Altera is back

[pcprogrammer]

They also knew to put "AI" into the announcement in some form or fashion - 

It's also secretely in the logo on the end.

The name should be ailtera. That will be 500K USD please for consulting.

Nah, because the French will then think it is a dish with garlic (ail) 

[Nominal Animal]

Wherever you see "AI" just replace it with "low precision vector dot products".

The funniest thing is when they start gushing about emergent behaviour when their model becomes large enough.

Apparently they've never heard of the combinatorial explosion, and are forever surprised by things like the birthday paradox, because actually understanding what is going on is so passé nowadays.  No wonder human brains are steadily shrinking.  Back to monke, I guess.

[Nominal Animal]

AI built in the fabric - does this statement makes sense to you guys?

🎶 Ai built the fabric 🎵
🎵 But ai did not builk the plebluplée 🎶
Or something.  So no.

[SiliconWizard]

Wherever you see "AI" just replace it with "low precision vector dot products".

The funniest thing is when they start gushing about emergent behaviour when their model becomes large enough.

Apparently they've never heard of the combinatorial explosion, and are forever surprised by things like the birthday paradox, because actually understanding what is going on is so passé nowadays.  No wonder human brains are steadily shrinking.  Back to monke, I guess.

Exactly. One issue with current "AI" approaches, which is really only about churning a lot of data through very simple algorithms, is that we can't know whatsoever how the models get a given answers from a given set of inputs. It's the epitome of the black box, and I guess our current fascination for this can only be explained by our immense stupidity.

[cfbsoftware]

AI built in the fabric - does this statement makes sense to you guys?

Sure - it must be the same fabric they used for the Emperor's new clothes.

[free_electron]

altera starts, partners with intel , intel sell pld division to altera, intel buys altera , intel spits out altera ... 'm okay

[glenenglish]

anyone here actively designing with Altera-i ?

[SiliconWizard]

altera starts, partners with intel , intel sell pld division to altera, intel buys altera , intel spits out altera ... 'm okay

Yeah, don't necessarily assume that it's all due to "strategic" reasons. Quite often, companies merging and splitting repeatedly is just done for optimizing tax, profits and expenditures. Purely financial stuff.

[Muxr]

altera starts, partners with intel , intel sell pld division to altera, intel buys altera , intel spits out altera ... 'm okay

Yeah, don't necessarily assume that it's all due to "strategic" reasons. Quite often, companies merging and splitting repeatedly is just done for optimizing tax, profits and expenditures. Purely financial stuff.

Yup. Intel is strapped for cash, to build out their fabs. They are prepping Altera for an IPO, same thing they did with Mobile Eye.

[tszaboo]

altera starts, partners with intel , intel sell pld division to altera, intel buys altera , intel spits out altera ... 'm okay

For the past two decades, everything (except cpu) Intel touched went broke. Microcontrollers, SSDs, the power management, IP phones, MP3 players. Everything it tries just shuts down.

[glenenglish]

and the reason I dont buy any Intel in anything I own , is what they did with Edison.

remember that ?  Lots of us designed long life embedded  products around a great platform

only for Intel to drop it like a stone  couple of years after release.

That's why I will never ever buy Intel, nor Intel owned Altera.

[SiliconWizard]

altera starts, partners with intel , intel sell pld division to altera, intel buys altera , intel spits out altera ... 'm okay

For the past two decades, everything (except cpu) Intel touched went broke. Microcontrollers, SSDs, the power management, IP phones, MP3 players. Everything it tries just shuts down.

I think that's not specific to Intel though, but to all very big companies that have become too unflexible to allow diversification.
And as we said above, they may never have even intended to make those ventures succeed. Often that's just a distraction and a way of circulating cash.

[floobydust]

Anything "AI" is attracting investors, who know nothing about the technology. NVIDIA stock is going nuts because of their AI IC's.

I think Intel is pathetically jumping on the bandwagon to boost their stock with this nothingburger press release, all about the graphic arts on the logo 
Let's give out the award for pathetic or is it pity to the old dinosaur. Nobody over there can innovate or keep up with technology.

[nctnico]

altera starts, partners with intel , intel sell pld division to altera, intel buys altera , intel spits out altera ... 'm okay

For the past two decades, everything (except cpu) Intel touched went broke. Microcontrollers, SSDs, the power management, IP phones, MP3 players. Everything it tries just shuts down.

I think that's not specific to Intel though, but to all very big companies that have become too unflexible to allow diversification.

I don't quite agree. Other semiconductor companies typically are good at absorbing integrating product lines from semiconductor firms they bought. Look at Analog, Microchip, TI, NXP for examples. OTOH Intel and AMD have been excessively bad at doing such integrations even those have designing & producing semiconductors as a core business.

[JPortici]

They also knew to put "AI" into the announcement in some form or fashion - 

It's also secretely in the logo on the end.

Midjourney: Design a minimalistic inclusive logo for the brand "altera" and include ai somehow

[KE5FX]

They also knew to put "AI" into the announcement in some form or fashion - 

It's also secretely in the logo on the end.

Midjourney: Design a minimalistic inclusive logo for the brand "altera" and include ai somehow

Honestly, GPT4 didn't do a bad job with your prompt:

[SiliconWizard]

The logo looks like a nasty bug.

[KE5FX]

[e0ne199]

Anything "AI" is attracting investors, who know nothing about the technology. NVIDIA stock is going nuts because of their AI IC's.

I think Intel is pathetically jumping on the bandwagon to boost their stock with this nothingburger press release, all about the graphic arts on the logo 
Let's give out the award for pathetic or is it pity to the old dinosaur. Nobody over there can innovate or keep up with technology.

AI nowadays is starting to look like a snake oil..

[KE5FX]

I made fun of it above, but for the record: that's not snake oil.  Snake oil can't kneecap a 9-dan Go master.  AI will get better, while the people who designed the real Altera logo won't. 

See also the question of whether transformer networks can "code."  Yes, and they're not great at it... but they will get better. 

The first couple of time derivatives are what matter when it comes to things like this, not the current state.  Everything is about to change, and Intel isn't entirely wrong in acknowledging that.  At least from a marketing perspective.

[Nominal Animal]

See also the question of whether transformer networks can "code."

They do not generate anything new.  They can only combine parts from their original dataset; not as in copy-paste, but in "this fits here" fashion.
The end result may look novel, but if the standard US abstraction-filtration-comparison test is applied to both (the "novel" result, and all of the original dataset), you always find 1:1 correlation.  This is why they are transformer networks, and not general intelligence.  It is also why transformer networks cannot author patents in the USA, according to recent judgments.

There are very strong arguments for general intelligence requiring the ability to perceive and manipulate the environment (environment including other members of similar species; language being just a tool for manipulating others using the same language).  This means that if we survive long enough as an intelligent species, general intelligence will grow out of either embodied robots, or from simulations, not from LLMs or transformer networks.

As to games (9-dan Go master), a pure imperative program can do that, if we were smart enough to write one.  We did that for chess already; Go just has much more permutations of the game state.

[KE5FX]

They do not generate anything new.  They can only combine parts from their original dataset; not as in copy-paste, but in "this fits here" fashion.
The end result may look novel, but if the standard US abstraction-filtration-comparison test is applied to both (the "novel" result, and all of the original dataset), you always find 1:1 correlation.  This is why they are transformer networks, and not general intelligence.  It is also why transformer networks cannot author patents in the USA, according to recent judgments.

There are very strong arguments for general intelligence requiring the ability to perceive and manipulate the environment (environment including other members of similar species; language being just a tool for manipulating others using the same language).  This means that if we survive long enough as an intelligent species, general intelligence will grow out of either embodied robots, or from simulations, not from LLMs or transformer networks.

As to games (9-dan Go master), a pure imperative program can do that, if we were smart enough to write one.  We did that for chess already; Go just has much more permutations of the game state.

There's so much here that's "not even wrong" that it's difficult to refute.  Stick around, things are going to get interesting.

[dferyance]

They are the company that came up with the brilliant name of Max+Plus II. Let's call it "Max" so people know how great it is. But we need more, how about max plus? Naw, needs two pluses, Max+Plus. That still doesn't pack enough punch, we'll call it Max+Plus II. But the 2 won't be the version number, that comes after.

[Nominal Animal]

There's so much here that's "not even wrong" that it's difficult to refute.  Stick around, things are going to get interesting.

I'm familiar with these models, how they work, and the math involved.  When I first studied them, Kohonen nets were the rage.  (I investigated them for OCR, as that was still a "hard task" back then, especially with new, untrained font styles.)

What you see as miraculous performance, I see as a logical result of the size and training done on the networks.

[Altera] are the company that came up with the brilliant name of Max+Plus II.

It would be funny to create some new programming language, and name it See+Plus.  Anyone trying to do search engine optimization on it would tear out their hair for sure.

[KE5FX]

You should spend some time with GPT4 before drawing conclusions based on 40-year-old concepts like Kohonen nets.  (Had to look that up, they aren't even mentioned in the current deep-learning texts.)  Given your assertion that these networks don't create anything new, you may be running a bit behind the times.  There's some really cool stuff being done.

Those logos I just posted suck, but they are new. Arguing that the output of these models isn't 'new' forces you into a God-of-the-gaps scenario, one that will force you to confront the fact that humans aren't great at coming up with 'new' stuff, either.   There are other threads better suited to this sort of argument, admittedly.