Intel Stuck with 14nm Processors Till Holiday 2019

[btarunr]

Wrap your head around this: at some point in 2019, AMD will be selling 7 nm processors while Intel sells 14 nm processors. That how grim Intel's 10 nanometer silicon fabrication process development is looking. In the Q&A session of its Q2-2018 Earnings Call, Intel stated that the first products based on its 10 nm process will arrive only by Holiday 2019, making 14 nm micro-architectures hold the fort for not just the rest of 2018, but also most of 2019. In the client-segment, Intel is on the verge of launching its 9th generation Core "Whiskey Lake" processor family, its 5th micro-architecture on the 14 nm node after "Broadwell," "Skylake," "Kaby Lake," and "Coffee Lake."

It's likely that "Whiskey Lake" will take Intel into 2019 after the company establishes performance leadership over 12 nm AMD "Pinnacle Ridge" with a new round of core-count increases. Intel is also squeezing out competitiveness in its HEDT segment by launching new 20-core and 22-core LGA2066 processors; and a new platform with up to 28 cores and broader memory interface. AMD, meanwhile, hopes to have the first 7 nm EPYC processors out by late-2018. Client-segment products based on its architecture, however, will follow the roll-out of these enterprise parts. We could see a point in 2019 when AMD launches its 7 nm 3rd generation Ryzen processors in the absence of competing 10 nm Core processors from Intel. Posted below is an Intel slide from 2013, when the company was expecting 10 nm rollout by 2015. That's how much its plans have derailed.



View at TechPowerUp Main Site

 

[Caring1]

"10 nm process will arrive only by Holiday 2019"
Is that the 13th month for those of us not in the U.S.?

 

[Voluman]

Hm, thats interesting. Any chance to know whats the reason/cause of the delay? I mean in general why Intel cant deal with 10nm? Are they want to do some big? Or just simple in a terms of 'lame' ?

 

[agent_x007]

Intel wants to do true 10nm : LINK, problem is - it wasn't feasable on materials (or method) they wanted to use originally.
I think up to 2016/2017 they wanted to do 10nm on "tweaked" old technology (it's cost effective method, never failed so far) - it failed this time.
Now they are using EUV lithography (expensive AF, also tweaking takes A VERY long time because it's new technology - it was never used to mass produce high perf. CPU dies).
They made laptop chips so far, however I bet 4,5GHz+ clock is giving them a hard time
Why clock is important ?
Because launching a CPU with slower per thread performance than last gen is not something Intel wants to do. I think they ran out of ideas for great IPC gains, and AMD is catching up too fast for them to respond effieciently.

 

[randomUser]

"10 nm process will arrive only by Holiday 2019"
Is that the 13th month for those of us not in the U.S.?

Its probably chinese new year holiday. So not much left to wait. 6 months~
Well it may be easter holidays too.
But then might also be summer holidays.
Or it could be christma holidays.

I bet there are more holidays for the various religions and nations out there.

But you could be correct. It may be a 13th month we didn't know about.

 

[btarunr]

"10 nm process will arrive only by Holiday 2019"
Is that the 13th month for those of us not in the U.S.?

No, December 2019 (sixteen months from now).

 

[RejZoR]

I think Intel thought they are so ahead of AMD they didn't even plan anything smaller for realz. It's why Ryzen caught them totally off guard, EPYC is hammering them on all levels, nodes smaller and smaller ahead of even TSMC. AMD is finally getting in the loop with things.

 

[Xzibit]

Intel also can't ask for help fab wise. Their stock would plummet if they even think about it.

 

[Raevenlord]

Intel also can't ask for help fab wise. Their stock would plummet if they even think about it.

And there lies the crux of the issue, I believe.

 

[oxidized]

We just have to wait and see how they do, and also see what GloFo and Samsung do...

 

[Vayra86]

I think Intel thought they are so ahead of AMD they didn't even plan anything smaller for realz. It's why Ryzen caught them totally off guard, EPYC is hammering them on all levels, nodes smaller and smaller ahead of even TSMC. AMD is finally getting in the loop with things.

What. This is not even about AMD, its about process and fab development more than anything else. AMD is not even playing that game to begin with. And that 'choice' is starting to pay off for them now, while it did hurt them until recently.

Even so its not like Intel suddenly cannot compete when AMD puts out product on 7nm. Their 14nm isn't thát far off to make it obsolete. They are going to have to play the pricing game.

 

[W1zzard]

"10 nm process will arrive only by Holiday 2019"
Is that the 13th month for those of us not in the U.S.?

"Holiday season" is the time around xmas during which people buy presents to give away, or spend the money they got for xmas. I'd say basically the month of December

 

[ppn]

Here I was hoping for 1-3nm graphene by then.

 

[dorsetknob]

An unmentioned consideration might be the urgent need to Redesign the CPU to avoid those all-ready Discovered Back-doors
like Spectre and Meltdown ( AND THOSE THAT INTEL HAVE NOT DISCLOSED).
WE ALL KNOW that Intel have to rework upcoming CPU's to mitigate those flaws found in all Current designs

 

[Vya Domus]

Even so its not like Intel suddenly cannot compete when AMD puts out product on 7nm.

Actually this does represent an inability to properly compete against AMD , Intel is extremely reliant on their nodes and that sits at the core of their strategy. I mean that's just obvious when you look at the past decade and more specifically the last 2-3 generations in which they did nothing but iterate on 14nm. There is so much you can squeeze out of it.

 

[iO]

By this time they might consider scraping 10nm all together and focus on the development of the 7nm process.
Especially as AMD confirmed that Zen2 will be fabbed at TSMC..

 

[Minus Infinity]

What a joke. I said in 2013 the move to 10nm by 2015 was complete and utter BS and utterly unattainable. And now it'll basically 2020 before we see them finally make the move. Makes me smile.

 

[MT66]

I read that Intel was trying to be aggressive with their optical shrink to 10nm and had to dial it back alittle but now I wonder if their 14nm is so improved with clock speeds over the years that they have been just creating more work to try and get most of that clock speed improvement to carry over to 10nm along with the optical shrink. Maybe improving 14nm so much was counter intuitive.

 

[Vayra86]

Actually this does represent an inability to properly compete against AMD , Intel is extremely reliant on their nodes and that sits at the core of their strategy. I mean that's just obvious when you look at the past decade and more specifically the last 2-3 generations in witch they did nothing but iterate on 14nm. There 8s so much you can squeez out of it.

Of course, in the long term that is true. But not for the 12-17 months they have until 10nm 'is supposed' to launch. They can get by, if they compete on price. Such irony

 

[Caqde]

What. This is not even about AMD, its about process and fab development more than anything else. AMD is not even playing that game to begin with. And that 'choice' is starting to pay off for them now, while it did hurt them until recently.

Even so its not like Intel suddenly cannot compete when AMD puts out product on 7nm. Their 14nm isn't thát far off to make it obsolete. They are going to have to play the pricing game.

The problem is it isn't just the Fabrication process that is going to hurt Intel. Intel had to hold off on releasing new Architectures that they otherwise would have released on the 10nm process. Not only that they likely had to divert resources to make new 14nm designs (Coffeelake the 6 core 14nm chip along with this new 14nm 8 core chip, and their monolithic versions). I'm sure they would have rather been competing against AMD's Zen with Cannonlake and Icelake instead of Kabylake and Coffeelake.

Now with this delay Intel is going to have problems where it will hurt the most the Server and HEDT market. I'm sorry I can't see a 14nm chip competing against AMD's 7nm Zen2 and Zen3 because be honest the late 2019 release of 10nm chips along with the statement that these will be desktop and client only means that there is no 10nm hedt or server chips in 2019. So those will likely be 2020 but when in 2020. AMD will have Zen 3 based 7nm+ chips out at that time which seems like Intel will be competing with their 14nm chips. This is a joke and if investors are seeing this then that is probably why they are jumping ship and rightfully so. Intel needs a chiplet design if they are going to compete again and that likely won't happen until 2022 at the earliest if they started shortly after Jim Keller got hired. They will be hurting for the foreseeable future.

Another thing about this situation though is that due to AMD's chiplet design their cost's and release timeframes are considerably condensed compared to Intel. AMD only needs two chips designs one an APU the other a pure CPU. This allows AMD to reduce costs due to reduced defect rates and make it easier to obtain higher clocked Server/HEDT chips early on in newer processes while Intel has to wait for process maturity to reach a high point for the release of their Monolithic dies.

 

[Vayra86]

The problem is it isn't just the Fabrication process that is going to hurt Intel. Intel had to hold off on releasing new Architectures that they otherwise would have released on the 10nm process. Not only that they likely had to divert resources to make new 14nm designs (Coffeelake the 6 core 14nm chip along with this new 14nm 8 core chip, and their monolithic versions). I'm sure they would have rather been competing against AMD's Zen with Cannonlake and Icelake instead of Kabylake and Coffeelake.

Now with this delay Intel is going to have problems where it will hurt the most the Server and HEDT market. I'm sorry I can't see a 14nm chip competing against AMD's 7nm Zen2 and Zen3 because be honest the late 2019 release of 10nm chips along with the statement that these will be desktop and client only means that there is no 10nm hedt or server chips in 2019. So those will likely be 2020 but when in 2020. AMD will have Zen 3 based 7nm+ chips out at that time which seems like Intel will be competing with their 14nm chips. This is a joke and if investors are seeing this then that is probably why they are jumping ship and rightfully so. Intel needs a chiplet design if they are going to compete again and that likely won't happen until 2022 at the earliest if they started shortly after Jim Keller got hired. They will be hurting for the foreseeable future.

Another thing about this situation though is that due to AMD's chiplet design their cost's and release timeframes are considerably condensed compared to Intel. AMD only needs two chips designs one an APU the other a pure CPU. This allows AMD to reduce costs due to reduced defect rates and make it easier to obtain higher clocked Server/HEDT chips early on in newer processes while Intel has to wait for process maturity to reach a high point for the release of their Monolithic dies.

You may be right. But we may also see Intel pulling out all the stops and making radical changes still. I mean, if your forecast would be 100% realistic then I'm quite sure they are going to adjust, radically. There is just too much at stake.

 

[Caqde]

You may be right. But we may also see Intel pulling out all the stops and making radical changes still. I mean, if your forecast would be 100% realistic then I'm quite sure they are going to adjust, radically. There is just too much at stake.

Well if their latest leaked roadmap is true then they certainly don't seem to have anything planned immediately and honestly it takes years to make a chip design so certainly they can tweak a few things here and their and maybe expand chip designs to create chips like the 8 core i9 that they are releasing, but what else can they really do on 14nm physics only goes so far and making a higher performing chip will either require more clockspeed (power/heat), or more cores (size which means higher cost/defects and more power and heat). They can only go so far and the first since they have already pretty much maxed out that area and the second is a deadend road as AMD's chiplet design cost so much less and at 7nm with 48 to 64 cores for 4 chips would mean Intel will need a 40 - 56 core chip to even somewhat compete. We are talking a Monolithic Coffeelake chip that is twice the area of a 698mm2 Skylake XCC based chip. Now I could see a 36-40 core Intel chip based on two HCC chips this might save some face, but Intel better hope those AMD Zen 2 CCX's are only 6 core and not the rumored 8 cores. So that their likely 200W 40 core Dual chip Xeon is competing against a 48 Core EPYC and not a Theoretically 64 core monster.

Either way we will see in 2020, but if my feelings are correct next year will be the start for year's of hurt for Intel, I just hope this time that AMD comes out of it better than they did when they were selling their Opteron's. And right now it seems they likely will as the OEM's don't seem to be acting like they have been payed off by Intel this time around.

Intel's leaked server roadmap -> https://www.anandtech.com/show/1311...ap-leaks-cooper-lakesp-ice-lakesp-due-in-2020

 

[ShockG]

The obsession with nodes sizes when it's but a single number is quite alarming. Well not really given how poor media has become technically.
Okay so when we say 10nm or 7nm what are we talking about?
Are we talking minimum feature size only and if so, why only that?
Here is a simple comparison for us all.

Intel 14nm vs Samsung 14nm (all in nm where lower is generally better)
(both Fin-Fet; Intel was first to market with these chips long before any other company was considering high density logic silicon on fin-fet)

Fin Pitch: 42nm vs 49nm
Fin Height: 42nm vs 39nm
Gate Pitch: 70nm vs 78nm
MMP: 52nm vs 67nm

So once again what are we talking about here as just saying 14,12,10,7nm says nothing.
Global Foundries 12nm processes is inferior to the above 14nm processes by all accounts save for the minimum feature size which is the magical 12nm which we obsess over.
Unless anyone here knows the ins and outs of Global Foundries 7nm* node then you can't compare and say who is ahead of what.
More over the obsession with the node as if it means anything directly is laughable as it clearly illustrates that a technical discussion is being had by people who aren't technical in the least and it ends up being conjecture based on little to zero knowledge. It's surprising that those that know better even in this very thread, do not say anything to correct this.

Look at the TDP for the 8-core Core i7 9900K. It's still 95W with two more cores and a higher clock speed and of course more cache (as compared to the 8700K). So if Intel can hit these performance targets on the same process, same TDP, why exactly is it so pressing for us end users to have a new process? The performance gains due to a smaller node are not always the same and many times are negligible depending on what else changes with the process. Straight optical shrinks rarely do anything at all for performance. Intel isn't Stuck* it's called engineering. Overcoming the difficulty is the engineering, which they pour billions of dollars and man hours into.
The expectationally shallow appreciation and conversation around deep and detailed engineering/physics phenomenon is quite perplexing.
It's okay to dislike and even loath Intel based on their business practices, but if you're going to engage their products from a technical perspective, let's be sure we know what we are talking about or at the very least cognisant of how little we know before we type.

 

[Captain_Tom]

Which actually means "Paper Launch December 2019 at the earliest." These probably will be laptop and low-power only like Broadwell.


Good lord by the time Intel has a true high performance 10nm product, AMD will be on 7nm+! Hopefully Intel can compete again by 2021, or AMD will stat price gouging hard.

You may be right. But we may also see Intel pulling out all the stops and making radical changes still. I mean, if your forecast would be 100% realistic then I'm quite sure they are going to adjust, radically. There is just too much at stake.

There's nothing they CAN DO after they launch the 8/16 "i9" @ 5GHz. That's their last hurrah for 2 years.

New architectures take time to make, and they can't even really nail one down until they get 10nm working.

 

[DeathtoGnomes]

everyone assumes Holiday means Christmas time. It really is a vague because there are so many official holidays here in 'murica. Pick one.