Intel 10nm Ice Lake to Quantitatively Debut Within 2019

[btarunr]

Intel put out interesting details about its upcoming 10 nanometer "Ice Lake" CPU microarchitecture rollout in its recent quarterly financial results call. The company has started qualification of its 10 nm "Ice Lake" processors. This involves sending engineering samples to OEMs, system integrators and other relevant industry partners, and getting the chips approved for their future product designs. The first implementation of "Ice Lake" will not be a desktop processor, but rather a low-power mobile SoC designed for ultraportables, codenamed "Ice Lake-U." This SoC packs a 4-core/8-thread CPU based on the "Sunny Cove" core design, and Gen11 GT2 integrated graphics with 64 execution units and nearly 1 TFLOP/s compute power. This SoC will also support WiFi 6 and LPDDR4X memory.

Intel CEO Bob Swan also remarked that the company has doubled its 10 nm yield expectations. "On the [10 nm] process technology front, our teams executed well in Q1 and our velocity is increasing," he said, adding "We remain on track to have volume client systems on shelves for the holiday selling season. And over the past four months, the organization drove a nearly 2X improvement in the rate at which 10nm products move through our factories." Intel is prioritizing enterprise over desktop, as "Ice Lake-U" will be followed by "Ice Lake-SP" Xeon rollout in 2020. There was no mention of desktop implementations such as "Ice Lake-S." Intel is rumored to be preparing a stopgap microarchitecture for the desktop platform to compete with AMD "Matisse" Zen 2 AM4 processors, codenamed "Comet Lake." This is essentially a Skylake 10-core die fabbed on existing 14 nm++ node. AMD in its CES keynote announced an achievement of per-core performance parity with Intel, so it could be interesting to see how Intel hopes 10 "Skylake" cores match up to 12-16 "Zen 2" cores.



View at TechPowerUp Main Site

 

[chaosmassive]

Intel Investors = "BOB, DO SOMETHING !!!!"

 

[lexluthermiester]

While this is good progress, that graphic... "Evolving Culture"? "One Intel"? Really? Are they aiming those presentations at morons? Tripe. Intel needs to stop spouting gibberish, state facts & milestones and let them speak for themselves...

 

[stanleyipkiss]

Intel Investors = "BOB, DO SOMETHING !!!!"

I too play Overwatch.

 

[GoldenX]

"At last, the promise has been made".

 

[Octopuss]

At this point I don't even care about Intel anymore, because with SO friggin many architectures/designs/CPU names/whatever I don't even know what is what and I wouldn't be able to choose anything even if I wanted to.
Screw it, anything AMD will be a great upgrade from 3770K.

 

[kastriot]

They got lazy, too much money+no competition kills need for development and just calls for unwanted products made for milking customers.

 

[Vayra86]

Its coming in 2019, but we need a third stopgap architecture for desktop

In other words, its not coming in 2019. Its getting a paper launch + maybe some halo mobile product nobody will want. Well played At least the sun is shining in Intel's tiny little Cove with 2012-era core counts. Real roadmaps tell us another story: 2021 at best.

 

[SL2]

They got lazy

No, the transition to 10 nm hit a brick wall years ago, something they haven't been able to fix.

If they were just lazy they'd sell 10 nm 4 core CPU's for desktop -K series. Intel benefits from transitioning to 10 nm, they get more chips from every wafer.
Intel does not benefit from trying to transition to 10 nm for >4 years or whatever, that's just crazy expensive. If they were lazy they wouldn't even try going beyond 14 nm.

This usually never holds true:
"Brand X is lazy!" "Brand X is doomed!" "Game over, brand X!" "Product Y is a product Z killer!!!!!111!!!!!"

 

[notb]

No, the transition to 10 nm hit a brick wall years ago, something they haven't been able to fix.

If they were just lazy they'd sell 10 nm 4 core CPU's for desktop -K series. Intel benefits from transitioning to 10 nm, they get more chips from every wafer.

Less.
Yields are bad. They get less CPUs. They get more candidates for CPUs - many of which go to the bin.

In other words, its not coming in 2019. Its getting a paper launch + maybe some halo mobile product nobody will want.

4 core U CPUs are one of their best selling products, so I don't really get what you're trying to say.
This is why their focusing their 10nm capabilities on this segment.

 

[SL2]

Less.
Yields are bad. They get less CPUs. They get more candidates for CPUs - many of which go to the bin.

I meant in the long run, i.e the very reason why smaller process nodes are developed at all. The hindsight of Intels 10 nm nightmare doesn't really explain their motivations for developing it in the first place.

 

[notb]

I meant in the long run, i.e the very reason why smaller process nodes are developed at all. The hindsight of Intels 10 nm nightmare doesn't really explain their motivations for developing it in the first place.

In the long run we'll be communicating with quantum-computing cloud via mind-cotrolled pocketable terminals. Or something like that.

Intel has to think about future (short and long term), but also about current cashflow.

Their 10nm wasn't cost-effective for mainstream desktop chips and may still not be.
Since the final retail product costs $100-200, production cost would have be $30-50 at most.
Apple reportedly pays TSMC $72 for a single A12, which is less than half the size of a 9700K.

 

[londiste]

"Evolving Culture"? "One Intel"? Really? Are they aiming those presentations at morons?

Shareholders...

Their 10nm wasn't cost-effective for mainstream desktop chips and may still not be.

If they really intend to mass-produce at 10nm in the end of 2019 as they are saying, something has to be fixed. Speculations have been going around for a while (partially confirmed by Intel's own slide) that 10nm will not clock as high as current iteration of 14nm. Also, (much) worse yields and 10nm is very likely more expensive - roughly comparable TSMC 7nm was twice as expensive for the same area (for Zen2's small 80mm² dies) in December 2017 according to AMD. While all this may be worth it for the mobile due to efficiency, desktop still centers on performance and 10nm might not be cut out for it (hopefully, yet).

 

[SL2]

In the long run we'll be communicating with quantum-computing cloud via mind-cotrolled pocketable terminals. Or something like that.

Ok, you're dodging my point really hard. I'm done.

 

[PanicLake]

No, the transition to 10 nm hit a brick wall years ago, something they haven't been able to fix.

Hit a brick wall because they where sitting on their pedestal thinking "we got time..." while in reality they didn't!

Ok, you're dodging my point really hard. I'm done.

And I think you are missing his point. We are talking about the actual situation, Intel needs 10nm NOW, not in the long run.

 

[unikin]

So 14nm++++++++++++ until 2021? Nice one Intel. You leave us no choice but to buy Zen 2/Zen 3. That's what resting on the laurels brings to you. Well deserved. Time to short Intel and buy AMD stocks again I guess.

 

[SL2]

Hit a brick wall because they where sitting on their pedestal thinking "we got time..." while in reality they didn't!

Again, no. Intel doesn't delay a roadmap for years just to be lazy, it doesn't work like that.
It's possible that Intel underestimated AMD, I'm not saying anything about that. My point is, if it was laziness then they'd still be at 10 nm with a dumb shrunk Skylake.

Delaying the next big CPU since Nehalem or Sandy Bridge because of lack of competition: maybe possible.
Delaying 10 nm because of lack of competition: no way.

The two are tied together, but still separate.

And I think you are missing his point. We are talking about the actual situation, Intel needs 10nm NOW, not in the long run.

Nope, you'd better read before you reply. He strictly replied to ---> 10 nm YIELDS.

-I said Intel benefit from moving to 10 nm.
-He said no, the yield is too low, which is true.
-I said I meant when the yield is high it's better in the long run than 14 nm.
-He went offtopic.
-You explained WHEN 10 nm is needed, also offtopic, it's got nothing to do with what I wrote.

It's pretty obvious when 10 nm was needed.
The 8700K was a great CPU that worked well, 14 nm wasn't really a problem from a consumers point of view.
The 9900K on the other hand could have benefitted from 10 nm even though it's very fast as it is.

So yes, mass production of 10 nm was very much needed in 2018.

 

[iO]

If that leaked roadmap from yesterday is correct then there won't be any 10nm desktop chips until at least 2022.

 

[Vayra86]

Less.
Yields are bad. They get less CPUs. They get more candidates for CPUs - many of which go to the bin.


4 core U CPUs are one of their best selling products, so I don't really get what you're trying to say.
This is why their focusing their 10nm capabilities on this segment.

So they can put a volume product on a more expensive node? I think thats a double edged blade but its not the main reason they need 10nm...

You said it yourself: low yields. So this product is not a sign 10nm ia ready... its an attempt to make us believe Intel made good progress over the last six months. But we knew a part like this would be coming already. The lack of announcing everything else is the real message here.

Another real message is them readjusting fab lines towards continued 14nm. Thats a mid term investment.

If that leaked roadmap from yesterday is correct then there won't be any 10nm desktop chips until at least 2022.

Exactly, its a recent slide and similar to other ones weve seen lately, and they all tell the same story.

 

[R0H1T]

Less.
Yields are bad. They get less CPUs. They get more candidates for CPUs - many of which go to the bin.


4 core U CPUs are one of their best selling products, so I don't really get what you're trying to say.
This is why their focusing their 10nm capabilities on this segment.

Intel were selling 2 core ULV i3, i5 & i7 models as well just a couple of years back! The quad core U chips are in fact a rarity, still though with previous SKU being phased out they will dominate Intel sales eventually.

https://en.wikichip.org/wiki/intel/cores/kaby_lake_u

Again, no. Intel doesn't delay a roadmap for years just to be lazy, it doesn't work like that.
It's possible that Intel underestimated AMD, I'm not saying anything about that. My point is, if it was laziness then they'd still be at 10 nm with a dumb shrunk Skylake.

Delaying the next big CPU since Nehalem or Sandy Bridge because of lack of competition: maybe possible.
Delaying 10 nm because of lack of competition: no way.

The two are tied together, but still separate.

You could define lazy as being content with providing just quad cores for well over a decade or in fact overcharging for HEDT. They weren't as lazy as much as greedy IMO - quad cores since forever, dual core ULV also being the norm since at least half a decade! They could've released more cores & yet they didn't want to, of course because they didn't need to.

 

[Berfs1]

Y’all are uh, a little mislead. 10 cores for mainstream is coming out Q2’2020 (Xeon E 22xx in Q1’2020), while 10nm for mainstream is coming as early as 2022. Knowing intel, we can expect it to be at least 2 quarters behind the predicted schedule.

Intel were selling 2 core ULV i3, i5 & i7 models as well just a couple of years back! The quad core U chips are in fact a rarity, still though with previous SKU being phased out they will dominate Intel sales eventually.

https://en.wikichip.org/wiki/intel/cores/kaby_lake_u
You could define lazy as being content with providing just quad cores for well over a decade or in fact overcharging for HEDT. They weren't as lazy as much as greedy IMO - quad cores since forever, dual core ULV also being the norm since at least half a decade! They could've released more cores & yet they didn't want to, of course because they didn't need to.

At the time, more cores would have been more expensive. Ever wondered why 0-60in3second cars in the 1980s were really expensive? Because they didn’t have the tech lnology to make them at a relatively low cost. Yes, intel can add literally speaking as many cores as they want up to architectural limits (which is more than 10 btw), it’s yields that will be a legitimate issue.

 

[R0H1T]

And more cores aren't expensive now? Intel went from 65nm to 14nm with quad cores as mainstream limit & dual cores for ULV, then suddenly with Ryzen we had their core counts literally doubled in a year. Do you honestly believe they couldn't manage 6 cores on 32nm or 22nm, quad cores in laptops by then?

 

[notb]

You said it yourself: low yields. So this product is not a sign 10nm ia ready...

Why not? -U CPUs are a mainstream, important and popular product.
The slides say that Intel won't be able to go full 10nm in this segment.

Another real message is them readjusting fab lines towards continued 14nm. Thats a mid term investment.

For majority of their CPUs 14nm is perfectly fine. As long as 10nm is more less profitable, there's no reason to switch.

It's a different story with AMD. They're strategy is strongly based on 7nm, which lets them make high-core CPUs and implement a chiplet design. But even they "admit" TSMC 10nm is either expensive or limited by staying on 14nm with the I/O die.

If AMD didn't have access to 7nm, they would have to redesign the Zen CCX to fit more cores on the AM4 socket, which totally kills the whole cost-cutting idea.
Intel still has a lot of space in the LGA1151 package - especially if they drop GPU for high-core models (which is just disabled in existing -F parts).

Intel has a big lead in single-thread performance - something they can sacrifice on the way.
i5-8600 matches a heavily overclocked 2700X in single-thread, but stays within 65W TDP under full 6-core load.
In other words: they can make a 130W 12-core if they need one (HT will add another 20W).

Sure, AMD Zen2 12C/24T will use 120W instead of 150W at similar clocks. Some believe this will make people move from Intel to AMD. I think we'll simply see slightly larger coolers. ;-)

Exactly, its a recent slide and similar to other ones weve seen lately, and they all tell the same story.

It's a corporate offer roadmap. Mass market CPUs for OEMs.
In this market Intel desktop CPUS are competing with AMD APUs. 14nm is fine for 8 cores.

The quad core U chips are in fact a rarity

They're in most laptops that cost over $500...

 

[Vayra86]

Why not? -U CPUs are a mainstream, important and popular product.
The slides say that Intel won't be able to go full 10nm in this segment.


For majority of their CPUs 14nm is perfectly fine. As long as 10nm is more less profitable, there's no reason to switch.

It's a different story with AMD. They're strategy is strongly based on 7nm, which lets them make high-core CPUs and implement a chiplet design. But even they "admit" TSMC 10nm is either expensive or limited by staying on 14nm with the I/O die.

If AMD didn't have access to 7nm, they would have to redesign the Zen CCX to fit more cores on the AM4 socket, which totally kills the whole cost-cutting idea.
Intel still has a lot of space in the LGA1151 package - especially if they drop GPU for high-core models (which is just disabled in existing -F parts).

Intel has a big lead in single-thread performance - something they can sacrifice on the way.
i5-8600 matches a heavily overclocked 2700X in single-thread, but stays within 65W TDP under full 6-core load.
In other words: they can make a 130W 12-core if they need one (HT will add another 20W).

Sure, AMD Zen2 12C/24T will use 120W instead of 150W at similar clocks. Some believe this will make people move from Intel to AMD. I think we'll simply see slightly larger coolers. ;-)

It's a corporate offer roadmap. Mass market CPUs for OEMs.
In this market Intel desktop CPUS are competing with AMD APUs. 14nm is fine for 8 cores.


They're in most laptops that cost over $500...

Its also a low performance product, so it does not justify the jump to 10nm. You dont need to explain profitability to me.. that was not the point. The point was this CPU doesnt tell us much about the state of 10nm. And rest assured this CPU will be scarce and cost a pretty penny because of it. Not directly the greatest perks for a mainstream product...

 

[efikkan]

The continued problems for 10nm is a sad tale, since Intel is sitting on a new faster architecture they can't use (yet).

But in all the Zen(2) hype, it's easy to forget that despite all these troubles, Intel still have the fastest core performance, and is quite likely going to do well against Zen 2 for the mainstream platform. The only "hole" will be the lack of a direct competitor to AMD's expected 12-core CPU, but that shouldn't matter a lot. Coffee Lake (v2) will not be inferior when Zen 2 launches, the only sad part is that Intel will be stagnant and seemingly have nothing new in this segment this year.

I'm more curious in what lies ahead. We still don't really know much about Comet Lake-S, and as far as I can see there are three possibilities;
1) Just another iteration of Skylake with no real IPC changes, but minor tweaks in the layout to achieve better thermals and lower voltage, which is utilized for marginal clock gains. This is very much possible, but boring.
2) Some architectural tweaks (like Cascade Lake-SP) to achieve small IPC gains.
3) A "backport" of Ice Lake/Sunny Cove to 14nm.
(or a combination of these)
The only indicator I've seen so far is the support in Core Boot and Linux indicating it may be a relative of Skylake.