Intel "Rocket Lake-S" Desktop Processor Comes in Core Counts Up to 8, Gen12 iGPU Included

[btarunr]

Intel's 11th generation Core "Rocket Lake-S" desktop processor will come in core-counts only up to 8, even as its predecessor, "Comet Lake-S," goes up to 10. Platform descriptors for Intel's next four microarchitectures surfaced on the web, detailing maximum values of their "S" (mainsteam desktop), "H" (mainstream notebook), "U" (ultrabook), and "Y" (low power portable) flavors. Both "Comet Lake-S" and "Rocket Lake-S" are 14 nm chips. "Comet Lake-S" comes with core counts of up to 10, a TDP of up to 125 Watts, Gen 9LP iGPU with 48 execution units, and native support for up to 128 GB of DDR4-2667.

The "Rocket Lake-S" silicon is interesting. Rumored to be yet another derivative of "Skylake," it features up to 8 CPU cores, the same 125 W maximum TDP, but swanky Gen12 iGPU with 32 execution units. The memory controller is also upgraded, which supports DDR4-2933 natively. There is no "Ice Lake-H" or "Ice Lake-S" in sight (no mainstream notebook or mainstream desktop implementations), ditto "Tiger Lake." For the foreseeable future, Intel will only make quad-core designs of the two 10 nm microarchitectures. "Rocket Lake-S" is slated for 2021 when, hopefully, we'll see Intel escape the 14 nm black hole.



View at TechPowerUp Main Site

 

[Object55]

Literally started going backwards

 

[davideneco]

RKL may be sunny cove or willow cove based

 

[Crowley]

For the foreseeable future, Intel will only make quad-core designs of the two 10 nm microarchitectures. "Rocket Lake-S" is slated for 2021 when, hopefully, we'll see Intel escape the 14 nm black hole.

This is just sad, 2021 and we are "hoping" that we can get to 10nm on the S series of processors. SMH

 

[birdie]

Sky Lake v6! Or is it 7 or 8 already? I've lost count.

 

[davideneco]

This is just sad, 2021 and we are "hoping" that we can get to 10nm on the S series of processors. SMH

Wait

Nobody hope to get a 10nm with 4c at <4ghz in 2021

 

[Freaky_Snuke]

It feels as if Intel is trying to stress how long the world will still say "What is this company called that makes every cpu? Ah yes Intel!" (indicating Intels massive dominance in this regard to the average joe) when AMD is continuously stealing market and mind share compared to Intel's countless refreshes that are just embarrassing and sad at this point.

 

[yeeeeman]

It feels as if Intel is trying to stress how long the world will still say "What is this company called that makes every cpu? Ah yes Intel!" (indicating Intels massive dominance in this regard to the average joe) when AMD is continuously stealing market and mind share compared to Intel's countless refreshes that are just embarrassing and sad at this point.

Intel is doing its best to make things progress on the 10nm front as well as keep revenue's high. People can't really comprehend how big of a problem is to screw up things with a fabrication process. First of all, they had issues from 22nm. It was a sign of the fact that if they don't change something on their fab recipe, they will have bigger and bigger issues. On 14nm, they had even more issues and delays. They managed to find solutions, whatever the problem was. But going smaller to 10nm and adding lots of brave changes did send them to a total failure that we see. Now, imagine yourself having a recipe for a process. A flow that you use that becomes not enough, non scalable, not working. You need to understand why, you need to find a solution or in a the worst case, start from scratch. We don't know exactly what route they took, but I am suspecting it is the first, but they see their fixes are still not enough, hence this indefinite delay. Maybe with this failure people will understand how amazing is the fact that foundries execute so fast each and every year and how hard is to keep this going.
Intel doesn't have too many options. One of it is very bad, going fabless. Another is to keep working on finding a solution for their problem. Maybe buy some ideas from Samsung foundries.

 

[lemonadesoda]

Sky Lake v6! Or is it 7 or 8 already? I've lost count.

Intel's 11th generation Core "Rocket Lake-S" with Gen12 GPU, what a mess

 

[notb]

Literally started going backwards

Why exactly?
They can't make more fast cores on 14nm. They work with the technology they have. 8 core processors will remain mainstream for years.
AMD will dominate the high-end and Intel will keep making more money.

 

[efikkan]

All it takes is a random picture from the great Internet and we got the Intel bashing train rolling again…
This information is either incomplete or incorrect. Rest assured, Intel will not replace a lineup with something worse.

Also keep in mind that we don't know if Intel will retain their current product segmentation(OEM, retail, etc.), if they will release a platform consisting of both 10nm and 14nm parts, and when we'll start to see MCM CPUs on the desktop.

 

[R0H1T]

MCM CPUs are still some ways away, I predict 2021 & 7nm at the earliest.

 

[Crackong]

Comet Lake -> Rocket Lake ->>>> shooting star Lake ?

 

[ratirt]

Lets hope Intel will make this one better. 32U and 48U for graphics. That is not much though. AMD APUs will be faster.

 

[Vayra86]

This is just sad, 2021 and we are "hoping" that we can get to 10nm on the S series of processors. SMH

Not happening, I'll spoil it for you.

Intel will axe 10nm and jump to 7nm probably even at another fab as their own. They don't want to. But they will have to. Meanwhile, they're busy covering their 14nm woes, supply issues, completely uninteresting portfolio, and while they may sell, they won't be extracting fat margins. The clock is ticking, it is a matter of time before Intel makes a move that makes sense.

Quad cores on 10nm are a lost case to begin with, performance isn't really there as clocks are low, nice for low power but never going to be cost effective, Intel's 10nm profit will have to come off enterprise which is also what they said they'd try for 10nm first and foremost and where lower clocks don't hurt. "Just quads' also doesn't mix well with having a product stack where lesser chips can easily move down the stack.

But I'm very glad they managed to upgrade the GPU once again, and it still isn't capable of much beyond the standard fare

Comet Lake -> Rocket Lake ->>>> shooting star Lake ?

More like Smoking Crater lake

All it takes is a random picture from the great Internet and we got the Intel bashing train rolling again…
This information is either incomplete or incorrect. Rest assured, Intel will not replace a lineup with something worse.

Also keep in mind that we don't know if Intel will retain their current product segmentation(OEM, retail, etc.), if they will release a platform consisting of both 10nm and 14nm parts, and when we'll start to see MCM CPUs on the desktop.

Say what now... they've already replaced things with worse alternatives more than once. Forgot about Broadwell's Iris IGPs? Or the QLC 660p versus its predecessors? And if I start digging, I can surely expand the list.

I fail to see how they can replace it with something better when all we have is more of the same, on the same node, with minor IGP bumps. All I really read is another round of creative tweaking of specs and how those are explained to us, and 'benched' by Intel. Because effectively that is what everything post Coffee Lake has been: misdirection, bad information, confusion and vagueness, combined with bad user experiences of hot chips barely doing what they are specced for, while Intel is ready to respond to it with disclaimers "don't OC a K chip"...

Sry man, credit where it is due

 

[notb]

Intel will axe 10nm and jump to 7nm probably even at another fab as their own. They don't want to. But they will have to.

No, they won't.
They'll make CPUs using 3 different nodes - assigning them for best results.

Quad cores on 10nm are a lost case to begin with, performance isn't really there as clocks are low, nice for low power but never going to be cost effective

In the same sentence you've called it a lost case and nice for low power. So which one is it?
Why would they not be cost effective? A 4-core mobile SoC is way more expensive (and profitable) than a 4-core desktop one.
Check pricing on Intel ARK.

But I'm very glad they managed to upgrade the GPU once again, and it still isn't capable of much beyond the standard fare

What exactly is it not capable of?
Aside from gaming, which it's not made for. It also can't sing or bring you coffee. You expect too much.

 

[ratirt]

Maybe Intel will still stay with 14nm +++ node and not go 10 nor 7 The 10nm is still faulty and can't go high end desktop so who knows.

 

[kapone32]

Lets hope Intel will make this one better. 32U and 48U for graphics. That is not much though. AMD APUs will be faster.

Hardware unboxed did a video on the Mobile part and it was just as fast or faster than the 2500U so we will have to wait and see if that is indeed the case. We also don't know what Intel's plans are for the GPU(s) they are producing. It would not surprise me to see another (APU) released from Intel with updated Graphics. Unless AMD's next APU has RDNA instead of Vega.

 

[ratirt]

Hardware unboxed did a video on the Mobile part and it was just as fast or faster than the 2500U so we will have to wait and see if that is indeed the case. We also don't know what Intel's plans are for the GPU(s) they are producing. It would not surprise me to see another (APU) released from Intel with updated Graphics. Unless AMD's next APU has RDNA instead of Vega.

Yes it was but it had 64U packed. You think Intel is going to update the graphics? Since the 5 years time what Intel did is bump clocks 200 Mhz. Intel's last mobile got more graphics units and now they are cutting it again to 48. Besides it was faster in some scenarios but is others wasn't so it also depends.
With the GPU Intel is going for OEM I think. Servers? Workstations? I doubt it will be desktops for games maybe much further in the future but I really doubt it.

Watch this. It explains some stuff.

 

[notb]

Maybe Intel will still stay with 14nm +++ node and not go 10 nor 7 The 10nm is still faulty and can't go high end desktop so who knows.

Why exactly is 10nm faulty?
They started making mainstream large-volume SoCs. They work perfectly fine.

Of course they can't make as many 10nm chips as they'd want to, but they need a bit more time to get their 10nm fabs fully operational. But that's really the only issue left.

 

[efikkan]

Intel will axe 10nm and jump to 7nm probably even at another fab as their own. They don't want to. But they will have to. Meanwhile, they're busy covering their 14nm woes, supply issues, completely uninteresting portfolio, and while they may sell, they won't be extracting fat margins. The clock is ticking, it is a matter of time before Intel makes a move that makes sense.

They will not skip 10nm. 7nm will start low volume production in 2021 at the earliest, and will not reach a high volume until 2022 or later.
Intel is currently having more 14nm production lines than ever, with great yields on 14nm++, and even then can't meet the demand from either the consumer market or the enterprise market. This is poor planning from their part, but is in part caused by the delays of 10nm, since this have caused increased pressure on a single node.

We should reserve judgement about Intel's 10nm+ node until it reaches volume production next year. Intel recently claimed the "yields [are] ahead of expectations for client and data-center products". But even if 10nm+ yields are good enough, I'm still wondering if the total volume will still be a restriction, which may lead to having product lineups consisting of both 10nm and 14nm products, especially for OEMs, since the volumes required there are huge. And to make matters worse, Intel have decided to launch their new dedicated GPUs at the worst possible point in time…

I fail to see how they can replace it with something better when all we have is more of the same, on the same node, with minor IGP bumps. All I really read is another round of creative tweaking of specs and how those are explained to us, and 'benched' by Intel.

Well, they can't do much more on 14nm alone, not for the higher performance products anyway.
If they had planned better and at least had Sunny Cove on 14nm as well, it would have helped, but anything beyond 8 cores would still have been bottlenecked by the node.

Because effectively that is what everything post Coffee Lake has been: misdirection, bad information, confusion and vagueness, combined with bad user experiences of hot chips barely doing what they are specced for, while Intel is ready to respond to it with disclaimers "don't OC a K chip"...

Now you're not even serious any more.
Coffee Lake are excellent chips that have been doing better than expected, especially considering them being an "improvised" backup-plan.

 

[Vayra86]

Now you're not even serious any more.
Coffee Lake are excellent chips that have been doing better than expected, especially considering them being an "improvised" backup-plan.

Of course its part jest, but still, since Kaby Lake, K CPUs have been running very hot and OC headroom has shrunk, TDPs are being played with handily through board makers, etc etc. Excellent, yes, but also clearly at a limit. Beyond the first batch of CFL things got worse despite improvements under the IHS.

As for 10nm... who knows, a lot of things can happen in the meantime, but I'm not holding my breath. Its not like TSMC and others won't keep at improving their nodes either. Can and does Intel want to keep trailing that, or lead again?

 

[ratirt]

Why exactly is 10nm faulty?
They started making mainstream large-volume SoCs. They work perfectly fine.

Of course they can't make as many 10nm chips as they'd want to, but they need a bit more time to get their 10nm fabs fully operational. But that's really the only issue left.

I haven't seen 10nm Intel chip for desktop market. What we got is laptop and that is all Intel can do. Maybe faulty was a little bit of a stretch from my side. Let's just say not suitable for desktop. Allthough what Intel produced or tried in the desktop market with 10nm was faulty since it didn't make it to the market.
As far as I know and you probably remember earlier this year Intel's CPU supply problem. The problem is with 14nm and yet Intel must fulfill the contracts.

They will not skip 10nm. 7nm will start low volume production in 2021 at the earliest, and will not reach a high volume until 2022 or later.
Intel is currently having more 14nm production lines than ever, with great yields on 14nm++, and even then can't meet the demand from either the consumer market or the enterprise market. This is poor planning from their part, but is in part caused by the delays of 10nm, since this have caused increased pressure on a single node.

Are you sure Intel won't skip the 10nm for desktop? 7nm will start in 2021 and you have to admit we have 2020 already. Roadmaps from Intel show they are still going for 14nm (delay 10nm over an over) and this will happen in the 2020. The 10nm will only be for laptops as it is now and for desktops, if any, another worthless refresh on 14nm node.

 

[efikkan]

Are you sure Intel won't skip the 10nm for desktop?

Let me be precise so people don't misunderstand me;
I'm sure Intel is planning to ship desktop parts on 10nm. Delays can happen, plans can evolve, but as of right now, that is certainly their plan.
We know Intel recently said 10nm desktop was still on the roadmap, we know Intel has added driver support for 95W Tiger Lake desktop parts, and we know Ice Lake-X exists, so at least something should be coming, unless disaster strikes. And to be clear, this will be late in 2020, not early.

The big question is to what extent will Intel ship 10nm on the desktop. While Intel claims yields are good on 10nm+, production volume may still be an issue. Due to large commitments and strong market demand, Intel may prioritize server and mobile, at the expense of mainstream desktop. We may very well see a mix of 10nm and 14nm desktop parts, I just hope they prioritize well. HEDT should at least be fairly safe, as these use CPUs unsuited for Xeons.

7nm will start in 2021 and you have to admit we have 2020 already.

7nm parts will start to ship in the second half of 2021, which is almost two years away, and will not reach high volumes until 7nm+ in 2022, almost three years away. Intel also mentioned this May that 10nm++ is coming in 2021, so this will be the high volume node in 2021.

Roadmaps from Intel show they are still going for 14nm (delay 10nm over an over) and this will happen in the 2020. The 10nm will only be for laptops as it is now and for desktops, if any, another worthless refresh on 14nm node.

Which roadmaps are you talking about?

 

[notb]

I haven't seen 10nm Intel chip for desktop market.

Correct. Intel is not making any desktop CPUs using their 10nm.
I just fail to see why this makes a node faulty...

What we got is laptop and that is all Intel can do. Maybe faulty was a little bit of a stretch from my side. Let's just say not suitable for desktop.

But how do you know that?
Desktop CPUs is not a prioritized segment. They use the 10nm supply they have for CPUs that benefit the most - mobile.

The second segment that should get 10nm are servers, but Intel likely doesn't have the production capabilities to make the switch. There's no point in making 10% of Xeons on 7nm, which on paper will leave them unattractive compared to AMD.

Desktops will be the last to update, IMO likely already to 7nm.