Intel Confirms: Arc Mobile Rollout Facing Delays, Desktop Debut On Track for Q2-2022

[kapone32]

This will be like their last GPU launch too little, too late with little uptake with independent consumers.

 

[mechtech]

It's not your foundry Intel, it's the TSMC, get in line like everyone else.

 

[80251]

Am I the only one that is having trouble keeping track of these things ? ARC, Alchemist 3-series, 5-series, 7-series, DG1 Xe Iris, Ponte Vecchio, you'd never guess these things are related to each other in some way in a million years as an average consumer. Not to mention they used to have alternative names for each one of those types of GPU for some reason like Xe-LP, Xe-HPC, etc, what a nightmare.

I wonder if it's on purpose so that you're always confused about which of them gets delayed.

Perhaps Intel is engaging in the ages old game of: if you can't dazzle 'em with data, baffle 'em with bullsh!t.

Is the Intel Arc (aka. Alchemist?) an Intel only franchise or will there be AIB's?

 

[thestryker6]

Q2 means pushed back to December.

Honestly, nobody is surprised at this point. Intel have been postponing Xe Arc dGPU launch by about 1 year per year of actual elapsed time since, uh, 2018?

I remember Intel demoing Xe dGPU prototype for "imminent launch" at CES 2018, with Raja comparing it to a 1050Ti/1060 at the time. It was running Destiny 2, something I remember because I was playing it a lot in 2017.

Here we are, four years later, getting ready for the Xe dGPU's "imminent launch".

Does anyone remember the Infineon Labs Phantom gaming console? Yeah....

The 2018 demonstration you're "remembering" was the laptop dGPU debut from CES 2020.

Am I the only one that is having trouble keeping track of these things ? ARC, Alchemist 3-series, 5-series, 7-series, DG1 Xe Iris, Ponte Vecchio, you'd never guess these things are related to each other in some way in a million years as an average consumer. Not to mention they used to have alternative names for each one of those types of GPU for some reason like Xe-LP, Xe-HPC, etc, what a nightmare.

I wonder if it's on purpose so that you're always confused about which of them gets delayed.

Kind of like RTX 30xx, Ampere, GA102, GA104, GA106 and their specific derivatives. Guessing this is just another troll post like the majority of the ones in this news piece as Intel seems to be the popular target these days.

Missing the targets they've officially announced is disappointing, but seemingly par for the course these days. It just seems like they should have been more cautious with the release window since they obviously didn't have everything ready to go yet. On the lower end laptop side it sounds a lot like software holding them back as they've got what will be retail units for testing (see the PC World test on the A370m).

 

[Vayra86]

It's the same anti-AMD idiots that doubted Zen after AMD spent half a decade selling steamy bulldozer piles.

Wrong

The 2018 demonstration you're "remembering" was the laptop dGPU debut from CES 2020.


Kind of like RTX 30xx, Ampere, GA102, GA104, GA106 and their specific derivatives. Guessing this is just another troll post like the majority of the ones in this news piece as Intel seems to be the popular target these days.

Missing the targets they've officially announced is disappointing, but seemingly par for the course these days. It just seems like they should have been more cautious with the release window since they obviously didn't have everything ready to go yet. On the lower end laptop side it sounds a lot like software holding them back as they've got what will be retail units for testing (see the PC World test on the A370m).

Time to market is everything. Even AMD saw how that worked in the GPU space not too long ago., losing the entire top half of the stack to Nvidia.

The starting point for Alchemist right now is far worse than what Polaris at the time was for AMD. They havent got mature drivers, they havent got a competitive (small) die for the performance on tap, they havent got volume in stock and the whole stack stops at midrange performance of 2020. Meanwhile AMD and Nvidia are going to relegate that midrange (~ 3060, the optimistic take on Intel's numbers) to entry level come the next release which is by now, likely to happen earlier. People think Intel might compete on price, but I think they wont even be able to.

 

[80251]

@Vayra86,
That's a disappointing take on Intel's entry into dGPU marketplace, but probably accurate. I just hope Intel can weather the storm of being relegated to entry level(?) or midrange(?) competition and become a real competitor to the existing duopoly.

 

[Vya Domus]

Kind of like RTX 30xx, Ampere, GA102, GA104, GA106

Kind of not like that because those are chip codenames not actual product names, so you guessed wrong.

 

[Vayra86]

@Vayra86,
That's a disappointing take on Intel's entry into dGPU marketplace, but probably accurate. I just hope Intel can weather the storm of being relegated to entry level(?) or midrange(?) competition and become a real competitor to the existing duopoly.

Yeah me too, but this needs long term commitment, and so far, Intel hasn't been committed for any longer than one-off products. It happened a few times before...

I mean would you EVER trust a company that releases a product every once in a while, and not in a somewhat fixed cadence? If there is no constant flow of refreshes and improvements, the whole support chain is a dead end too. That also applies to game developers, they won't trust and thus won't be willing to optimize for Intel, which is what Intel is experiencing right now.

Nvidia and AMD's drivers aren't just 'the drivers' but a huge box of tweaks applied over decades to virtually every game on the market, to keep their GPUs competitive in the marketplace, and quite often they got to that point by working with studios all over the place - actual engineers doing actual work in tandem with devs. Intel can at the very best apply some catch-all, blunt instruments to mimic that, but it will never EVER get to the level of refinement its competition has built up over time. The devil is in the details between camp green and red. Here comes Intel, with a sledgehammer-like object trying to wrestle control from two giants that turned sledgehammers into toothpicks a decade ago.

Even now, the postponing of product... Intel is still betting on some sliver of margins but that chance slips out of their fingers much faster by delaying product. No product is worse than 'a' product in the hands of gamers, because they'll never get started on building that commitment. Are we really smarter than Intel by saying and seeing this? Yes, and for the simple reason the company is stuck between rock and hard place, every choice they make right now is already going into the red numbers big time. If they're not committed to that for now and the next 2-4 generations of GPU, they might as well stop today. They do seem to have planned ahead though, but that planning is also moving further away from us by postponing the initial stuff, making it less realistic, even as a business case on its own.

 

[r9]

Larrabee


To be thinking intel being a savior for us is a big mistake, their prices will be right up there with nvidia and AMD.

Save me Intel so I have sinned.
AMD and Nvidia realized that there is no need for them to go into price wars.
Nvidia releases the new product first and set price then AMD release theirs and they price them exactly to relation to NVIDIA so if their card in class is 5% faster it will cost 5% more.
All I'm saying is that with 3 players that might not be so easy to do.

If AMD and NVIDIA learned any lesson from this is how not to get caught.

 

[thestryker6]

Kind of not like that because those are chip codenames not actual product names, so you guessed wrong.

Xe-LP, Xe-HPC, Xe-HPG, Alchemist and Ponte Vecchio are all codenames so you're just mostly wrong rather than trolling I apologize.

Time to market is everything. Even AMD saw how that worked in the GPU space not too long ago., losing the entire top half of the stack to Nvidia.

The starting point for Alchemist right now is far worse than what Polaris at the time was for AMD. They havent got mature drivers, they havent got a competitive (small) die for the performance on tap, they havent got volume in stock and the whole stack stops at midrange performance of 2020. Meanwhile AMD and Nvidia are going to relegate that midrange (~ 3060, the optimistic take on Intel's numbers) to entry level come the next release which is by now, likely to happen earlier. People think Intel might compete on price, but I think they wont even be able to.

I completely agree with you with regards to the retail discrete market. However you cannot forget the massive OEM access Intel has and they'll be able to move every part they can make whether it's positioned well or not. I'd like to dream that we're going to get a third player, but it certainly doesn't look like it's going to be this generation.

 

[Vya Domus]

Xe-LP, Xe-HPC, Xe-HPG, Alchemist and Ponte Vecchio are all codenames

It seems you mostly can't read, those are not chip codenames. Nvidia never refers to or advertises their products as "GA102" or whatever. So no, again, not "like that".

 

[Chrispy_]

The 2018 demonstration you're "remembering" was the laptop dGPU debut from CES 2020.

No I distinctly remember that one you're talking about too - that was where Steve Burke and someone else were taking bets on guessing the framerate because Intel didn't have FRAPS or any other counter running for whatever dumb reason. IIRC it was a little under 30fps and those heroic nerds decided to do a frametime analysis on the recorded footage to work that out!

 

[ModEl4]

So probably we will have the following?

Desktop ARC3:
June - China OEMs
July - China retail
August - Worldwide OEM
September - Worldwide retail

Desktop ARC5/ARC7:
ARC3 plus one month (at least)

So worldwide retail ARC5/ARC7 missing even the Q3 deadline? (and possibly mobile ARC5/ARC7 missing the back-to-school period? (worldwide))

At this rate, for desktop they may as well wait for Intel's/AMD's next gen products to launch first and at least announce competitive SRPs according to next gen, while at the same time using the mean time to polish the drivers/software.

 

[phanbuey]

Wrong

We'll see. I've been right so far on AMD @ $2.50, and apple M1 to desktop replacing x86...

I'm feeling the new intel - the prior attempts were under different, and poor, leadership that was complacent. The culture of the company is shifting back to engineering under Pat and they are gunning for the GPU segment with lessons from the failures of past products.

Is it going to be pretty, and on time, led by Raja, the first release out of the gate? Hell no... lol- it's going to be a mediocre product late to market, but it will come, and then the second, and then the third. Intel needs this not for just gamers but to compete in HPC space and there will be no stopping; battlemage will probably end up being what Alchemist was supposed to be, but I'm not seeing that any of these things will stop them. They're trending in the right direction.

The parallel stands: AMD was YEARS behind intel in the CPU space, and had no top stack to compete with nvidia, had horrible track record with vendors of delivering and with game developers for bad drivers and support utilities etc. They were leasing back their campus just to stay financially afloat. But they were engineering focused and were trending in the right direction.

The first versions of Zen were brushed off by many; just like Alchemist will be. And Intel will be in the same place in the GPU space: years behind with a relatively sucky product, the second one will suck slightly less, but once they apply disaggregated packaging and the designs from the cpu space we will have a very interesting product.

 

[Steevo]

I feel bad for them, they can’t seem to get anything quite right, between too high of power consumption on CPUs, flexing sockets, failure to launch and perform GPUs, getting countries to suck their balls for a foundry just doesn’t do it for them like it used to, their stuck process node, screwing AMD by buying up TSMC allocation for imagined products, bribing vendors to not use AMD…..

They sound like a junky who can’t get a fix and is about to start with the autoerotic asphyxiation…..

 

[TheoneandonlyMrK]

We'll see. I've been right so far on AMD @ $2.50, and apple M1 to desktop replacing x86...

I'm feeling the new intel - the prior attempts were under different, and poor, leadership that was complacent. The culture of the company is shifting back to engineering under Pat and they are gunning for the GPU segment with lessons from the failures of past products.

Is it going to be pretty, and on time, led by Raja, the first release out of the gate? Hell no... lol- it's going to be a mediocre product late to market, but it will come, and then the second, and then the third. Intel needs this not for just gamers but to compete in HPC space and there will be no stopping; battlemage will probably end up being what Alchemist was supposed to be, but I'm not seeing that any of these things will stop them. They're trending in the right direction.

The parallel stands: AMD was YEARS behind intel in the CPU space, and had no top stack to compete with nvidia, had horrible track record with vendors of delivering and with game developers for bad drivers and support utilities etc. They were leasing back their campus just to stay financially afloat. But they were engineering focused and were trending in the right direction.

The first versions of Zen were brushed off by many; just like Alchemist will be. And Intel will be in the same place in the GPU space: years behind with a relatively sucky product, the second one will suck slightly less, but once they apply disaggregated packaging and the designs from the cpu space we will have a very interesting product.

You sound like raja kadouri three years ago.

Difference= zen 1 came out the first year it was due not the third.

The parallel fails AMD released just another CPU as they already had and did (zen1)

Unless we count the 740 or larrabee Intel released zip didly squat.

 

[80251]

Yeah me too, but this needs long term commitment, and so far, Intel hasn't been committed for any longer than one-off products. It happened a few times before...

I mean would you EVER trust a company that releases a product every once in a while, and not in a somewhat fixed cadence? If there is no constant flow of refreshes and improvements, the whole support chain is a dead end too. That also applies to game developers, they won't trust and thus won't be willing to optimize for Intel, which is what Intel is experiencing right now.

Nvidia and AMD's drivers aren't just 'the drivers' but a huge box of tweaks applied over decades to virtually every game on the market, to keep their GPUs competitive in the marketplace, and quite often they got to that point by working with studios all over the place - actual engineers doing actual work in tandem with devs. Intel can at the very best apply some catch-all, blunt instruments to mimic that, but it will never EVER get to the level of refinement its competition has built up over time. The devil is in the details between camp green and red. Here comes Intel, with a sledgehammer-like object trying to wrestle control from two giants that turned sledgehammers into toothpicks a decade ago.

Even now, the postponing of product... Intel is still betting on some sliver of margins but that chance slips out of their fingers much faster by delaying product. No product is worse than 'a' product in the hands of gamers, because they'll never get started on building that commitment. Are we really smarter than Intel by saying and seeing this? Yes, and for the simple reason the company is stuck between rock and hard place, every choice they make right now is already going into the red numbers big time. If they're not committed to that for now and the next 2-4 generations of GPU, they might as well stop today. They do seem to have planned ahead though, but that planning is also moving further away from us by postponing the initial stuff, making it less realistic, even as a business case on its own.

Maybe Intel can cater to cryptominers instead of gamers? I've read the Radeon VII is still in demand by cryptominers.

 

[phanbuey]

You sound like raja kadouri three years ago.

Difference= zen 1 came out the first year it was due not the third.

The parallel fails AMD released just another CPU as they already had and did (zen1)

Unless we count the 740 or larrabee Intel released zip didly squat.

zip didly squat yes? They definitely don't make graphics, ur right...

 

[Vayra86]

I completely agree with you with regards to the retail discrete market. However you cannot forget the massive OEM access Intel has and they'll be able to move every part they can make whether it's positioned well or not. I'd like to dream that we're going to get a third player, but it certainly doesn't look like it's going to be this generation.

Oh they will move units, of that we can be certain, but discrete is what matters in terms of competition. Shitty laptops will always be shitty vendor lock ins, even with three players, and its pretty easy to make a bad laptop, whether they can even push Intel GPUs in there on top of the throttling monsters they already are is the question

zip didly squat yes? They definitely don't make graphics, ur right...

View attachment 247060

Gosh, so that is what ten years of selling all CPUs with an IGP looks like, who could've thunk it. The real question is, who games on an Intel IGP, and what do they game on it. I think the much safer take away from that pie chart is that about a quarter of the PC market is comprised of actual gaming PCs - and that's already being generous because lots of laptop dGPU is also just there because its there. Intel's IGP is primarily used for desktop view, some video, and lots of browser. I get your earlier post about them probably sticking to it, and I think earlier on I essentially said the same thing. They will fail the first time, and they'll need to keep going at it for a second, third and onward. But 'don't make graphics', no they definitely didn't do a whole lot of work in the gaming graphics space. No pioneering of any kind even though you would think with such a massive market share, they'd be the first to push that IGP for gaming

 

[Deleted member 24505]

zip didly squat yes? They definitely don't make graphics, ur right...

View attachment 247060

Bet's on reply's saying that does not count

 

[phanbuey]

Oh they will move units, of that we can be certain, but discrete is what matters in terms of competition. Shitty laptops will always be shitty vendor lock ins, even with three players, and its pretty easy to make a bad laptop, whether they can even push Intel GPUs in there on top of the throttling monsters they already are is the question


Gosh, so that is what ten years of selling all CPUs with an IGP looks like, who could've thunk it. The real question is, who games on an Intel IGP, and what do they game on it. I think the much safer take away from that pie chart is that about a quarter of the PC market is comprised of actual gaming PCs - and that's already being generous because lots of laptop dGPU is also just there because its there. Intel's IGP is primarily used for desktop view, some video, and lots of browser. I get your earlier post about them probably sticking to it, and I think earlier on I essentially said the same thing. They will fail the first time, and they'll need to keep going at it for a second, third and onward. But 'don't make graphics', no they definitely didn't do a whole lot of work in the gaming graphics space. No pioneering of any kind even though you would think with such a massive market share, they'd be the first to push that IGP for gaming

Point is - they do make a GPU, they do have market share, this isn't their first GPU and it certainly isn't 'diddly squat zip'. 10 years of selling IGPs and developing for DirectX/Ogl/Vulcan compliance is not nothing.

does UHD 770 suck absolute balls for gaming? yes! But it DOES run games.

 

[Vayra86]

Point is - they do make a GPU, they do have market share, this isn't their first GPU and it certainly isn't 'diddly squat zip'. 10 years of selling IGPs and developing for DirectX/Ogl/Vulcan compliance is not nothing.

does UHD 770 suck absolute balls for gaming? yes! But it DOES run games.

And yet they could not scale the way those games run (not too inconsistent at all, just slower) to their beautiful Xe even after three years of excessive trying. Its an inconsistent stuttery mess. So does their IGP run games, sure. But it doesn't speak for their discrete stuff or even the new architecture they have, which IS new and IS built to scale. This is the whole point. Making a weak GPU is easy, trailing the competition is easy, you have lots of ways to keep up (bigger dies, moving units a tier down, clocking, mature process, etc.), but chasing the cutting edge AND doing it efficiently is another story completely. If your arch is not scaling properly, you're fucked. We've seen this every single time when there were clear winners and losers each gen. Fury vs 980ti because Hawaii XT could not go further is a fantastic example. Radeon VII after Vega, fail upon fail, late to market, and AMD was literally stuck at that perf level for years on end. Even RDNA's first iteration could not save the day just yet.

 

[thestryker6]

Oh they will move units, of that we can be certain, but discrete is what matters in terms of competition. Shitty laptops will always be shitty vendor lock ins, even with three players, and its pretty easy to make a bad laptop, whether they can even push Intel GPUs in there on top of the throttling monsters they already are is the question

I meant desktop OEM, because yeah laptops are always going to be nothing but what's cheapest to put together at each price point. They ought to be able to control distribution more in the desktop OEM market than retail so the people looking to buy a certain performance range don't pay a huge markup. That alone may leave a good impression among the non-DIY market and then the performance/$ just has to be reasonable. Even if they're great performance/$ they aren't making enough of this generation to make any retail impact under best case scenario. It just seems like leaving a good impression of the actual product once people get their hands on it is the most important part and then ramping up production for the subsequent generations.

 

[Steevo]

Bet's on reply's saying that does not count

It doesn't count.

What do I win boss man? Will Intel suck my balls for buying a iGPU that fails to deliver any real performance beyond it does word real good?

I bet you get excited for the Kia car announcements and had a Kia on your wall as a kid instead of a supercar, plain looking people excite you, not supermodels, you like cream without the ice as it hurts your sensitive tummy, vanilla is for wild people that like to live on the edge?

Topic: Intel and their failure at delivering a Dgpu, typical Fanboi response, yeah but that Igpu thou.........

Also no offense man, but you are off topic here.

 

[TheoneandonlyMrK]

zip didly squat yes? They definitely don't make graphics, ur right...

View attachment 247060

Great example of the "D"GPU amounts out there, now how much of that Intel slice are ADD in Discrete GPUS instead of the kind of tat ENTHUSIASTS refuse to use.

Genius argument ,I laughed my arse off.

Reads thread title again, hmnn so no, Google translate isn't that shit surely.