Intel Core i5-L16G7 is the first "Lakefield" SKU Appearance, Possible Prelude to New Nomenclature?

[btarunr]

Intel Core i5-L16G7 is the first commercial SKU that implements Intel's "Lakefield" heterogenous x86 processor architecture. This 5-core chip features one high-performance "Sunny Cove" CPU core, and four smaller "Tremont" low-power cores, with an intelligent scheduler balancing workloads between the two core types. This is essentially similar to ARM big.LITTLE. The idea being that the device idles most of the time, when lower-powered CPU cores can hold the fort; performance cores kick in only when really needed, until which time they remain power-gated. Thai PC enthusiast TUM_APISAK discovered the first public appearance of the i5-L16G7 in an unreleased Samsung device that has the Userbenchmark device ID string "SAMSUNG_NP_767XCL."

Clock speeds of the processor are listed as "1.40 GHz base, with 1.75 GHz turbo," but it's possible that the two core types have different clock-speed bands, just like the cores on big.LITTLE SoCs. Other key components of "Lakefield" include an iGPU based on the Gen11 graphics architecture, and an LPDDR4X memory controller. "Lakefield" implements Foveros packaging, in which high-density component dies based on newer silicon fabrication nodes are integrated with silicon interposers based on older fabrication processes, which facilitate microscopic high-density wiring between the dies. In case of "Lakefield," the Foveros package features a 10 nm "compute field" die sitting atop a 22 nm "base field" interposer.



Intel's nomenclature for the Core i5-L16G7 is fascinating. It condenses Intel's lengthy "i9-10980XE" (7-character) model names down to 5, besides the main brand extension (i3, i5, i7, i9). The first character probably represents the product type, followed by a numerical model number, further followed by "G" denoting the presence of integrated graphics, and a numeral next to "G" denoting its tier. Here's hoping this nomenclature holds, because Intel now has three: the 10th gen "Comet Lake" mobile processors retain the classic nomenclature (eg: Core i7-10710U), while the 10th gen "Ice Lake" has a slightly improved nomenclature (eg: Core i7-1065G7), and now "Lakefield" brings in the shortest of the three (eg: Core i5-L16G7).

View at TechPowerUp Main Site

 

[bug]

F*ck nomenclatures, everybody should just use QR codes and be done with it

 

[notb]

Heterogenous x86 cores so quickly... unbelievable.
+ GPUs
+ OneAPI

Nice.

 

[GlacierNine]

Heterogenous x86 cores so quickly... unbelievable.
+ GPUs
+ OneAPI

Nice.

But what if they suck?

 

[R0H1T]

Heterogenous x86 cores so quickly... unbelievable.

Hardly, I talked about this what ~2 years back & then ARM had it for 7+ years IIRC?

Intel Could Develop its own big.LITTLE x86 Adaptation

big.LITTLE is an innovation by ARM, which seeks to minimize power-draw on mobile devices. It is a sort of heterogeneous multi-core CPU design, in which a few "big" high-performance CPU cores work alongside a few extremely low-power "little" CPU cores. The idea here is that the low-power cores...

www.techpowerup.com

 

[bug]

Hardly, I talked about this what ~2 years back & then ARM had it for 7+ years IIRC?

Intel Could Develop its own big.LITTLE x86 Adaptation

big.LITTLE is an innovation by ARM, which seeks to minimize power-draw on mobile devices. It is a sort of heterogeneous multi-core CPU design, in which a few "big" high-performance CPU cores work alongside a few extremely low-power "little" CPU cores. The idea here is that the low-power cores...

www.techpowerup.com

I think he meant it's unbelievable Intel went big.LITTLE so fast. Though I'm not sure we have an indication of when exactly Intel started working on it.
Then again, considering about 3 decades ago my computer was Z80 based, where we are today is pretty unbelievable in its entirety.

 

[R0H1T]

Tbh it looks like one of those humbling moments, likewise "the glue" which they're also trying to perfect. Big little & it's subsequent iterative revisions clearly are the key to greater efficiency in the ULP tablet & convertible space. It's one thing Intel would've learnt especially in the last half of the decade is that hubris takes you nowhere, we've seen empires befall on their own swords & Intel isn't even 0.1% of that.

 

[windwhirl]

F*ck nomenclatures, everybody should just use QR codes and be done with it

*random conversation about which CPU someone should upgrade to*

"Oh yeah, that one is your best bet, the Intel Core i7-*beatboxing gibberish*"

My head tried to think how to pronounce a QR code and was brought back to the Post Office scene in Men in Black 2.

 

[SRB151]

I can see how multiple chips on the same package can be useful, but I'm not sure I see the advantage of this setup other than power savings. Currently, multi-core cpus power gate the unused cores, but they're all the same and all high performance. While "little" cores use less power, if you truly need to ramp up for more demanding workloads, you wind up with a couple of high performance cores working along side some slower ones. Kind of going backwards from a performance perspective.

 

[R0H1T]

Kind of going backwards from a performance perspective.

It's not really, for the kind of workloads this is gonna be used for it's likely gonna be the best in class ~ in terms of a balanced design & approach. Although I do expect these chips to be more expensive (than say equivalent pentium or celeron chips) & the devices sporting them to be uber expensive, at least at first.

 

[GlacierNine]

I can see how multiple chips on the same package can be useful, but I'm not sure I see the advantage of this setup other than power savings. Currently, multi-core cpus power gate the unused cores, but they're all the same and all high performance. While "little" cores use less power, if you truly need to ramp up for more demanding workloads, you wind up with a couple of high performance cores working along side some slower ones. Kind of going backwards from a performance perspective.

Not necessarily.

An efficient low power core could still be used in say, a game, for a less demanding thread. It's greater efficiency would make it produce less heat than a high performance core doing the same job, and performance would be unaffected for that thread.

But because it's not generating as much heat, your package temp as a whole is lower, which would mean the high performance cores may be able to clock higher for longer, increasing overall performance.

Sure, for a workload like Blender or Cinebench it's going to be slower, since there are no threads in those workloads that are of lower priority than other threads, but for a mixed workload like a game it could have advantages, especially if the low power cores are physically placed between the high power ones, thus helping to reduce the thermal density of the CPU die overall.

 

[GeorgeMan]

They finally found a way to complicate the names even more!

 

[R0H1T]

You just wait till the full lineup is revealed

 

[Tomorrow]

Heterogenous x86 cores so quickly... unbelievable.
+ GPUs
+ OneAPI

Nice.

You're an Intel fanboy. That much is clear from prevoius news comments. Go buy Lakefield, go buy DG1, use OneAPI - oh that's right - you can't.
That's what Intel is these days. A powerpoint company who has cool tech behind closed doors (or so they say) but no actual compelling products in retail.

 

[notb]

You're an Intel fanboy. That much is clear from prevoius news comments. Go buy Lakefield, go buy DG1, use OneAPI - oh that's right - you can't.

I don't understand why it's fine to drool over leaks of AMD's Cinebench scores, but you find it so weird that someone looks forward to using some really interesting Intel stuff.
Care to explain?

no actual compelling products in retail.

Reality disagrees. You've missed the financial reporting, right?

 

[I No]

You're an Intel fanboy. That much is clear from prevoius news comments. Go buy Lakefield, go buy DG1, use OneAPI - oh that's right - you can't.
That's what Intel is these days. A powerpoint company who has cool tech behind closed doors (or so they say) but no actual compelling products in retail.

Rude and uncalled for tbh, WCCF type of comment, even though I agree with some points in the second part of it the first part just trashes the rest of the read.
On Topic: I'm really curious about the market this would cater to and cannot wonder if they're not to late to the party, time will tell though.

 

[Vayra86]

You're an Intel fanboy. That much is clear from prevoius news comments. Go buy Lakefield, go buy DG1, use OneAPI - oh that's right - you can't.
That's what Intel is these days. A powerpoint company who has cool tech behind closed doors (or so they say) but no actual compelling products in retail.

What is your contribution other than bashing like this?

I think this is new technology and its apparently one attempt for Intel to move CPU forward. I don't see what's wrong with that, and I'll take this sort of news a hundred times over the next Lake-on-fire. Its a bench of working, new tech...

 

[Tomorrow]

I don't understand why it's fine to drool over leaks of AMD's Cinebench scores, but you find it so weird that someone looks forward to using some really interesting Intel stuff.
Care to explain?

Atleast those are scores for real, existing products in retail. It literally does not matter how intresting Intel's stuff is if they can't produce it and it's not in retail. Like i said - a powerpoint company.

Reality disagrees. You've missed the financial reporting, right?

...revenue based on aging 14nm tech and server sales. Nothing too exciting. What's you point with this sentance?

I think this is new technology and its apparently one attempt for Intel to move CPU forward. I don't see what's wrong with that, and I'll take this sort of news a hundred times over the next Lake-on-fire. Its a bench of working, new tech...

10nm desktop and server = not delivered. Only some mobile CPU's that top out at 4c/8t
7nm = not delivered until 2022 at the earliest.
Foveros = not delivered
Lakefield = not delivered
EMIB = not delivered. Only a single product from years ago.
Optane 2nd gen = not delivered
PCI-E 4.0 = not delivered
GPU's = not delivered. Only a single low end development vehicle.
Willow Cove = not delivered
Golden Cove = not delivered
Sapphire Rapids = not delivered

Intel. A company that overpromises and underdelivers.

 

[biffzinker]

Lakefield = not delivered

It's in the new/unreleased Microsoft Surface Neo.

A variation of Samsung's Galaxy Book S with Lakefield is set to release this year.

 

[GlacierNine]

It's in the new/unreleased Microsoft Surface Neo.

That doesn't sound very delivered

 

[notb]

On Topic: I'm really curious about the market this would cater to and cannot wonder if they're not to late to the party, time will tell though.

You mean heterogeneous cores?
They really improved battery life in smartphones. No reason why something similar wouldn't happen in laptops.
Late to the party - clearly not - there's no competition on x86. And it'll make incoming ARM ultrabooks less attractive as well.

In fact, I'd love to have control over this. Like a big, mechanical switch on the side of my laptop for limiting the CPU just to the little cores (we used to get that for WiFi).
Multiple random events make the CPU spike from time to time - eating battery and generating more cooling noise.
But when I'm just reading a text or doing basic browsing, I'm fine with a 2-core Atom limited to 3W.
Also, since such a 3W chip would be fine on laptop's passive cooling capabilities (with big margin), one could safely keep the notebook running in a bag or even a sleeve.

Possible use cases are quite simple to imagine. But let's wait and see what kind of product we actually get.

 

[SRB151]

Not necessarily.

An efficient low power core could still be used in say, a game, for a less demanding thread. It's greater efficiency would make it produce less heat than a high performance core doing the same job, and performance would be unaffected for that thread.

But because it's not generating as much heat, your package temp as a whole is lower, which would mean the high performance cores may be able to clock higher for longer, increasing overall performance.

Sure, for a workload like Blender or Cinebench it's going to be slower, since there are no threads in those workloads that are of lower priority than other threads, but for a mixed workload like a game it could have advantages, especially if the low power cores are physically placed between the high power ones, thus helping to reduce the thermal density of the CPU die overall.

Yeah, with some current games that run mostly on 1-2 threads, but games are increasingly taking advantage of more cores, and, as we've seen with Ryzen, windows scheduler has enough issues with keeping cores and threads straight. I can't imagine it trying to keep straight which threads have to run fast vs those that don't. Besides, if a game, or any software for that matter, can benefit from more, faster cores, why would they bother to sort out which one goes where?

 

[notb]

Yeah, with some current games that run mostly on 1-2 threads, but games are increasingly taking advantage of more cores

That's a misunderstanding.
A game, or any complex software, runs many threads (even tens or hundreds). They will be run on all available cores.
The problem is: some threads are heavy and some are extremely light.
If a program runs 2 compute threads and 10 tiny "utility" threads, it'll use 2 cores and just a marginal part of the rest.

Besides, if a game, or any software for that matter, can benefit from more, faster cores, why would they bother to sort out which one goes where?

Because of context switching.
So in the Blender example mentioned earlier: heavy rendering would be run on the "big" cores, while light background stuff (OS, networking etc) could be kept on the small ones. That means less switching and less performance lost.

And on the other hand: that would help PC remain responsive (usable) even under heavy load (also improving stability!). If your mouse/keyboard, your music player and browser are run by separate cores, you can still do something useful with a PC that is otherwise totally clogged by a computing task.

 

[GlacierNine]

snip

Right but back to my earlier question - there's no hardware and no benchmarks. How are you so confident in saying intel are not developing a product that's going to suck?

 

[Vayra86]

Intel. A company that overpromises and underdelivers.

Thanks but you can state the obvious without getting personal, too. We all know this, don't we?

Regardless of all they don't deliver, I think its good to keep an eye out for the developments that DO matter, and BIG.Little on x86... I say bring it on! We've already seen its merit.