Do you think Apple will go all out to annihilate Intel CPUs or hold back a bit?

[robco74]

Intel doesn't really need to worry about Apple specifically. What they, and potentially Microsoft, need to worry about is other companies following Apple's lead. There may be others willing to dedicate more R&D to start taping out their own custom chips. To compete, Intel, AMD, and MS need to work together more closely than before to provide more customized solutions. Apple has been one of the few willing to make chips specifically for tablets and wearables.

But in addition to PPW, also consider that every new Mac also has ML acceleration, something developers can now target, since they don't have to wonder if the capability will be present or not. The new GPU baseline is pretty good. Apple is willing to design chipsets for user workflows, something Intel and AMD haven't really done before. Instead of churning out large volumes of general purpose processors, the shift will be toward smaller runs of specialized silicon. If ARM makes the ISA licensing easier, or RISC-V starts gaining traction in the future, that would be a bigger risk. Not to mention Intel's manufacturing hiccups.

If I were Intel, I would worry more about Qualcomm, Ampere, and Amazon. Or a company like Samsung putting more resources into their chip and software development to really start competing with Apple. We've already heard whispers of Microsoft getting into the game as well.

 

[iPadified]

So naive, maybe you should give the clearly incompetent engineers at AMD and Nvidia a call to help them improve their performance and crush their competition, they are sure to offer you an amazing job. Even better, give Intel a call - "just increase the number of your iGPU cores by a factor of 10, the performance will linearly increase and you will easily beat anything AMD and Nvidia have to offer." I guess the engineers have never thought of this strategy, but thankfully we have the Macrumors couch GPU architecture engineers to teach them how to do their job.
I mean, it has totally never happened (except literally every single GPU generation) that scaling up the number of processing units in a GPU lead to diminishing or even negative gains in performance. The GPU archs have definitely never been memory bandwidth starved or had other inefficiencies that only became overwhelming while scaling up. I mean, Intel Larrabee has been a fantastic success that has dominated the GPU markets for the last decade, right?

Just to put things into perspective: an entire M1 SoC has 68.25GB/s memory bandwidth, while a single 3090 has a 936.2 GB/s bandwidth to its memory.


And while we are at it, let's have @iPadified school the CPU designers by explaining how easily Apple could kill the whole CPU industry and how the performance scales linearly with the number of cores. It's not like already 32-core threadrippers easily get memory starved with their quad RAM channels and the same goes for the high core count Epyc CPUs with 8 memory channels. LOL just the part of the chip dealing with routing data in and out of an Epic CPU has 5 times the area of a whole M1 SoC.


I'm not saying M1 is bad, honestly, it's great. But it's a low power consumer SoC with no advanced functionality somebody buying Epyc CPUs or Tesla GPUs would expect. High end stuff is a whole different world. And Apple might do well there too, but you know, other companies are doing well too and it's not trivial. It's insane to see how many people here suddenly became hardware experts while it's obvious you've never read an in-depth dive into arch design in your life, which, honestly, isn't surprising among Mac users.

So please educate me. Being unpleasant without giving any real information is not the way to go. I commented on the incorrect statement that performance per watt automatically will decrease with increased performance. It is not so simple, is it?

Are you seriously thinking that Apple will not be able to make a 32 core chip that is much more efficient that AMD/Intel when starting out with a the M1 as base?

 

[turbineseaplane]

This is all great and fun, but for me (desktop usage) ASi isn't yet competitive with Intel based options on the GPU front.

I really hope that changes with the next machines released.

I'd love to see a Mac Mini Pro -- and/or what the next iMac might look like here.

 

[JohnnyGo]

iPhone market share is partly subsidized/supported by wireless carriers offering promotions and 2+ year device installment plans.

Majority of x86 computers sold are of the cheap variety and I don't think Apple has any plans to compete in that tier.

+1

PC market is determined by enterprise/corporate purchases.

Besides, Apple’s market share for premium computers/laptops (>$1000) is significantly higher (20-30%).

It’s also a fact that worldwide, cheap PCs and unlicensed software are the rule. If you take only OECD countries or US+Europe market share, Apple has a more significant presence.

 

[JohnnyGo]

Apple will probably introduce DDR5 onboard Apple Silicon in 2022. That will lead to even better performance and higher base memory configurations.

I expect AS to have two configs for PRO models:
16-cores with 32-64Gb RAM (2 memory chips)
- geared towards laptops (8-12 performance cores)
32-cores with 128-256Gb RAM (4 memory chips)
- geared towards desktops (24-28 performance cores)

“DDR5 will arrive with a minimum speed of at 4800Mbit/s, which works out to 76.8GB/s of bandwidth in a dual-channel configuration. DDR5-5600 would support 89.6GB/s, while DDR5-8400 — a still-theoretical configuration that Hynix has pledged to hit — would deliver 134.4GB/s of bandwidth.”

 

[philpalmiero]

I don't even see how AMD is a competitor. Plain and simple Apple's chips will run on Macs and AMD will not be running on Macs so there's no contest because even if AMD proves to be faster that wouldn't matter unless a person would prefer Windows to Mac.

Exactly. I doubt Apple has plans to sell its silicon wholesale to other computer makers like Dell, Assus, HP, etc. Therefore, Intel and AMD have nothing to worry about even if Microsoft ports Windows to M1.

 

[leman]

Apple will probably introduce DDR5 onboard Apple Silicon in 2022. That will lead to even better performance and higher base memory configurations.

Even more, if Apple wants to continue using unified memory, DDR5 is almost a necessity for their higher-end machines. They need RAM that would let them achieve GPU-like bandwidth level, and DDR4 just doesn’t cut it.

 

[nothingtoseehere]

Considering the great performance of the Macbook Air and with rumors flying around of 32 Core M1 versions and 64 core GPUs I kinda wonder if Apple will go all out to annihilate Intel performance per watts wise or if they're going to throttle themselves a bit now that they can easily outpace the competition.

I still can't quiet imagine what performance their 3000 usd pro laptop will deliver if their 900 usd one is already running circles around their old 3000 usd machines in certain conditions.

what do you think? will we really already see a 16 or even 32 core Macbook Pros or do you think it will be more like 10 or 12 cores + better gpus?

I think Apple will annihilate Intel from its own lineup - in the long run, it should be much more efficient to maintain macOS on one hardware platform instead of two.

But even this is still a long way to go and Apple has not the time to throttle themselves as the competition does not sleep.

I am sure, and reports say so (cf. https://techcrunch.com/2020/12/07/a...e-count-apple-silicon-chips-for-high-end-macs), that is testing chips with more cores (CPU performance, CPU efficiency, GPU, neural). The viable maximum will go to an Apple Silicon Mac Pro, the minimum is in the M1 series we already know of, and everything in between remains to be seen...

 

[Wokis]

I don't really expect to see their design scale linearly with increased wattage targets. They will be good for sure but Intel/AMD can probably compete fine on the high-end given access to a similar manufacturing node. It doesn't really surprise me that the big mac pro might just stay Intel for a while.

Wanting chips in the 5-45W range, x86 is looking less and less interesting for the laptop- and thus the majority of the regular consumer market. That is a huge blow by itself.

 

[Jpoon]

From the perspective of light based rendering engines, I'm pretty excited for what Metal is going to bring to macOS even without M1 in the short term on the 7,1 Mac Pro. It's amazing what companies like Maxon and Davinci are doing with metal with Redshift in Cinema 4D and Davinci Resolve Studio 17.

If the metal M1 Mac Pros are even better in this regard, we'll be buying a lot of them in the entertainment sector .

 

[jerryk]

Apple will produce the fastest high-performance SOCs and other packages they can at a reasonable cost. They are not gating their efforts on Intel products. They are basing their designs on what the architecture can achieve and deliver in high volumes at a reasonable price. The bottom line is Apple is not very concerned about what Intel does.

 

[leman]

I don't really expect to see their design scale linearly with increased wattage targets. They will be good for sure but Intel/AMD can probably compete fine on the high-end given access to a similar manufacturing node. It doesn't really surprise me that the big mac pro might just stay Intel for a while.

In regards to single core performance, I fully agree with you. Current Apple CPU sweet spot seems to be around 3 ghz, they might be able to push it to 3.5 ghz or slightly above, but then the power consumption will skyrocket. However, Apple has a huge potential in horizontal scalability. Professional Intel and AMD CPUs have to run each core at low frequencies to avoid overheating. E.g. a 18-core Xeon W2295 core clock is 3.0ghz to match its target TDP of around 160 watts. A hypothetical 18-core Apple CPU using current technology could also run at 3.0 ghz per core, but only consume around 100 watts. Moreover, Apple's 3 ghz offers performance comparable to that of Intel's 4.5 ghz.

Wanting chips in the 5-45W range, x86 is looking less and less interesting for the laptop- and thus the majority of the regular consumer market. That is a huge blow by itself.

I'd say that x86 actually does better in this range, since it only uses few CPU cores and single-threaded performance is still important. Apple's more efficient CPU cores will make a bigger difference in large multi-core designs. The more cores you use, the slower x86 has to be per core, the bigger the performance gap. At same power consumption, best x86 cores is about half the performance of an Apple performance core...

The big problem of course is memory... if you have dozens of super-fast cores, you need to supply them with a lot of memory bandwidth. Which is again why unified memory makes perfect sense for Apple. They need their CPUs have access to GPU-typical memory bandwidth figures.

 

[Maconplasma]

Exactly. I doubt Apple has plans to sell its silicon wholesale to other computer makers like Dell, Assus, HP, etc. Therefore, Intel and AMD have nothing to worry about even if Microsoft ports Windows to M1.

It's not necessarily about Intel and AMD having nothing to worry about (because in a way they kinda do), it's their customer base, the media and the tech world that is going to push a great deal of pressure in their faces saying if Apple can go from selling Intel-based Macs that "overheat" causing constant fan noise, draw a great deal of power from the battery and not offering maxed advertised performance without throttling.....then end with their own Apple Silicon Macs that do all of the above in the opposite direction then they will expect that type of performance for Windows machines.

So while Apple is not necessarily competing in this fashion they are certainly lighting a torch under AMD and Intel's fanny's.
One thing to note, while Apple has stated that Microsoft is the road block to why there's no Windows on M1, I doubt that Apple is in any hurry to expedite this. Pretty sure Bootcamp is officially dead. They may offer Windows under VM but with Apple's major advances they absolutely want Macs to be used as Macs running MacOS. Apple has put forth too much time and R&D with Big Sur and Apple Silicon only to "fix" their Macs to run Windows. Not gonna happen.

 

[ian87w]

Looking at Apple’s history, imo Apple will self sabotage their huge lead in some way or another. I mean just look at the rumors. We are barely at the first generation, and Apple is already going to make thinner MacBooks that will undoubtly shave whatever battery life advantage the M1 has. That’s like giving the competitors room to match it. With thinner design, comes thermal compromises, which will affect sustained performance rate.

And through time, Apple will likely to keep 8GB RAM as standard for a looong time, giving room for competitors to offer better value by having 16GB RAM as standard. In my country, we can only get the default configurations of current M1 Macs, so we can only buy models with 8GB of RAM. That’s BS. Apple’s incessant desire to profit from BTO upgrades will make the base models uncompetitive. Worse, in some markets Apple Macs become undesirable at all due to lack of desired configurations. Just look at how long Apple kept the iMac with default slow 5400rpm hard drives.

So Apple might have the technological advantage, but decisions like I mention above will surely allow the competitors to catch up.

 

[senttoschool]

iPhone market share is partly subsidized/supported by wireless carriers offering promotions and 2+ year device installment plans.

Majority of x86 computers sold are of the cheap variety and I don't think Apple has any plans to compete in that tier.

Average price of laptop sold is about $600-$700. I mentioned many times in this forum that Apple is likely to introduce a $700-$750 Macbook SE.

 

[Macalway]

Like everyone says, there isn't competition, as Apple doesn't sell silicon. I agree that crushing Intel sounds like good fun, but that just may be a side effect of robust sales

 

[thenewperson]

and Apple is already going to make thinner MacBooks that will undoubtly shave whatever battery life advantage the M1 has

Says who? Also the competition are pretty happy to go thinner as well so I don't think much will change where they stand.

 

[ian87w]

Says who? Also the competition are pretty happy to go thinner as well so I don't think much will change where they stand.

Right now, the competitors won't be able to go thinner due to intel's limitation. But Apple will according to the rumors, probably negating whatever battery life lead the M1 has. Worse, thinner design means worse thermal dissipation, which also means worse sustained performance. Just look at iPhone 12 throttling. Imo sometimes that's Apple's issue, overdesigning their own stuff beyond sensible engineering. If that's the case, then we will see the competitions catching up quicker than we imagined.

 

[senttoschool]

Right now, the competitors won't be able to go thinner due to intel's limitation. But Apple will according to the rumors, probably negating whatever battery life lead the M1 has. Worse, thinner design means worse thermal dissipation, which also means worse sustained performance. Just look at iPhone 12 throttling. Imo sometimes that's Apple's issue, overdesigning their own stuff beyond sensible engineering. If that's the case, then we will see the competitions catching up quicker than we imagined.

When Apple transitions to TSMC's 3nm in 2022, their SoCs will require ~30% less electricity at the same speed. This means Apple can make devices thinner without sacrificing speed or battery life.

So if Apple does decide to make a thinner Macbook Air/Macbook Pro, they're probably planning for 3nm, 2nm nodes since chassis designs usually last 4-5 years.

Keep in mind that the current chassis are designed for power-hungry Intel chips that can use as much as 125w. A 5nm M1 MBP uses about 20w max. A 3nm M1 would use 14w and maintain the same speed. So if Apple decides to make their laptops 10% thinner, it will likely still have similar/better battery life and faster performance.

 

[thenewperson]

Right now, the competitors won't be able to go thinner due to intel's limitation. But Apple will according to the rumors, probably negating whatever battery life lead the M1 has. Worse, thinner design means worse thermal dissipation, which also means worse sustained performance. Just look at iPhone 12 throttling. Imo sometimes that's Apple's issue, overdesigning their own stuff beyond sensible engineering. If that's the case, then we will see the competitions catching up quicker than we imagined.

Honestly I thinking you're getting ahead of yourself. Rumours don't say that they'll negate whatever battery life lead the M1 has (how would they know?), just that it'll get thinner. Those same rumours are also pointing to the MBA redesign coinciding with 3nm launch, which likely means more efficient chips. Unlike in the Intel days, Apple is actually designing their upcoming Macs for what they *know* they'll have, as opposed to what they hope Intel can deliver (or at least promises to deliver ).

And what iPhone 12 throttling?

 

[NotTooLate]

Exactly. I doubt Apple has plans to sell its silicon wholesale to other computer makers like Dell, Assus, HP, etc. Therefore, Intel and AMD have nothing to worry about even if Microsoft ports Windows to M1.

Intel and AMD has a LOT to worry about , Microsoft are looking to build their own CPU`s (starting at server side , which is a big blow right away but a different topic) and are starting to push ARM , Qualcomm just bought Nuvia to start competing at the PC/Server ARM space and Nvidia is going go buy ARM , I assume to help them get back into making servers/computers and not just GPU`s and small Tegra SoC.

So even if Apple are not selling their silicon to other OEM`s , the other giants of Silicon Valley have been scrambling to catch up and for some its new opportunities (Qualcomm ,Nvidia) to generate new revenue stream , they wont let this opportunity go.

 

[senttoschool]

Intel and AMD has a LOT to worry about , Microsoft are looking to build their own CPU`s (starting at server side , which is a big blow right away but a different topic) and are starting to push ARM , Qualcomm just bought Nuvia to start competing at the PC/Server ARM space and Nvidia is going go buy ARM , I assume to help them get back into making servers/computers and not just GPU`s and small Tegra SoC.

So even if Apple are not selling their silicon to other OEM`s , the other giants of Silicon Valley have been scrambling to catch up and for some its new opportunities (Qualcomm ,Nvidia) to generate new revenue stream , they wont let this opportunity go.

Yep.

In the near future, Intel & AMD (x86) will have a ton of competition from Qualcomm, Nvidia, Microsoft themselves, and maybe even Samsung.

Apple gave the ARM transition a huge boost with Apple Silicon. Now software makers such as Docker must support ARM which will make it easier for Microsoft's internal ARM chips, Qualcomm, Nvidia to compete directly with Intel/AMD.

Even worse for Intel/AMD, Amazon is hellbent on making its own server chips with ARM via their Graviton2 chip. Amazon will always favor its own internal chips over Intel/AMD. Amazon's AWS owns 40% of the cloud market.

I've sold my AMD/Intel stocks because of this.

 

[Joe The Dragon]

Yep.

In the near future, Intel & AMD (x86) will have a ton of competition from Qualcomm, Nvidia, Microsoft themselves, and maybe even Samsung.

Apple gave the ARM transition a huge boost with Apple Silicon. Now software makers such as Docker must support ARM which will make it easier for Microsoft's internal ARM chips, Qualcomm, Nvidia to compete directly with Intel/AMD.

Even worse for Intel/AMD, Amazon is hellbent on making its own server chips with ARM via their Graviton2 chip. Amazon will always favor its own internal chips over Intel/AMD. Amazon's AWS owns 40% of the cloud market.

I've sold my AMD/Intel stocks because of this.

Microsoft are looking to build their own CPU they better not lock it down to windows only and store only as they will fail big and anti trust will smack them down big time.

And killing years and years of old software will not go over that good. There are people who can't go cloud only and it's one thing to have your own VM on AWS but it's an other to have cloud SQL as the only choice. Some apps need full control over there own DB.
Other people need to know where there data is hosted at. Other must keep it in house.

 

[EntropyQ3]

Microsoft are looking to build their own CPU they better not lock it down to windows only and store only as they will fail big and anti trust will smack them down big time.

And killing years and years of old software will not go over that good. There are people who can't go cloud only and it's one thing to have your own VM on AWS but it's an other to have cloud SQL as the only choice. Some apps need full control over there own DB.
Other people need to know where there data is hosted at. Other must keep it in house.

Which is why x86 will stick around for a very long time. Even if Microsoft starts to take Windows on ARM seriously (still no x86-64? ), only parts of the user base will be able to, or want, to transition. The x86 market is kinda stuck in a slow downward spiral though. And as in all shrinking markets, the only way to keep revenues up is to raise prices - which in turn lowers attractiveness and the spiral down gets steeper.
But all is not roses on the ARM side of the fence either. If Nvidia owns ARM, they effectively own the ISA of the entire mobile market - a mobile Intel without an AMD. While there are a few players with an architectural liscense up until this point of the ISA, what happens later is anybodys guess. But there are obviously reasons why Nvidia buys the company and controls the ISA as opposed to just having a license that lets them control their own fate.
Their public statements don’t hold water.

Being under Nvidias thumb is not necessarily any better than being under Intels. Won’t affect Apple much, but may disrupt the ARM eco system as a whole.

 

[pshufd]

Which is why x86 will stick around for a very long time. Even if Microsoft starts to take Windows on ARM seriously (still no x86-64? ), only parts of the user base will be able to, or want, to transition. The x86 market is kinda stuck in a slow downward spiral though. And as in all shrinking markets, the only way to keep revenues up is to raise prices - which in turn lowers attractiveness and the spiral down gets steeper.
But all is not roses on the ARM side of the fence either. If Nvidia owns ARM, they effectively own the ISA of the entire mobile market - a mobile Intel without an AMD. While there are a few players with an architectural liscense up until this point of the ISA, what happens later is anybodys guess. But there are obviously reasons why Nvidia buys the company and controls the ISA as opposed to just having a license that lets them control their own fate.
Their public statements don’t hold water.

Being under Nvidias thumb is not necessarily any better than being under Intels. Won’t affect Apple much, but may disrupt the ARM eco system as a whole.

I'm quite happy that Apple is leading the charge here as it means that they will have a significant lead over the competition.

I can see myself without x86 in a few years. But there's one program I need to run on it. Not a problem because I have a system that I can use until they do a port. If they never do a port, then I will look at using different software. That's the way it goes. At some point, the technical advantages with ARM will overwhelm the requirement for x86. That transition point is different for every person or company. In some cases, some might not be able to transition at all.