fan noise down (sonnet eGPU) because it got so annoying. I would like to think that I fairly easily hit the "limits" of that processor due to fan noise.

I would say those Intel Macs were stuck on 14nm and that led to bad efficiency. If M1 Max was based on Intel's 14nm node I am 100% sure it would be as loud or even louder than the 2020 Intel i7 13" under load.

The M1 Max is capped at 35 Watts for the CPU and is also on 5nm node.
Intel Meteor's Lake laptop CPUs will be based on Intel 4 which is close to 5nm TSMC node and I really do think that will be the best comparison.

One thing for sure tho M1 has better IPC than Intel's chip and Intel will have to make the biggest redesign since the Core arch for their CPUs. Intel is rumoured to do so in 2025/26 and that is why Intel claims perf/watt king then because it will likely be a big CPU change.

 

[Kelly Jones]

Good take. I 100% can’t go back to a device that makes fan noises randomly regardless of what I’m doing. I love working in my now silent office and being able to listen to audiophile tower speakers to their full potential instead of headphones.

Yep. Now that laptops are fast enough for most tasks, other factors such as fan noise, screen quality, sound quality, webcam quality, battery life, trackpad quality, and keyboard quality are deciding factors for me. I do think that PC laptops are getting better in these respects and are headed in the right direction. That said, the MacBook Pro 16" is the best laptop I've ever owned. However, it is *expensive*.

 

[exoticSpice]

Yep. Now that laptops are fast enough for most tasks, other factors such as fan noise, screen quality, sound quality, webcam quality, battery life, trackpad quality, and keyboard quality are deciding factors for me. I do think that PC laptops are getting better in these respects and are headed in the right direction. That said, the MacBook Pro 16" is the best laptop I've ever owned. However, it is *expensive*.

You are right. Intel needs to improve their noise pollution.

 

[exoticSpice]

Good take. I 100% can’t go back to a device that makes fan noises randomly regardless of what I’m doing. I love working in my now silent office and being able to listen to audiophile tower speakers to their full potential instead of headphones.

Intel's needs to fix their loud fans. AMD got it 80% percent sorted for ultrabooks.

When AMD moves to 5nm TSMC it will be nice to see.

 

[aParkerMusic]

Forever!!!!!

 

[jav6454]

Intel's needs to fix their loud fans. AMD got it 80% percent sorted for ultrabooks.

When AMD moves to 5nm TSMC it will be nice to see.

Intel needs to fix lots of things, especially their power hungry performance. It's hurting them in the long run. Also, AMD is using TSMC for GPUs. Not sure who their CPU fab is.

 

[Argon_]

The only likely situation I see for Intel regaining Apple's business would be ASi fabbed on an Intel process node, were they to match or exceed TSMC. Seems unlikely at the moment, but not impossible. Competition benefits us all.

 

[deconstruct60]

I doubt it had much to do with 5nm node. Sure, the node helps with efficiency, but the gap is just too wide.

Apples design philosophy is very different from Intel or AMD. Apple relies on very wide execution backends and extremely large caches. They start with low power consumption and extend it to performance. Intel does it exactly the other way round.

You make it sound like "5nm" and "extremely large caches" are completely decoupled from one another. Pragmatically, the are not. Using denser fab process allows the cache size to grow larger without causes some losses in other areas. ( e.g., sacrifice core count to have larger caches with a given fixed transistor budget and die size).

Apple can in part get away with making larger caches because they are on a fab process that is denser than their competitors. When Apple's competitors can make caches of similar size than that relative advantage will erode. It won't erode to "zero value add" but it probably won't be as big of a advantage lever as it is now.

For example both AMD and Nvidia GPUs are rumors to substantively boost their L2 caches in the next gen GPUs coming at the end of the 2022. Is that going to instantly give them good iGPUs, maybe not. However, even if Apple catches the 6900/3090 AMD and Nvidia will be 'gone' from those levels.

AMD has already got 3D cache augment technology. When applied gets a decent multithread boost ( thermal constraints make that not a good single threading drag racing augment). But it is clearly illustrative that once AMD gets to put more cache on the same die with N5/N4 that will be a substantive gain.


The "5nm" plays a smaller but still contributing role in the "go super wide execution" design also. It allows provisioning extra wide units that may sit idle on long stretches of conditional and co-dependent code fragments. Some tight computation section in a long loop and it goes 'high traction'. Similar issue with the AMX units. On "Dick , Jane , Spot" normal code it does nothing. However, since have an extra large transistor budget it doesn't hurt much in trade-offs to carry it around on the die.

Therefore it is a bit of smoke that if Apple was on Intel 14nm+++ that the M1 would do what it does on TSMC N7. To hit the same fab manufacturing costs the die wouldn't be much bigger and Apple would have had to made some unit transistor budget trade offs. Would they still have the edge in some areas? Yes. Would they have all of the edge advantages they have over the AMD/Intel implementations? No. The gaps would shrink ( or grow where AMD/Intel are just throwing lots more power at higher clock speeds. )

There is no indication that Intel is any close in reaching M1 in efficiency, so unless they come out with radically new core design that takes lessons from how Apple does things I wouldn’t worry about it. Similarly, Intels fastest enthusiast CPUs are around 20% faster in single core than M1 - while consuming close to 10x power. There is not much spare room left there and it’s not a method you can utilize for laptops (and you can clearly see that premium mobile Alder Lake barely outperforms even the old M1).

But the playing field isn't just about laptops. Apple and Intel mostly sell laptops , but that isn't the whole market. Especially where "single threaded top end performance" is a strong selling point. Apple also was talking 'smack' about how they conquered desktop performance. Can't move the goal posts and crawl back into solely laptop land without a retreat there.

Putting a higher performance GPU on the same memory bus as a CPU core trying to hit "beat everybody" single threaded throughput is a dual edged sword. That is why Apple puts a bandwidth cap on the CPU cores. It is a graphical user interface operating systems so at some point the GPUs 'wins' the limited bandwidth contrast when both sides want "too much".

The GPU less AMD desktop CPUs and smallish iGPU Intel desktop product tip the 'tie breaker' in the other direction (not even possible to 'tie' in the AMD case). So if even up the fab process and level up the caches and push the GPU off the memory bus... the notion that Apple has a 'slam dunk' win there in single threaded isn't clear at all.
Large enough micro-op translation cache and take the quirks of the literal x86 opcodes off the table.

In the general desktop market it isn't likely that most of the buyers are going to be keen to throw modularity out the window for efficiency. Some of the efficiency trade-offs here are markets that Apple is tossing aside. As long as Intel and AMD are shooting at broader market coverage, they will likely continue to make different trade offs.

So yeah, first I want to see Intel getting anywhere, because so far, they are not. Alder Lake is Great of course, but it doesn’t bring any noteworthy increases in efficiency and it’s praised performance improvements boil down to the massive increase in cores which gives you good results in some popular benchmarks. If that’s where the wild is blowing, well, Apple can easily add a couple of CPU cores to the next gen and get back ahead. The real trick is getting thst kind of per-core perf at 5watts, and so far only Apple can do that.

Actually, without a much denser fab process ... no Apple can't easily add cores and hit the same price points they are now. And as long as the costs for the bleeding edge processes keep going up each generation that is a dual edge sword also. Apple is going to fight a multiple front "war". Trying to keep up with dGPUs. Trying to keep up with discrete AI/ML solutions , trying to keep up in media de/ecode ( trailing on AV1) , trying to keep up with discrete CPU packages. CPU cores may not get huge transistor budget increases in both cache and core count.

That would have been a terrible choice. Sure, sticking with Intel could give us marginally faster desktops (and even that’s not guaranteed) today. But the real strength of Apple Silicon is a new unified programming model. CPU, GPU, vector coprocessors, ML coprocessors, unified memory. Developing and testing becomes simpler and unlocks new programming paradigms.

That real strength doesn't come for 'free'. It is all integrated... but it is all fixed integrated. Again scoped down to just laptops a more reasonable trade-off than as scale up the desktop product space.

Apple Silicon is a truly heterogeneous system with multiple programmable processors thst can work in unison. Can’t really do that with x86 - yet. There are indications that they feel the pressure and want to get there.

Errr..... it isn't like Intel and AMD don't have fused on die solutions. "x86" in and of itself doesn't inhibit heterogeneous compute solutions. Neither AMD or Intel have "bet the whole farm" on it, but it isn't like they haven't worked on it. (and some of this has to do with operating support and security.... not CPU core design. )

 

[jav6454]

The only likely situation I see for Intel regaining Apple's business would be ASi fabbed on an Intel process node, were they to match or exceed TSMC. Seems unlikely at the moment, but not impossible. Competition benefits us all.

I highly doubt Intel will ever regain that business. Apple's PA Semi acquisition was the nail on the coffin for Intel's dominance on the Mac.

Furthermore, the entire Apple Silicone is way too advanced and has Mac-oriented features at this point that it makes no sense to switch back to an Intel CPU.