NotebookCHECK - Notebook Forum

It would be nice to have some useful content in this article. Is the "AI" going to run locally? That is very feasible on Apple silicon (on general x86 chips, not so much).

I wish and Apple fans would stop thinking that Apple is the only company in the world.

Microsoft did this with Visual Studio a few years ago and even VSCode has a similar plug in.

Apple is not revolutionising anything here - they're just playing catch-up.

And it's a tacit admission that Apple was last to the party.

(Also, why does this article read more like a press release than an actual article?)

Quote from: davidm on February 16, 2024, 18:44:51It would be nice to have some useful content in this article. Is the "AI" going to run locally? That is very feasible on Apple silicon (on general x86 chips, not so much).

Oh please. I have lots of friends who are running Stable Diffusion and similar genLLMs on their x86 PCs and that's doing generative graphics. Microsoft integrated AI predictive coding into Visual Studio and VSCode almost a year ago and it runs just fine (well, if you like genLLM code assists).

Quote from: The Werewolf on February 17, 2024, 00:09:19Microsoft did this with Visual Studio a few years ago and even VSCode has a similar plug in.

a) there are "plugins" for everything, including xcode
b) you can not use those "plugins" in any real corporate dev environment except your in-house pet projects
c) apple is rumored to do public AI that owns its training data and is actually usable in real dev, which is different from existing AIs trained on public data

Quote from: The Werewolf on February 17, 2024, 00:12:58I have lots of friends who are running Stable Diffusion and similar genLLMs on their x86 PCs and that's doing generative graphics

yeah ask them to generate 8K image on x86

why are you even commenting, you have no idea about the topic

If anyone comes across this, the point is full scale *local* AI. Not from the cloud, not on a subscription, not with any privacy or secrecy concerns. The article didn't go to that depth (my original point), but that is the direction and where Apple has some strengths, since with their unified memory design Apple silicon is years ahead of x86 for this type of application.
People can already run good quality LLMs on any Apple system from M1 up that would be impractical on a base x86 system, though more than 16GB RAM is essential, more than 32GB helpful.

Quote from: davidm on February 18, 2024, 13:03:45full scale *local* AI. [...] where Apple [...] with their unified memory design Apple silicon is years ahead of x86 for this type of application.

Your statement is absolutely wrong! The truth is: It depends on the (e.g., AI) software, its usage and for both on the needed amounts of RAM, VRAM or Unified Memory. If any of the volatile storage is insufficient for the software in its usage, execution is very slow or impossible. If volatile storage is sufficient for the software in its usage, it depends on a) whether the software is available for the hardware, OS and libraries, b) the kind of cores are particularly suitable for the software and c) the libraries for the cores are particularly suitable for the software.

There are examples of interesting versions of AI software, such as LLM, for which insufficient VRAM prevents execution while Unified Memory, if it is sufficiently large, enables execution. There are also Nvidia server GPUs or servers with many such GPUs with enough VRAM for such LLMs.

There are examples of AI softwares, such as Katago, that run mainly on RAM, hardly need VRAM, profit from Nvidia tensor and CUDA cores and profit from Nvidia libraries for these cores. Such softwares are dozens of times faster on x64 than Apple M, which lacks TDP, suitable cores and suitable libraries for such suitable cores. In your words, x64 with Nvidia GPUs is many years ahead of Apple M for such softwares.

For the citation in my previous message, I forgot to point out for the user A:

Apple propaganda detected!

Apple is the best in everything. EOD.

Quote from: RobertJasiek on February 18, 2024, 15:42:15Apple propaganda detected!

Basically this, but Apple. (https://imgur.com/a/rqmEYif)

lmao poor apple haters still can't adopt the reality.
poor go cheater even thinks his go cheating app is an ai lmao and calls llms "interesting version of ai".
nbc really has to do something about these low-quality shitposters in comments, last two messages are just pure hate with zero arguments.

Quote from: lmao on February 18, 2024, 16:07:04lmao poor apple haters still can't adopt the reality.
poor go cheater even thinks his go cheating app is an ai lmao and calls llms "interesting version of ai".
nbc really has to do something about these low-quality shitposters in comments, last two messages are just pure hate with zero arguments.

Retard level 9000. Congrats on achieving it.

Quote from: Neenyah on February 18, 2024, 16:40:10Retard level 9000. Congrats on achieving it.

third shitpost in a row from the same two personas full of hate
no offense taken lmao dont worry guys we know that its your envy to people who achieved something real in their life and have money for apple tech makes you do this

Quote from: lmao on February 18, 2024, 16:48:42

Quote from: Neenyah on February 18, 2024, 16:40:10Retard level 9000. Congrats on achieving it.

third shitpost in a row from the same two personas full of hate
no offense taken lmao dont worry guys we know that its your envy to people who achieved something real in their life and have money for apple tech makes you do this

Admit it, you love Apple more than your own mother. Only that would explain that ongoing mental breakdown of yours. Resume therapy please.

Quote from: Neenyah on February 18, 2024, 17:01:25Admit it, you love Apple more than your own mother. Only that would explain that ongoing mental breakdown of yours. Resume therapy please.

lmao what i love is watching people triggered by word "apple" and running around trying to insult people because of their hate
nbc really has to do something about you low-iq low-effort commenters

Quote from: lmao on February 18, 2024, 16:07:04lmao poor apple haters still can't adopt the reality.

El MAO - apple user detected! - haha! El MAO EL MAO

"lmao" defending Apple: "8gb is enough" -> ELMAO ELMAO

low-effort bro is supporting himself by pretending he is someone else lmao
grabbing popcorn

Quote from: lmao on February 18, 2024, 16:07:04lmao poor apple haters

The language of an Apple defender.

Quotepoor go cheater [...] low-quality shitposters

Insults by an Apple defender.

Quotehis go cheating app

You make the same mistake as the user A: it is not my software but it is software that also I use.

A software generating Go moves and evaluations of Go moves and Go positions is, first of all, just this: a software generating Go moves and evaluations of Go moves and Go positions. Usually, users of such software use it for these purposes. Only if such use occurs while playing (especially tournament or rated games), it can be cheating. Not only do I not do this but I have also written now adopted Go tournament rules that prohibit AI use while playing and been similarly active for decades.

Similarly, other AIs used for other purposes, such as programming or media creation, are not cheating softwares per se. Cheating is if one uses them while pretending not to use them or while performing an activity that is supposed to be without help by third persons or uneligible tools.

Quotecalls llms "interesting version of ai".

You misunderstand. I call only those LLMs interesting that have a sufficient quality. Garbage does not impress me.

Quoteenvy to people who achieved something real in their life and have money for apple tech makes you do this

Widen the scope of your understanding of the world: it is possible to have game-related jobs. I use aforementioned AI daily to make money. (Needless to say, without using Apple tech for this purpose.)

***

Furthermore, learn that I have mentioned software I understand the best. There must be countless AIs for professional uses but I can evaluate the other AIs easily because I do not use them. I can, however, evaluate Go AIs because I use them and have informed myself about them.

Quote from: RobertJasiek on February 18, 2024, 18:46:23

Quote from: lmao on February 18, 2024, 16:07:04lmao poor apple haters

The language of an Apple defender.

I like it how poor is his most creative "insult", lol. Little brain can't think about anything else. Bill Gates is also poor though, that guy can't afford a MacBook and enjoy all benefits of memeOS 😥

Quote from: RobertJasiek on February 18, 2024, 18:46:23Apple defender

Quote from: RobertJasiek on February 18, 2024, 18:46:23Apple defender

say something smart already i am bored lmao
i dont see any "apple defenders" running around and posting low-iq crap in intel and amd news like you hater bros do all the time when you are triggered by the word "apple" in news headlines

Quote from: RobertJasiek on February 18, 2024, 18:46:23I use aforementioned AI daily to make money

HAHA i already figured you are cheating in online tournaments lmao
www.nordicgodojo.eu/post/33/how-to-limit-cheating-in-online-go

Quote from: RobertJasiek on February 18, 2024, 18:46:23evaluate Go AIs

it is not ai, at least learn the terminology before posting your useless low-effort comments under unrelated news about xcode developer ai you've never seen and don't really have any idea about.
who the f*** even cares what you have to say on unrelated topics, 4090 you don't have or your go cheating habits and programs you use for it haha

Quote from: Neenyah on February 18, 2024, 19:32:14I like it how poor is his most creative "insult"

maybe it is because i am not here to come up with creative ways to insult people and spread hate like you do lmao
and poor is not an insult, it's just a fact about the origins of your behavior

Katago plays Go very much stronger than the strongest human Go players who have spent their lifes on studying Go. This is sufficient reason for calling Katago an AI.

When you refer to Antti Törmänen's page Nordic Go Dojo and say "you are cheating in online tournaments", you hit ambiguity of English: you do not mean me but you mean the generic "you", that is, you have found a description that there are some cheating players in online tournaments. Indeed, there are. Not me. Although your citation context tries to suggest the opposite. For others, it is easy to see through your casting doubts scheme by noticing that the referred, linked page is not about me at all.

Quote from: RobertJasiek on February 18, 2024, 21:49:49Katago plays Go very much stronger than the strongest human Go players who have spent their lifes on studying Go. This is sufficient reason for calling Katago an AI.

ai is not about being superhuman, it is about thinking as human lmao.
your app has neural network only for one specific task unlike whats considered a real ai which is doing everything using neural networks.
why you keep commenting for the sake of commenting wth you probably never knew anything about ai except first link in google

Quote from: RobertJasiek on February 18, 2024, 21:49:49you hit ambiguity

lmao i do not hit any ambiguity just showing you i am well aware what those go engines are used for and how they are used to earn money

Quote from: davidm on February 18, 2024, 13:03:45since with their unified memory design Apple silicon is years ahead of x86

They are ahead but this isn't the technical reason. UMA designs have existed decades before Apple started building silicon. Anything with an iGPU is UMA by definition.

Quote from: davidm on February 18, 2024, 13:03:45People can already run good quality LLMs on any Apple system from M1 up that would be impractical on a base x86 system, though more than 16GB RAM is essential, more than 32GB helpful.

I've made this argument before but I'll repeat it here again. Apple, much like Nvidia charge a lot for memory. The base / entry Mac's with 8GB of ram are useless for such applications. For the same price of a 16GB Mac you can easily get a 32GB windows machine. For the price of a 32GB Mac you can get a 64GB windows machine.

It's true once you start going past 64 GB, to like 96GB-128GB there isn't really anything in the x86 landscape that can compete. But do you know how much such systems cost? $5000. So yes, for that kind of money you can beat an x86 system. I guess you could make the argument "well just stop being poor and low IQ then" but it's irrelevant because who do you is the target audience for such a system? Certainly not amateur hobbyists. The kind of people who require such systems work in big tech firms or startups and work on this professionally / get paid for a living for it. I doubt these people even pay for their work machines. The machine could cost $25k and it still wouldn't matter or affect them since their company is paying for it all.

So yeah, I can see why so many are recommending MBP for LLMs. It makes complete sense if you're getting them for essentially free or are running a corporation/company. For the average end-user tho who isn't so into AI, it doesn't make as much sense.

Quote from: lmao on February 18, 2024, 22:27:33ai is not about being superhuman, it is about thinking as human

The scope of AI differs from weak to strong. From barely human-like in specific tasks to - in the future - more capable than mankind.

Quoteyour app has neural network only for one specific task unlike whats considered a real ai which is doing everything using neural networks.

Welcome to learning that there are different AIs with different scopes. Currently most AIs are task-specific. General AIs equivalent to all capabilities of the human brain do not exist yet. Different AIs achieve their artificial intelligence by different means, such as neural nets, deep learning, reinforcement learning etc. Current abstract game AIs use, in particular, all three of these. There are also AIs that use none of these.

Quotei am well aware what those go engines are used for and how they are used to earn money

There are many more uses of Go AIs than currently you imagine. Currently, the most frequent use by most users is hobby study attempting to improve by playing against the AI or letting it statistically quick-analyse one's own games played earlier. My major use differs: I use the AI to explore moves, variations and positions to help me studying theory of opening or middle game tactics and strategy very much faster than without using AI. (For endgame theory, my mathematical approach is much faster than AI.)

Quote from: AAGUC on February 18, 2024, 23:04:19It's true once you start going past 64 GB, to like 96GB-128GB there isn't really anything in the x86 landscape that can compete. But do you know how much such systems cost? $5000. So yes, for that kind of money you can beat an x86 system.

It depends on whether >64GB must be RAM or VRAM or alternatively Unified Memory.

If it can be RAM, using 128GB is easy and 192GB possible even for ordinary x64 desktops with suitable mainboard, CPU and RAM kits, which are ~€200 per 64GB so ~€600 for 192GB. With Threadripper (very expensive), much more RAM is possible.

If it must be VRAM, it becomes impractical with 4 to 8 water-cooled GPUs in a desktop in the €5.000 ~ €30.000 price range, which is similar to expensively configured Macs (MBPs are only at the lower end of such prices, but also below in speed.) Or, instead of just impractical, it becomes extremely expensive by using two or more Nvidia workstation or server GPUs, each for €10.000 ~ €100.000.

Where MBPs excel in €4.000 ~ €9.000 price range is to enable VRAM storage equivalent in the 64 ~ 192GB range (Or is 128GB the limit? I forgot the exact number.) in the notebook form factor. Such MBPs have two prominent use cases: a) medium quality LLMs; b) demanding video editing (because Apple has gimped bandwidth of all less expensively priced M3 MBPs compared to M2 MBPs, so one should prefer M2 to M3, see Tech Notice and Just Josh).

Quote from: RobertJasiek on February 19, 2024, 08:13:05Welcome to learning that there are different AIs with different scopes

lmao what part of "everything has to be done with neural networks for something to be considered an ai" exactly evades your understanding. 'zero' game engines are ai, your crap isnt, get over it

Quote from: RobertJasiek on February 19, 2024, 08:13:05There are many more uses of Go AIs than currently you imagine

HAHAHA yeah right

Quote from: RobertJasiek on February 19, 2024, 08:13:05My major use differs: I use the AI to explore moves, variations and positions to help me studying theory of opening or middle game tactics and strategy very much faster than without using AI

HAHAHAHA what a nice way to describe online tournament cheating

Quote from: RobertJasiek on February 19, 2024, 08:37:09Where MBPs excel in €4.000 ~ €9.000 price range is to enable VRAM storage equivalent in the 64 ~ 192GB range (Or is 128GB the limit? I forgot the exact number.) in the notebook form factor. Such MBPs have two prominent use cases: a) medium quality LLMs; b) demanding video editing (because Apple has gimped bandwidth of all less expensively priced M3 MBPs compared to M2 MBPs, so one should prefer M2 to M3, see Tech Notice and Just Josh).

lmao ranting again about something you dont know anything about, every sentence in your post is a pile of bullshit.
every m3 is better than corresponding m2 for ai, community did all tests already, read the fucking reddit instead of your intel shilling youtubers. its not only about ram bandwidth lol, it's a balance between gpu speed and ram bandwidth.
nbc really has to deplatform you low-effort low-quality commenters, you are a scarecrow for healthy community

Quote from: AAGUC on February 18, 2024, 23:04:19Anything with an iGPU is UMA by definition.

lmao one more expert

Quote from: lmao on February 19, 2024, 11:35:10"everything has to be done with neural networks for something to be considered an ai"

Your too restricted understanding of what can be AI.

Quote'zero' game engines are ai

Wikipedia EN: "Artificial intelligence (AI) is the intelligence of machines or software, as opposed to the intelligence of other living beings, primarily of humans. It is a field of study in computer science that develops and studies intelligent machines. Such machines may be called AIs."

This definition acknowledges that the term AI is used for two major purposes: a property of machines / software; a field of study in computer science. We are discussing the former kind. That webpage does not define 'intelligence' though but only discusses various AI types. Remarkably, that webpage does write the following about one AI type:

"superhuman play and analysis in strategy games (such as chess and Go)"

There you have it. Even the weak Wikipedia EN webpage on AI has it.

Quote

Quote from: RobertJasiek on February 19, 2024, 08:13:05My major use differs: I use the AI to explore moves, variations and positions to help me studying theory of opening or middle game tactics and strategy very much faster than without using AI

what a nice way to describe online tournament cheating

It would be, if it were a) online and b) during tournament game play.

However, I only use the AI a) offline and b) never during tournament game play.

Quotecrap
[...]
every sentence in your post is a pile of bullshit.
[...]
fucking
[...]
scarecrow

The Apple defender advertising for Apple by insulting others.

Quoteevery m3 is better than corresponding m2 for ai,

1) Faster: probably.

2) Efficient: I am not sure about M2 versus M3 but reviewers have indicated M1 as more efficient.

3) In the context for bandwidth, I have written about video editing and have not written about AI. The Youtube tech reviewers Tech Notice and Just Josh have found that M2 MBP is better than M3 MBP for video editing, unless one buys the most expensive M3 MBP that is not crippled in bandwidth by Apple in an attempt to upsell for a few excess thousand euros.

Quoteit's a balance between gpu speed and ram bandwidth.

Sure. But without sufficient bandwidth, there is no balance.

Quotenbc really has to deplatform you low-effort low-quality commenters

Whom do you think a social medium restricts first?

Somebody like you using insults, such as

crap
[...]
every sentence in your post is a pile of bullshit.
[...]
fucking
[...]
scarecrow

or somebody like me not using insults?

Quote from: RobertJasiek on February 19, 2024, 12:14:04Your too restricted understanding of what can be AI.

you are simply arguing for the sake of arguing despite you lost this debate long ago

Quote from: RobertJasiek on February 19, 2024, 12:14:04"superhuman play and analysis in strategy games (such as chess and Go)"
There you have it. Even the weak Wikipedia EN webpage on AI has it.

lmao exactly what i was talking about, they mean 'ZERO' series game engines that work using only neural network
and your crap isnt ai, its just a normal game engine with an added evaluation function using neural network, get over it and stop pretending you understand anything

Quote from: RobertJasiek on February 19, 2024, 12:14:041) Faster: probably.
2) Efficient: I am not sure about M2 versus M3 but reviewers have indicated M1 as more efficient.

pile of bullshit lmao, m3 have about the same power consumption for more compute power

Quote from: RobertJasiek on February 19, 2024, 12:14:04The Youtube tech reviewers Tech Notice and Just Josh have found that M2 MBP is better than M3 MBP for video editing, unless one buys the most expensive M3 MBP that is not crippled in bandwidth by Apple in an attempt to upsell for a few excess thousand euros.

who cares about your cherry picked reviewers with failed results lmao
you are a hater and you only reference info that is your echo chamber, i am pretty sure you will be running around telling everyone s24u has better battery life than 15promax referring to some youtubers who claimed it, despite gsmarena already disproved the fact

Quote from: RobertJasiek on February 19, 2024, 12:14:04or somebody like me not using insults?

lmao i dont care
insults are not a scarecrow for normal community, low-quality low-effort haters like you are
nbc will never create a healthy community until they deplatform you nikob neenyah and the likes, who just spread hate and misinformation lmao
like what are you doing in apple discussion you never even had a macbook lmao

Apple and you as one of its fans fear nothing more than free speech abouts its products. The fear is so great that you desire censorship and assign "hater" attributes to critics. History has taught that giving in to censorship and insults results in totalitarianism. Therefore, no number of calls for censorship or insults stops brave people.

As to ZERO series games, it is unclear what you mean. Zero-sum games? Deepmind game AI engines, such as AlphaGoZero, starting from only the game rules without prior human knowledge on tactics and strategy? Some other kind of games?

And yes, current Go AI engines use deep neural nets etc. (I have not said otherwise when speaking, more generally, about any AI.)

Quote from: RobertJasiek on February 19, 2024, 13:26:16Apple and you as one of its fans

when i see yet another local low-effort commenter appealing to "you are a fan" rhetoric i immediately skip whole low-iq paragraph lmao its clear the guy lost the debate
ever heard me saying anything about "x86 fans"? get you s*** together and learn to admit you were wrong

Quote from: RobertJasiek on February 19, 2024, 13:26:16As to ZERO series games, it is unclear what you mean. Zero-sum games? Deepmind game AI engines, such as AlphaGoZero, starting from only the game rules without prior human knowledge on tactics and strategy? Some other kind of games?

HAHAH question from "ai expert" posting on every ai-related article pretending he knows a thing or two
"what is a zero series of ai" wth how low can it go
i dunno man make your "educated guess", you are the "expert" right? lmao

Quote from: RobertJasiek on February 19, 2024, 13:26:16And yes, current Go AI engines use deep neural nets etc.

for the third time - useing small neural net for a small part of the task doesn't make it ai lmao
ai has to use neural nets for everything aka "think" using exactly the same processes as human
or maybe you think human is going through 24359024850439358 board combinations one by one and building decision trees in their "RAM" lmao

Quote from: lmao on February 19, 2024, 14:45:11ai has to use neural nets for everything aka "think" using exactly the same processes as human

Shows you do not understand modern AIs for abstract games.

Pure neural net Go programs reached intermediate human strength. Afterwards, Go programs have combined an increased number of different kinds of data structures (Not all of which are neural nets! E.g., some of them are tree search.) and algorithms and therefore surpassed top human strength.

LOL, troll.

Quote from: RobertJasiek on February 19, 2024, 21:41:09LOL, troll

a day of writing low-effort low-iq comments just to fail all your arguments and call someone a troll because he knows more than you?
nbc really has to get rid of the likes of you to make this a normal community