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

941–950 of 1001 posts

Re: Apple introduces M4 chip

#941

Post by ethagnawl »

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…

> Processing data at the edge also makes for the best possible user experience because of the complete independence of network connectivity and hence minimal latency.

I know a shop who's doing this and it's a very promising approach. The ability to offload the costs of cloud GPU time is a tremendous advantage. That's to say nothing of the decreased latency, increased privacy, etc. The glaring downside is that you are dependent upon your users to be willing and able to run native apps (or possibly WASM, I'm not sure) on bleeding edge hardware. However, for some target markets (e.g. video production, photography, designers, etc.) it's a "safe" assumption that they will be using the latest and greatest Macs.

I've also been hearing people talk somewhat seriously about setting up their own training/inference farms using Macs because, at least for now, they're more readily available and cheaper to buy/run than big GPUs. That comes with a host of ops problems but it still may prove worthwhile for some use cases and addresses some of the same privacy concerns as edge computing if you're able to keep data/computation in-house.

Re: Apple introduces M4 chip

#942

Post by fennecfoxy »

cafed00d wrote: post #527
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.

I like to think back to 2011 and paraphrase what people were saying: "Is anyone seriously using gpu hardware to write nl translation software at the moment?" "No, we should be use cheap commodity abundantly available cpus and orchestrate then behind cloud magic to write our nl translation apps" or maybe "no we should build purpose built high performance computing hardware to write our nl translation apps" Or perhaps…

I don't think of examples really apply, because it's more a question of being on "cutting edge" vs personal hardware.

For example, running a local model and access to the features of a larger more capable/cloud model are two completely different features therefore there is no "no we should do x instead".

I'd imagine that a dumber local model runs and defers to cloud model when it needs to/if user has allowed it to go to cloud. Apple could not compete on "our models run locally privacy is a bankable feature" alone imo, TikTok install base has shown us enough that users prefer content/features over privacy, they'll definitely still need SoA cloud based models to compete.

Re: Apple introduces M4 chip

#943

Post by throwaway1194 »

_zoltan_ wrote: post #930
throwaway1194 wrote: post #920

I dread using a Mac for any serious work, you lose a lot of the advantages of Linux (proper package and window management, native containers, built-in drivers for every hardware out there, excellent filesystems support, etc). And you get what exactly?

Best hardware on the world best desktop OS. I wouldn't trade my MBP .

Best according to whom and for what reason? Give me some real reasons for why it is "best". That's what Apple's marketing department would say.

Re: Apple introduces M4 chip

#944

Post by dabbz »

andsoitis wrote: post #539
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…

> 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…

When I was at Apple for a short time, there was a small joke I hear from the ex-amazonians there who would say "What's the difference between an Apple software engineer and an Amazon software engineer? The Amazon engineer will spin up a new service on AWS. An Apple engineer will spin up a new app". Or something along those lines. I forget the exact phrasing. It was a joke that Apple's expertise is in on-device features, whereas Amazon thrives in the cloud services world.

Re: Apple introduces M4 chip

#946

Post by losvedir »

jameshart wrote: post #579
SpaceManNabs wrote: post #551

Earlier quoted context omitted.

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 pr…

Everyone’s strategy? The biggest players in commercial AI models at the moment - OpenAI and Google - have made absolutely no noise about pushing inference to end user devices at all. Microsoft, Adobe, other players who are going big on embedding ML models into their products, are not pushing those models to the edge, they’re investing in cloud GPU. Where are you picking up that this is everyone’s strategy?

Of course Google is. That's what Gemini Nano is for.

Re: Apple introduces M4 chip

#947

Post by faeriechangling »

onlyrealcuzzo wrote: post #39
praseodym wrote: post #10

"With these improvements to the CPU and GPU, M4 maintains Apple silicon’s industry-leading performance per watt. M4 can deliver the same performance as M2 using just half the power. And compared with the latest PC chip in a thin and light laptop, M4 can deliver the same performance using just a fourth of the power." That's an incredible improvement in just a few years. I wonder how much of that is Apple engineering a…

On one hand, it's crazy. On the other hand, it's pretty typical for the industry. Average performance per watt doubling time is 2.6 years: https://newsroom.arm.com/blog/performance-per-watt#:~:text=T... .

It's a shame performance per watt doesn't double every 2.6 years for modems and screens.

Re: Apple introduces M4 chip

#949

Post by faeriechangling »

themagician wrote: post #421
simonbarker87 wrote: post #21

Seeing an M series chip launch first in an iPad must be result of some mad supply chain and manufacturing related hangovers from COVID. If the iPad had better software and could be considered a first class productivity machine then it would be less surprising but the one thing no one says about the iPads is “I wish this chip were faster”

Welcome to being old! Watch a 20-year old creative work on an iPad and you will quickly change your mind. Watch someone who has, "never really used a desktop, [I] just use an iPad" work in Procreate or LumaFusion. The iPad has amazing software. Better, in many ways, than desktop alternatives if you know how to use it. There are some things they can't do, and the workflow can be less flexible or full featured in some…

It's not a matter of being young or old, it's that iPadOS is not tooled to be a productive machine for software developers, but IS tooled to be productive machine for artists.

Re: Apple introduces M4 chip

#950

Post by bmitc »

jascination wrote: post #794
bmitc wrote: post #734

Earlier quoted context omitted.

> [Apple's] privacy-first strategy That is a marketing and advertising strategy.

I'm a cross-platform app developer and can assure you iOS is much stronger in terms of what data you can/can't get from a user

That says nothing about what Apple does with data.
Post reply on HN