Hacker News new | past | comments | ask | show | jobs | submit login
Apple's M1 processor spotlights Intel's chip challenges (cnet.com)
35 points by eddieoz 7 months ago | hide | past | favorite | 78 comments



This reads a lot like Palm’s CEO back in the day:

“We’ve learned and struggled for a few years here figuring out how to make a decent phone,” he said. “PC guys are not going to just figure this out. They’re not going to just walk in.”

https://web.archive.org/web/20061205211900/http://www.mercur...


Yes, it’s fun to see echoes of the reaction to the iPhone announcement so many years hence.

I suppose you could argue, then as now, that the disbelief is performative and not a real reaction. I have my doubts, though.


I am still of the view that the real issue here one that Apple has never overcome. Software. This was also very obvious during the recent presentation, very few companies highlighted and of those how many did you recognize?

For current Mac users we still have no means by which to look at our installed software and know if Rosetta will handle it or in many cases if the developer is even planning on supporting it in the future. Catalina did a good job of punting a lot of old software to the side and decimated gaming on the Mac.

I know I am in a minority here but for me I want only one desktop/laptop type machine for both productivity and games. As above, Catalina decimated the gaming environment on Mac and frankly the iOS game environment isn't comparable.

So, yes in some limited cases Intel sounds a bit like Palm but the difference is in phones it was a wide open environment to exploit and the current desktop/laptop market is heavily entrenched with established 3rd party software that must go along for the ride and that condition did not exist for phones, though now it does.

Now if Apple could find inroads into the server market.


> ...the difference is in phones it was a wide open environment to exploit and the current desktop/laptop market is heavily entrenched...Now if Apple could find inroads into the server market.

These observations of yours actually go together.

The phone business was highly entrenched in 2007 with carriers having all the control (especially in North America) over a highly sclerotic system that sucked almost all the value out of any third party apps. Apple truly disrupted them all: phone mfrs, carriers, and the app environment.

Apple has an opportunity to do the same with the cloud business. They have three lines of attack, ones they've used before. 1 - a lot of cloud code is platform agnostic (in Python or js etc) so for that the switching cost will be lower 2 - they have an edge in privacy protection and, on a cost basis, their NN chips vs Nvidia who has been lucky enough to enjoy a quasi-monopoly. People will pay for this, especially when/if privacy rules start to have teeth. 3 - Since they are vertically quite integrated and build their own chips, Apple can build web hardware optimized for high performance networking, lots of NNs, etc with kernels to go with. Their cost structure is potentially a lot lower even compared to open connect folks because they already have everything set up to build lots of machines so they can be quite profitable while undercutting the Amazons of the world.

Local server machines? Not a chance. And will they do the cloud? It seems likely they will be driven to it, but given their decades of poor execution on anything but hardware it seems better than even odds that they would flub it despite putting great people on it.


Every single major cloud providers builds their own chips and most have gone even further with custom ASICs for ML. I don't see any reason to believe that Apple has the organizational or technical ability to enter the cloud business successfully. I don't think they even see data center ops as a core competency and you can't go into the cloud business without expertise there. AFAIK they primarily rely on public clouds.


> highlighted and of those how many did you recognize?

Are you kidding? All of them. I mean, Adobe, Microsoft?

> we still have no means by which to look at our installed software and know if Rosetta will handle it or in many cases

It will, the cases were Rosetta 2 won’t work are extremely rare, hand tuned vector instructions from high performance software that is probably well supported anyway.

As for games, yes, if you care about them at all, don’t buy a Mac. At least not until Apple changes its attitude towards it.

> Now if Apple could find inroads into the server market.

Yes, one can dream.


> during the recent presentation, very few [Software ] companies highlighted and of those how many did you recognize?

I don't know if I'm an outlier but I recognized many of them. I have no idea if they (apart from some real biggies like MS or Adobe) have any Windows presence, and I know there are big Windows companies I don't know exist.


> Now if Apple could find inroads into the server market.

That would be catastrophic


How so?


The M1 is a revolution and upcoming M2, M3 and so forth will be strong evolutions. Apple has a good track record of improving their chips year over year. Look at initial comments of the iPhone by the competition. They were all laughing. A decade later they are not.

With full control over the hardware and not relying on Intel for the processor, Apple is aiming to reproduce its iPhone success in PCs. They have everything going for them.

The future looks bleak for Intel and AMD. They are stuck evolving a 30 year-old architecture.

In fact, this is where Apple shines in terms of long-term strategy and execution. They are able to create a revolution with a new product/architecture (iPod, iPhone, iPad and now PC), then iterate to make it evolve over time. In doing so, they create new markets. I really feel like the M1 is an inflection point in the personal computers market.

Can't wait to see how Qualcomm and Nvidia (with the acquisition of ARM), will respond.


> With full control over the hardware and not relying on Intel for the processor, Apple is aiming to reproduce its iPhone success in PCs.

With Mac desktops and mobile devices getting similar in hardware the question is how much longer it will take, till you can plug your phone to a screen, keyboard and mouse and end up with a decent desktop for most use cases.

(There are tricky questions in that for instance whether one can plug it out for a break or something and how well such a thing would recover and a question whether one would be able to separate work and private life in such a world, but all should be solvable)


It is possible to do that wirelessly with samsung dex already. I use it regularly, with an external mouse/keyboard for dex desktop on a TV, and I can use the phone normally. They supported linux on Dex for a while, but that is unsupported now. If Samsung brings a linux desktop with Miracast as an alternative to their own desktop, I might just switch full time to that.


There is a difference between the MacOS environment and existing software vs. Android apps on a desktop-like environment. If Apple executes that well the transition could be mostly fluent, hardly noticed. With the transition to Apple Silicon right now being the most radical step in the process.


I built a system years ago that did this one better.

imagine your iphone (with its 128 or 256GB) of storage being the storage device of your computer. But not necessarily the CPU. When you're at a more powerful computer (say a laptop or mini), you can plug in and all your processes now run on faster computer. When you want to go remote, you migrate the processes to the portable device.

imagine being able to carry a single computer in your pocket like this betwen home and work and not having to carry a laptop.

http://systems.cs.columbia.edu/files/wpid-compsac2006-fordis...


Samsung DeX already does that. I plugged a colleagues phone in to this screen and it was pretty cool: https://www.laptopsdirect.co.uk/electriq-eiq-15fhdpmt-15.6-t...

The screen has a USB A port, so a keyboard and mouse were useable with the dongle. It also has two USB-C ports allowing you to charge the phone through the screen. Touch screen was excellent too.


PinePhone is getting there: https://www.youtube.com/watch?v=yBeza4UNOm8


I agree re Intel, but AMD's been doing some nice revolution/evolution type moves themselves with chiplet / Zen1/2/3. Architecture makes a difference, but I don't think it's as big of a factor as the internal innovations.

Dream setup, maybe AMD would hop on the RISC-V or OpenPOWER trains if it does end up looking like x86 is going the way of the dodo.


The competitors will be fine unless Apple broadens away from SoCs (which isn't in line with Apple's walled garden approach, so I doubt it). There will always be a market where footprint and performance per watt are not issues. ie. I don't think the SoC vs CPU debate is settled.


>There will always be a market where footprint and performance per watt are not issues.

Unfortunately even that is gone. If you need multiple cores, Pref Watt Scale up with each core adding to it. Unless you want absolute single thread performances. But even that Apple seems to be winning.

We have enter the age of TDP Computing.


I believe that market will shrink. SoC brings many advantages. I think it will overtake the majority of the market. Time will tell.


We thought that in the 1990s and a lot of VC money was spent on it. I'd like to see it happen for technical reasons, but still I wait to see it happen.


> They are stuck evolving a 30 year-old architecture.

ARM is older than you may think :) AFAIR the ARM instruction set was developed just a couple years after the Intel 8086, 1983 vs 1978. They're both nearing 40 years, but not that modern x86-64 or ARM64 are the same as those old ISAs.


True, but ARM, and apple in particular have been willing to discard compatibility with previous generations. I believe the M1 for instance is not 32 bit compatible. ARM also was willing to do some house cleaning when they designed the 64 bit architecture.

X86-64s on the other hand have compatibility not just to 32 bit, but also 16 bit. You can still boot dos 1.0 if you wanted.


AMD not had license to make ARM cpus ? https://www.extremetech.com/computing/181867-amds-project-sk...

Form my point of view, they only need to replace the decoding units and, perhaps, the simd units of a Zen core to have an arm cpu.


> The future looks bleak for Intel and AMD. They are stuck evolving a 30 year-old architecture.

ARM is over 30 years old as well.


64 bit Arm is ten years old - if you dispense with 32 bit as Apple has then there is no legacy to carry forward.


My point is that PC architecture still remains CPU + GPU + RAM mostly separate communicating over a large bus. The M1 combines this all in a SoC.


Not sure that talking about 'innovations that only Intel can do" and then mentioning two things that are not innovations at all and which other companies can in fact do is very reassuring.

I suspect that Intel's near term prospects are defined by how quickly TSMC can get fabs up and running - not a great place to be.


That was very worrisome to me, too: backwards compatibility is a real market but it’s an inherently shrinking one and profits are capped with multiple competitors, even including Apple - if that old software ran on old processors it’ll almost certainly run fast enough in a modern emulation/translation system on new hardware while users migrate to non-legacy software.

Similarly, the full-range of PCs comment reads like they’re hoping nobody noticed that the low-end market is increasingly using cheaper ARM chips. Here again x86 compatibility is worth something but for a lot of people it comes down to whether it runs Chrome and Office, so there’s an upper bound on how much money you can squeeze out with such effective competition.

Intel has the resources to do better than this but I no longer have confidence that they have management who can effectively run the company.


Need less talk and more actual innovation. I'd love to see something around the performance and power efficiency of the M1 without it all needing to be built into a single chip. The M1 definitely gets a lot of benefits from being all packaged together, but it unfortunately leads to fewer options since you can't mix-and-match CPU, GPU, and RAM.

I'm hopeful that M1 will drive both AMD and Intel to do better in the mobile space. AMD got a lot of praise for the 4000 series, but it's had nearly no availability.


Yes kudos to apple for delivering astonishing performance and performance/watt, but to use these chips you'll have to accept Apple's full embrace - the chip, the device it powers, the OS, the app ecosystem and restrictions.


Is that entirely true? I don't see any reason you couldn't load Linux on, for example, the Mac Mini. I would normally never consider Apple products, because I don't care for the lock-in either, but... the value proposition packed into these new M1 chips is certainly tempting given the <$700 price tag! For me personally, I could easily be persuaded once any teething pains (related to loading Linux on this new hardware) are figured out.


I could see this being the time when they prevent 3rd party OSes using some kind of code signing/hardware verification, idk what. But that's speculation, no need to prematurely come up with things to be mad/disappointed about when there's no evidence of that now.


IIRC, Apple said something about Apple Silicon not "supporting" other OSes, but it was ambiguous as to whether "support" meant "it won't work" or "we don't help you do it".


You cannot load Linux on an iPhone. So don't be surprised if you cannot on your Mac, either.


To be pedantic, it sure is possible to run Linux (and Android) on an iPhone 7, although it requires some security vulnerabilities: https://projectsandcastle.org


And a lot of stuff just doesn't work yet -- like graphics acceleration, audio, cellular networking, the camera... it's pretty rough, and the support matrix only starts looking worse for newer devices.

https://projectsandcastle.org/status


For sure, it will only run MacOS barring any jailbreak from the community.

Even previous gen Mac Mini's with the T2 chip made it difficult to reinstall even MacOS due to Find My Mac.


Yes, but to some degree, that’s _why_ Apple can deliver these chips with their level of performance while performing a near-seamless architecture transition.


What restrictions are you referring to for the Mac ecosystem ?


Apple has stated they will not support running alternative operating systems on Macs with Apple Silicon.


Not yet for mac, but appstore for iphone. I'm sure it's coming for M1 macs too.


Why are people so sure? The Mac is a different sort of device with a different use case and people who use it don't have the same expectations about it that they, and others, have about phones.


It’s not likely in the near term that Apple will lock down Mac software to the extent they do on iOS. But they’ve already taken many steps—over several macOS iterations—to either lock it down more than it had been, or to make overriding default restrictions difficult (SIP) or non-obvious (right click -> open to bypass gatekeeper). It’s unlikely they could take it drastically further without alienating a lot of developers, or without providing some kind of blessed escape hatch (e.g. you can run arbitrary software if you have a developer license; even that would alienate anyone developing software for non-Apple targets like web/server).


That’s the beauty of it.

To turn the phrase, there’s a lot of innovation in the M1 that only Apple can do.


If you look at the way MS works with NVIDIA, AMD, and Intel this advantage isn’t that large. AMD SoCs have Unified Memory, there’s DirectStorage to let SSDs load directly into GPU memory without the CPU, etc

There’s also a downside to being a walled garden, Apple is building itself a Great Wall, but the barbarians eventually got the better tech. So for example, AMD and NVIDIA GPUs have Directx ray tracing. How many generations before Apple GPUs catch up with Ampere or RDNA3?

To some extent having the GPU architectures discrete allows them to iterate much faster as they aren’t tied to other dependencies, or left with inferior thermal and power constraints.


I think you’re absolutely right if you need the highest absolute performance regardless of performance per dollar or performance per watt.

But I think a lot of people are more concerned with Perf/$ and Perf/Watt and will find Apple’s fully integrated approach is extremely hard to beat or even compete with.

In this case perf/$ is going to provide Apple an unbeatable margin, not necessarily the consumer an unbeatable price.


The problem with AMD in the mobile space is that changing the CPU on a laptop board is much more difficult than a desktop board. There is no commodity market for those. Intel has a real lock on laptops, which is a big problem for AMD.


Competition is good for everyone. Having Intel dominate the market was never a good idea (clearly AMD is making some inroads). For Apple, control of all suppliers has always been a goal of hardware design. It only took 20 years...


Reminds me of Acorn, where Sophie Wilson describes they travel to the US to visit a chip/CPU design shop.

Their expectations are not met when they find only a couple of guys in a plain office building.

They decide they can do it themselves and design the ARM1.


Apple collaborated with Acorn on the ARM610, back around 1990.

https://en.wikipedia.org/wiki/ARM_architecture#Advanced_RISC...


ARM was probably the only realistic CPU Apple could put in the Newton PDA, at about that time.

Apply had a large stake in the company. Amusingly, it is possible that selling that stake latter, saved Apple - this

https://www.cultofmac.com/97055/this-is-how-arm-saved-apple-...


The future looks bleak for Intel and AMD. They are stuck evolving a 30 year-old architecture.

This is a bit silly. Neither Intel nor AMD are particularly constrained by the underlying ISA that their chips are using.


Heh, dunno, notice that Apple is comparing themselves to the Intel chips, and in particular the i9. But NOT the AMD chips. On various metrics the Zen 3 looks pretty good compared to M1. AMD has Higher peak performance/throughput, similar per thread performance, and I've yet to see a GPU comparison.


This isn’t because they compare to AMD unfavorably, but because they habitually compare against their past offerings and against their goliath du jour (notice how they very seldom compare against other software vendors anymore since the iPhone has been a stable profit leader).


And AMD has released ARM chips as recently as 4 years ago


>> Intel said it's "relentlessly" focused on building leading chips. "We welcome competition because it makes us better," Intel said in a statement. "We believe that there is a lot of innovation that only Intel can do," including supplying chips that span the full price range of PCs and that can run older software still common in businesses.

-- IMO, the chip seems to bring some innovations, but nothing too new or disruptive for this market.


I think the real advantage of Apple so far is to have the kind of volumes and margins with iphones to be able to have privilegied access to TSMC fabs and processes, and therefore now it will automatically trickle down on Macs and their new SoCs granting them a head start on everybody else, even irrespective of any other kind of possible architectural advantage or design decision.


Title should be "Apple's M1 processor spotlights Intel's chip challenges" and while this picked quote is in there, it goes on to list legitimate value (although hardly innovations)


Reminds me of Ballmer's comments when the original iPhone was release.


"Apple and Intel are in the middle of a slow-motion breakup."

I'd say it's pretty decisive at this point.


Apple will continue to buy Intel chips in quantity for at least 18 months into the future.


Of course, but I'd say this is less a 'slow motion breakup', which has obviously already been going on for the last few years, as Intel has been cruising along oblivious, while Apple has been furiously building it's own capabilities and an escape, and more a case of 'divorce papers have now been filed, but these things take time to conclude'


Probably longer to meet support contract obligations. But the idea that they might prolong the transition more than the market will bear, and their operations can turn out ARM product, is silly. They invested years of R&D into this transition, they’ve made huge hardware architectural changes smoothly twice with tons of preparation, and if anything they let the decision linger longer than anyone watching for it would’ve expected given the A-series improvements and given Intel’s many disappointments over the last few years.

When they switched from PPC to Intel, they had clearly been desperately trying to keep PPC alive and viable. With Intel they’ve been just openly showing the world they can do better and just daring anyone try to keep up.


Let's see you ship that 5nm in 2023, Intel. At that point TSMC will be doing 3nm.


Intel’s 5nm is supposed to be around TSMC’s 3nm target that said Intel’s process has been stuck in muck for years now.


yeah. just like the "revolutionary" innovations done by zen 2 and zen 3 by AMD, which happens to not be intel. the same innovation by intel meant to keep active exploits in the market just because.


> There is a lot of innovation that only Intel can do

In other words, innovation which solves x86 problems doesn't exist on other architectures.


Except oh wait, AMD seems to have x86 innovation on lock anyway


hahaha, "pricing innovation" they mean, or perhaps "marketing innovation"


All of this is useless chatter for most folks that just want a decent gaming platform. Sadly, Intel on Windows is going to be where folks go.


No one sees Mac as a gaming platform at all. And gamers are a small subset of the user base.


However people do see iOS as a gaming market, and soon you can run those apps on a Mac (clearly limitations in interface might restrict it a little). Gaming has many segments. I play WOT on my Mac via Codeweavers emulation, and performance is 60FPS+ in my 2017 iMac in most cases, equivalent to a midrange PC. I am not sure how this will work on M1 however.


I wonder when Apple will finally ship a touchscreen Mac. It seems almost inevitable now that iOS apps will be running on MacOS.


To be fair, those gaming markets are almost completely non-overlapping on a venn diagram, and it’s not just the interface (as in user input) that drives that. It’s the form factor. Handheld touchscreen games and many platform/PC games are just drastically different in the interaction models they employ/require, and slapping a touch screen on a MacBook isn’t going to change that meaningfully without many other significant form factor considerations.


WOT was made to run on low end systems, not really that impressive. Can Valorant, Overwatch, or other triple-A content run?


I'm pretty sure gamers are not willing to see a gaming computer on Apple. And who uses Apple are not in the gaming field.


Most folks that want a decent gaming platform will buy a console, and they are much freer in their choice of processor.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: