Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Apple introduces M4 chip

apple.com

Post reply on HN
Print view View original

551–560 of 1001 posts

Re: Apple introduces M4 chip

#551

Post by SpaceManNabs »

rsp1984 wrote: post #457

Together with next-generation ML accelerators in the CPU, the high-performance GPU, and higher-bandwidth unified memory, the Neural Engine makes M4 an outrageously powerful chip for AI. In case it is not abundantly clear by now: Apple's AI strategy is to put inference (and longer term even learning) on edge devices. This is completely coherent with their privacy-first strategy (which would be at odds with sending dat…

This comment is odd. I wouldn't say it is misleading, but it is odd because it borders on such definition.

> Apple's AI strategy is to put inference (and longer term even learning) on edge devices

This is pretty much everyone's strategy. Model distillation is huge because of this. This goes in line with federated learning. This goes in line with model pruning too. And parameter efficient tuning and fine tuning and prompt learning etc.

> This is completely coherent with their privacy-first strategy

Apple's marketing for their current approach is privacy-first. They are not privacy first. If they were privacy first, you would not be able to use app tracking data on their first party ad platform. They shut it off for everyone else but themselves. Apple's approach is walled garden first.

> Processing data at the edge also makes for the best possible user experience because of the complete independence of network connectivity

as long as you don't depend on graph centric problems where keeping a local copy of that graph is prohibitive. Graph problems will become more common. Not sure if this is a problem for apple though. I am just commenting in general.

> If (and that's a big if) they keep their APIs open to run any kind of AI workload on their chips

Apple does not have a good track record of this; they are quite antagonistic when it comes to this topic. Gaming on apple was dead for nearly a decade (and pretty much still is) because steve jobs did not want people gaming on macs. Apple has eased up on this, but it very much seems that if they want you to use their devices (not yours) in a certain way, then they make it expensive to do anything else.

Tbf, I don't blame apple for any of this. It is their strategy. Whether it works or not, it doesn't matter. I just found this comment really odd since it almost seemed like evangelism.

edit: weird to praise apple for on device training when it is not publicly known if they have trained any substantial model even on cloud.

Re: Apple introduces M4 chip

#552

Post by robbomacrae »

legitster wrote: post #475
rsp1984 wrote: post #457

Together with next-generation ML accelerators in the CPU, the high-performance GPU, and higher-bandwidth unified memory, the Neural Engine makes M4 an outrageously powerful chip for AI. In case it is not abundantly clear by now: Apple's AI strategy is to put inference (and longer term even learning) on edge devices. This is completely coherent with their privacy-first strategy (which would be at odds with sending dat…

> This is completely coherent with their privacy-first strategy (which would be at odds with sending data up to the cloud for processing). I feel like people are being a bit naïve here. Apple's "Privacy First" strategy was a marketing spin developed in response to being dead-last in web-development/cloud computing/smart features. Apple has had no problem changing their standards by 180 degrees and being blatantly ant…

Having worked at Apple I can assure you it's not just spin. It's nigh on impossible to get permission to even compare your data with another service inside of Apple and even if you do get permission the user ids and everything are completely different so theres no way to match up users. Honestly its kind of ridiculous the lengths they go to and makes development an absolute PITA.

Re: Apple introduces M4 chip

#553

Post by winwang »

I would want to know if the LPDDR6 rumors are substantiated, i.e. memory bus details.

https://forums.macrumors.com/threads/lpddr6-new-beginning-of...

If M4 Max could finally break the 400GBps limit of the past few years and hit 600 GBps, it would be huge for local AI since it could directly translate into inference speedups.

Re: Apple introduces M4 chip

#554

Post by acdha »

seec wrote: post #485
legitster wrote: post #475

Earlier quoted context omitted.

> This is completely coherent with their privacy-first strategy (which would be at odds with sending data up to the cloud for processing). I feel like people are being a bit naïve here. Apple's "Privacy First" strategy was a marketing spin developed in response to being dead-last in web-development/cloud computing/smart features. Apple has had no problem changing their standards by 180 degrees and being blatantly ant…

Don't bother the fanboys have an Apple can't do anything wrong/malicious. At this point it's closer to a religion than ever. You would be amazed at the response of some of them when I point out some shit Apple does that make their products clearly lacking for the price, the cognitive dissonance is so strong they don't know how to react in any other way than lying or pretending it doesn't matter.

If you’re annoyed about quasi-religious behavior, consider that your comment has nothing quantifiable and contributed nothing to this thread other than letting us know that you don’t like Apple products for non-specific reasons. Maybe you could try to model the better behavior you want to see?

Re: Apple introduces M4 chip

#555

Post by Draiken »

bamboozled wrote: post #543
andsoitis wrote: post #539

Earlier quoted context omitted.

> In case it is not abundantly clear by now: Apple's AI strategy is to put inference (and longer term even learning) on edge devices. This is completely coherent with their privacy-first strategy (which would be at odds with sending data up to the cloud for processing). Their primary business goal is to sell hardware. Yes, they’ve diversified into services and being a shopping mall for all, but it is about selling lu…

As soon as the privacy thing goes away, I'd say a major part of their customer base goes away too. Most people use android so they don't get "hacked" if Apple is doing the hacking, I'd just buy a cheaper alternative.

At least here in Brazil, I've never heard such arguments.

Seems even more unlikely for non technical users.

It's just their latest market campaign, as far as I can tell. The vast majority of people buy iPhones because of the status it gives.

Re: Apple introduces M4 chip

#556

Post by jamesmontalvo3 »

bamboozled wrote: post #543
andsoitis wrote: post #539

Earlier quoted context omitted.

> In case it is not abundantly clear by now: Apple's AI strategy is to put inference (and longer term even learning) on edge devices. This is completely coherent with their privacy-first strategy (which would be at odds with sending data up to the cloud for processing). Their primary business goal is to sell hardware. Yes, they’ve diversified into services and being a shopping mall for all, but it is about selling lu…

As soon as the privacy thing goes away, I'd say a major part of their customer base goes away too. Most people use android so they don't get "hacked" if Apple is doing the hacking, I'd just buy a cheaper alternative.

Maybe true for a lot of the HN population, but my teenagers are mortified by the idea of me giving them android phones because then they would be the pariahs turning group messages from blue to green.

Re: Apple introduces M4 chip

#557

Post by spxneo »

adrian_b wrote: post #244
dhx wrote: post #158

M2's Neural Engine had 15TOPS, M3's 18TOPS (+20%) vs. M4's 38TOPS (+111%). In transistor counts, M2 had 20BTr, M3 25BTr (+25%) and M4 has 28BTr (+12%). M2 used TSMC N5P (138MTr/mm2), M3 used TSMC N3 (197MTr/mm2, +43%) and M4 uses TSMC N3E (215MTr/mm2, +9%).[1][2] [1] https://en.wikipedia.org/wiki/5_nm_process#%225_nm%22_proces... [2] https://en.wikipedia.org/wiki/3_nm_process#%223_nm%22_proces...

> The Most Powerful Neural Engine Ever While it is true that the claimed performance for M4 is better than for the current Intel Meteor Lake and AMD Hawk Point, it is also significantly lower (e.g. around half) than the AI performance claimed for the laptop CPU+GPU+NPU models that both Intel and AMD will introduce in the second half of this year (Arrow Lake and Strix Point).

damn bro thanks for this

here i am celebrating not pulling the trigger on M2 128gb yesterday

now im realizing M4 ain't shit

will wait a few more months for what you described. will probably wait for AMD

> Given that Microsoft has defined that only processors with an NPU with 45 TOPS of performance or over constitute being considered an 'AI PC',

so already with 77 TOPS it just destroys M4. Rumoured to hit the market in 2 months or less.

Re: Apple introduces M4 chip

#558

Post by victorbjorklund »

krunck wrote: post #460
rsp1984 wrote: post #457

Together with next-generation ML accelerators in the CPU, the high-performance GPU, and higher-bandwidth unified memory, the Neural Engine makes M4 an outrageously powerful chip for AI. In case it is not abundantly clear by now: Apple's AI strategy is to put inference (and longer term even learning) on edge devices. This is completely coherent with their privacy-first strategy (which would be at odds with sending dat…

Yes, that would be great. But without the ability for us to verify this who's to say they won't use the edge resources(your computer and electricity) to process data(your data) and then send the results to their data center? It would certainly save them a lot of money.

If you trust that Apple doesn't film you with the camera when you use the phone while sitting on the toilet. Why wouldn't you trust Apple now?

It would have to be a huge conspiracy with all Apples employees. And you can easily just listen to the network and see if they do it or not.

Re: Apple introduces M4 chip

#559

Post by xanderlewis »

dylan604 wrote: post #532
joelthelion wrote: post #461

Earlier quoted context omitted.

>n case it is not abundantly clear by now: Apple's AI strategy is to put inference (and longer term even learning) I'm curious: is anyone seriously using apple hardware to train Ai models at the moment? Obviously not the big players, but I imagine it might be a viable option for Ai engineers in smaller, less ambitious companies.

Does one need to train an AI model on specific hardware, or can a model be trained in one place and then used somewhere else? Seems like Apple could just run their fine tuned model called Siri on each device. Seems to me like asking for training on Apple devices is missing the strategy. Unless of course, it's just for purely scientific $reasons like "why install Doom on the toaster?" vs doing it for a purpose.

It doesn’t require specific hardware; you can train a neural net with pencil and paper if you have enough time. Of course, some pieces of hardware are more efficient than others for this.

Re: Apple introduces M4 chip

#560

Post by acdha »

deadmutex wrote: post #503
dheera wrote: post #483

Earlier quoted context omitted.

> This is completely coherent with their privacy-first strategy Apple has never been privacy-first in practice. They give you the illusion of privacy but in reality it's a closed-source system and you are forced to trust Apple with your data. They also make it a LOT harder than Android to execute your own MITM proxies to inspect what exact data is being sent about you by all of your apps including the OS itself.

Yeah, given that they resisted putting RCS in iMessage so long, I am a bit skeptical about the whole privacy narrative. Especially when Apple's profit is at odds with user privacy.

RCS is a net loss for privacy: it gives the carriers visibility into your social graph and doesn’t support end to end encryption. Google’s PR campaign tried to give the impression that RCS supports E2EE but it’s restricted to their proprietary client.
Post reply on HN