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

Apple M5 chip

apple.com

Post reply on HN
Print view View original

201–210 of 1001 posts

Re: Apple M5 chip

#201

Post by zozbot234 »

fooblaster wrote: post #176
hannesfur wrote: post #167

It’s unfortunate that this announcement is still unspecific about what they improved in the Neural Engine. Since all we know about the Neural Engine comes from Apple papers or reverse engineering efforts ( https://github.com/hollance/neural-engine ), it’s plausible that they addressed some quirks to enable better transformer performance. They have written quite interesting papers on transformers on the Neural Engine:…

MLX doesn't use the neural engine still right? I still wish they would abandon that unit and just center everything around metal and tensor units on the GPU.

Wrt. language models/transformers, the neural engine/NPU is still potentially useful for the pre-processing step, which is generally compute-limited. For token generation you need memory bandwidth so GPU compute with neural/tensor accelerators is preferable.

Re: Apple M5 chip

#202

Post by butlike »

mumber_typhoon wrote: post #44

The M5 MacBook Pro still gets the Broadcom WiFi chip but the M5 iPad Pros get the N1 and C1X (Sweet). All in all, apple is doing some incredible things with hardware. Software teams at apple really need to get their act together. The M1 itself is so powerful that nobody really needs to upgrade that for most things most people do on their computers. Tahoe however makes my M1 Air feel sluggish doing the exact same task…

I think it's probably a play to get you to upgrade for the new GPU computational power. I _do_ think that what we're seeing (and marketed as AI) will be the future, but I don't think it will look like what we're seeing now. Whatever that future holds will require the upgraded capabilities of these new GPU architectures, and this being a reason for the subtle nudge to upgrade from Apple makes sense to me.

It feels very much like how I imagine someone living in the late 1800's might have felt. The advent of electricity, the advent of cars, but can't predict airplanes, even though they're right around the corner and they'll have likely seen them in their lifetime.

Re: Apple M5 chip

#203

Post by pivo »

throw_this_one wrote: post #129
toddmorey wrote: post #6

The modern Apple feels like their hardware teams way outperforming the software teams.

Their software is literally falling apart. ios26 was the biggest trash ive ever experienced from a company this big

How so? Seriously asking because it works fine for me.

Re: Apple M5 chip

#204

Post by bfrog »

The big win would be a linux capable device. I don't have any interest in mac os x but the apple m parts always seem amazing.

In theory this would be where qualcomm would come in and provide something but in practice they seem to be stuck in qualcomm land where only lawyers matter and actual users and developers can get stuffed.

Re: Apple M5 chip

#205

Post by chedabob »

nik736 wrote: post #2

This is only the base model, no upgrades yet for the Pro/Max version. The memory bandwidth is 153GB/s which is not enough to run viable open source LLM models properly.

My guess would be those are going into the rumoured OLED models coming out next year.

Re: Apple M5 chip

#206

Post by butlike »

lelandfe wrote: post #138
mumber_typhoon wrote: post #44

The M5 MacBook Pro still gets the Broadcom WiFi chip but the M5 iPad Pros get the N1 and C1X (Sweet). All in all, apple is doing some incredible things with hardware. Software teams at apple really need to get their act together. The M1 itself is so powerful that nobody really needs to upgrade that for most things most people do on their computers. Tahoe however makes my M1 Air feel sluggish doing the exact same task…

As a UI/UX nerd, it’s a coin flip on intentionality. I’ve been noticing so many rough edges to Apple’s software when it used to astound. iOS Settings search will flash “No Results” as you begin to type which is comically amateurish. The macOS menu bar control panels can’t be keyboard navigated... It’s just silly. I’ve been debating making a Tumblr-style blog, something like “dumbapple.com,” to catalogue all the dumb…

iirc, there's a setting to make the menu bar navigatable. you just need to "alt+tab" to it with some weird button combo, like Ctrl + Cmd + 1 or something.

Re: Apple M5 chip

#207

Post by coredog64 »

pxc wrote: post #149
samwillis wrote: post #114

Earlier quoted context omitted.

Software is very easy to bloat, expand scope, and grow to do more than really needed, or just to release apps that are then forgotten about. Hardware is naturally limited in scope due to manufacturing costs, and doesn't "grow" in the same way. You replace features and components rather than constantly add to them. Apple needs someone to come in and aggressively cut scope in the software, removing features and product…

> pare down products and features macOS has way too many products but far too few features. In terms of feature-completeness, it's already crippled. What OS features can macOS afford to lose?

I would say it's less about losing and more about focus. Identify the lines of business you don't want to be in and sell those features to a third party who can then bundle them for $1/$10/$20. A $2T company just doesn't care, but I would bet that those excised features would be good enough for a smaller software house.

(I have the same complaint about AWS, where a bunch of services are in KTLO and would be better served by not being inside AWS)

Re: Apple M5 chip

#209

Post by fooblaster »

zozbot234 wrote: post #201
fooblaster wrote: post #176

Earlier quoted context omitted.

MLX doesn't use the neural engine still right? I still wish they would abandon that unit and just center everything around metal and tensor units on the GPU.

Wrt. language models/transformers, the neural engine/NPU is still potentially useful for the pre-processing step, which is generally compute-limited. For token generation you need memory bandwidth so GPU compute with neural/tensor accelerators is preferable.

I think I'd still rather have the hardware area put into tensor cores for the GPU instead of this unit that's only programmable with onnx.

Re: Apple M5 chip

#210

Post by scbzzzzz »

butlike wrote: post #164
thomascgalvin wrote: post #126

Earlier quoted context omitted.

> The modern Apple feels like their hardware teams way outperforming the software teams. There aren't a lot of tangible gains left to be made by the software teams. The OS is fine, the office suite is fine, the entertainment apps are fine. If "performance" is shoving AI crap into software that was already doing what I wanted it to do, I'd rather the devs take a vacation.

There were a few things on that page that made me excited for the future of where computing is going, but I do think we're going to hit a "lull" in terms of exciting new features until some of the really futuristic stuff comes to pass. Who knows, maybe the era of "exciting computing" is over, and iteration will be a more pleasant and subtle gradient curve of improvements, over the earth-shattering announcements of yo…

True. I would like to hijack this thread and wante d to discuss what we want for software that is not present. For me. All i can think of is ondevice , al/ml ( photo editing, video editing etc ) and not the ones the current companies are trying hard shove down our throats.

May be steve is true. We don't know what we want until some one shows it .

Post reply on HN