Apple Updates MacBook Pro With M6 Chips (theverge.com) 40
Ahead of the touchscreen MacBook Pro expected to debut tomorrow, Apple has refreshed its standard 14-inch MacBook Pro with the new M6 processor, promising up to 20% faster multithreaded performance, improved power efficiency, and more powerful on-device AI capabilities. The Verge reports: Announced this morning through a press release, the new 14-inch MacBook Pro features three Thunder 4 USB-C ports that all support charging and DisplayPort connectivity, a MagSafe 3 charging port, full-size HDMI, and an SDXC memory card slot accommodating photographers. Pricing for the new M6 version of the 14-inch MacBook Pro starts at $1,999 with 16GB of memory, the same as last year's M5 model. Preorders start on October 28th and shipping is expected to begin on November 6th.
The new M6 processor is Apple's first 2nm chip with a bump to a 12-core CPU, up from 10 cores in the M5, and the company says it offers up to 1.2x faster multithreaded performance when compared to its predecessor. The M6 also features a Dual 16-core Neural Engine to handle heavy on-device AI tasks, plus an upgraded 12-core GPU with a Neural Accelerator in each core. In addition to performance improvements, Apple says its M6 processor is more power efficient.
The new M6 processor is Apple's first 2nm chip with a bump to a 12-core CPU, up from 10 cores in the M5, and the company says it offers up to 1.2x faster multithreaded performance when compared to its predecessor. The M6 also features a Dual 16-core Neural Engine to handle heavy on-device AI tasks, plus an upgraded 12-core GPU with a Neural Accelerator in each core. In addition to performance improvements, Apple says its M6 processor is more power efficient.
Re: (Score:2)
Apple and the undocumented hardware registers :o (Score:2)
“If we try to describe this feature and how the attackers took advantage of it, it all comes down to this: they are able to write data to a certain physical address while bypassing the hardware-based memory protection by writing the data, destination address, and data hash to unknown hardware registers of the chip unused by the firmware.”
Re: (Score:1)
Your point being...?
https://www.csoonline.com/arti... [csoonline.com]
Re: (Score:2)
--
if thou gaze long into an abyss, the abyss will also gaze into thee.
Re: (Score:2)
There's nothing at the end of that line of reasoning. You can quite easily apply this to any processor ever in existence, past or future. "How do I know that won't have in the future?" You don't. And even if you knew on day one that an undocumented register existed in a chip, it tells you exactly nothing about future exploitability, and advises you not at all on whether or not to purchase it. I'm sure there are lots of undocumented "features" on chips from the past.
Re: (Score:2)
FYI, this included some items I was grammatically templating that used triangle brackets because I forgot slashdot doesn't sanitize inputs, but permits raw html.
"How do I know that (insert processor) won't have (insert condition) in the future?" You don't.
Re: (Score:2)
An undocumented register ... provides a backdoor ...
Hmkay!
Math (Score:4, Interesting)
Don't ignore the "UP TO" qualification. (Score:2)
They added 2 cores but the whole thing is still limited by thermal design limitations - whoop de doo.
Re: (Score:2)
The benchmarks already show its faster, throttled or not. So your comment is a big whatever.
Variables (Score:2)
If you're going to Math then you need to not omit variables. What's the thermal envelope change. If there's no increase in instructions per watt, and if the previous chip is (like all laptop chips) thermally limited, then the additional cores don't linearly scale. The burst performance of the chip may be 1.2x on cores alone, but can it burst for the same duration or does it clock down sooner?
To Math this you'll need to include about 20 additional variables. Cooling, efficiency, changes in the interconnect,
Heavy (Score:2)
"The M6 also features a Dual 16-core Neural Engine to handle heavy on-device AI tasks"
If it's supposed to handle heavy on-device AI tasks, why does it only have 16GB? Womp womp.
Re: (Score:2)
Maybe it only needs like a few GB for a tiny gemini or tts model..
Re: (Score:2)
Maybe it only needs like a few GB for a tiny gemini or tts model..
Of course it does, but that's not "Heavy" AI tasks. Lightweight model, lightweight task.
"Heavy" is not "Size" (Score:1)
Of course it does, but that's not "Heavy" AI tasks. Lightweight model, lightweight task.
"Heavy" refers to the amount of data being processed and the speed of that processing. Not the "Size" of the model.
Re: (Score:2)
"Heavy" refers to the amount of data being processed and the speed of that processing. Not the "Size" of the model.
I disagree. It's not heavy AI processing if you're not doing a lot of processing per request, which only happens when you have a large model.
Re: (Score:2)
Apple uses on-device AI for lots of things that are not chatbots. Video, audio and image processing can be computationally intensive with fairly small models.
Re: "Heavy" is not "Size" (Score:3)
All of which cannot be described as "heavy". You didn't read the summary or any of the comments in the thread to which you replied, where the same exact discussion was already had. Are you confused by RMBs as well?
Re: (Score:2)
You've got a shovel and you're determined to dig hey? Well, have at it!
Re: (Score:2)
The size of the app doing the processing is trivia.
Re: (Score:1)
"Heavy" refers to the amount of data being processed and the speed of that processing. Not the "Size" of the model.
I disagree. It's not heavy AI processing if you're not doing a lot of processing per request, which only happens when you have a large model.
Sometimes heavy is defined in terms of response time, in particular real-time or near real-time processing. That's why in computer science complexity can be measure in terms of space or in terms of time.
Re: (Score:2)
That's nothing compared to the fact that just plain lacks the necessary hardware to perform tensor math. Running stable diffusion on it is likely going to be pretty slow, though they probably mean shitty apple intelligence.
I'd say you should use your poo-agra falls attack on it, but it's already shitty.
Re: (Score:2)
If it's supposed to handle heavy on-device AI tasks, why does it only have 16GB? Womp womp.
Because it's a bad look if the least-expensive version of your laptop costs $2200.
Re: (Score:2)
Memory is expensive right now, but not THAT expensive. That's just apple overcharging for memory and storage, as it's already known for.
Not all AI models take a lot of RAM (Score:2)
If it's supposed to handle heavy on-device AI tasks, why does it only have 16GB? Womp womp.
There is more to AI models than LLM. There are small specialized models that run on Apple Watches, let alone a 16GB M6 based mac.
Re: (Score:2)
There are small specialized models that run on Apple Watches
Those aren't "Heavy" AI tasks. All that time you spent supporting child rapists has broken your brain to the point it's now not good for anything else, either.
Re: (Score:2)
The size of the model has nothing to do with the size of a task.
Simple speaking, you run similar tasks thousands of times ... the parts of the model that are not relevant: never gets swapped in.
More concrete: you load a model that is specific for your task. Worst case it only can find cats in pictures, and gives you the x1,y1 - x2,y2 coordinates in the picture where it finds cats ...
Such a model is probably not even 100NM.
Re: (Score:2)
The size of the model has nothing to do with the size of a task.
Obviously false as written
Re: (Score:2)
Nitpicking all the way down to the last response on your comment, are you not ?
Re: (Score:2)
Nitpicking all the way down to the last response on your comment, are you not ?
Welcome to Slashdot, YMBNH
Re: (Score:2)
Believe it or not, I am not new here. But contrary to your belief, not everyone is an asshole of your magnitude, even on Slashdot.
Re: Not all AI models take a lot of RAM (Score:2)
True. Many are much bigger assholes.
Re: (Score:2)
There are small specialized models that run on Apple Watches
Those aren't "Heavy" AI tasks.
"Heavy" refers to the amount of data being processed and the speed of that processing. Not the "Size" of the model.
Re: (Score:2)
Those aren't "Heavy" AI tasks.
They most definitely can be. The term "heavy" applies to scale, reasoning depth, and total workload. E.g. Image / video generation is defined as a "heavy" AI task, regardless if you use a model that can run locally in under 2GB of RAM or not. Likewise a large LLM model may be "light" if the reasoning depth is low.
Before insulting people it helps to have a fucking clue. That will help you look more intelligent, but it won't fix the fact you speak like an arsehole after burrito night.
Re: Not all AI models take a lot of RAM (Score:2)
You're more cowardly than an AC as you run away like a little bitch every time it's proven you were wrong. Not impressed by your sad attempt at cock swallowing to be relevant. Tim Apple will never give you gold.
Re: (Score:2)
No, he simply understands that it's pointless to wresle with a pig. You both get dirty, and the pig likes it.
Apple haters will resort to any travesty of logic and reason to rationalize their hatred. A company can be dispecable for many reasons, and their technology can still have technical merit. But blind haters have simple, binary minds. It's always black or white, all or nothing, good or evil.
Anyone can see from the tone of your post that your opinion of Apple is based purely on emotion, an nothing else.
Re: (Score:2)
Anyone can see from the tone of your post that your opinion of Apple is based purely on emotion, an nothing else.
That's what a stupid fuck might think based on their Apple worship.
I have loads of experience with Apple. That's why I despise them now, that and their sucking Trump's cock with a golden icon donation on the international stage. Feel free to pay them to abuse you and shit on our country.
Wow, the TDS comes out even in Apple convo (Score:2)
That's what a stupid fuck might think based on their Apple worship. I have loads of experience with Apple. That's why I despise them now, that and their sucking Trump's cock with a golden icon donation on the international stage. Feel free to pay them to abuse you and shit on our country.
LOL, the TDS even comes out in an Apple conversation. Man, Trump truly lives in your head, the discomfort that produces must drive you crazy. Explains a lot.
Re: (Score:2)
Swap, baby!