Intel Core "Meteor Lake" Confirmed NOT Coming to Desktops, Only AIOs and Mini PCs

[btarunr]

Intel's next generation Core "Meteor Lake" processor was confirmed by ComputerBase.de to not release on the desktop platform. The processor will not make it to a socketed desktop package such as the upcoming LGA1851. It will see a mobile-only (notebook and tablets only) launch, with select processor models based on the mobile BGA package being made available to PC OEMs to build all-in-one desktops and mini PCs as non-socketed processors.

The desktop platform presence of "Meteor Lake" has been surrounded by some controversy, owing mainly to its maximum CPU core count of 6P+16E, which is bound to fall short of the current 8P+16E, and AMD's 16P. A slide released by Intel added to the confusion, which indicated that "Meteor Lake" has a thermal range of 5 W to 125 W, with the latter being typically associated with the processor base power values of desktop Core K-series processors.



Last week, in a statement to PC World, Intel Client Computing Group head Michelle Johnston Holthaus, implied that "Meteor Lake" will address all client form-factors, with desktop processor models arriving in 2024, a surprising disclosure, given that the company is preparing to launch its 14th Gen "Raptor Lake Refresh" processors within Q4-2023. "I want one processor family top to bottom for both segments, doesn't everybody?"

ComputerBase.de sought more explicit clarifications from Intel on whether by "desktop," Johnston Holthaus meant mainstream desktop (i.e. socketed processors). After some back and forth, Intel clarified that by "desktop," they meant prebuilt all-in-one desktops, which tend to be internally similar to notebooks and use low-TDP mobile processors; and mini PCs, such as from the NUC brand that ASUS recently acquired from Intel.

This wouldn't be the first time an Intel microarchitecture completely skipped mainstream desktop. The company's "Ice Lake" and "Tiger Lake" microarchitectures were limited to notebooks, convertibles, tablets; and a specific few SKUs made it to AIO desktops and mini PCs.

As for the mysterious 125 W reference in that slide, it could just refer to the maximum turbo power value of a certain higher-spec H- or HX-segment SKU. H- and HX-segment SKUs have been known to come with maximum turbo power values above the 100 W mark for some time now.

View at TechPowerUp Main Site | Source

 

[Vayra86]

Oh no, look at this dry crater of disappointment, we're missing out on a Lake!

 

[theouto]

I doubt this socket will see another major launch that is not just a refresh, so eh, nothing lost really.

 

[Hyderz]

guess this meteor will have no impact on desktop enthusiasts

 

[Vayra86]

guess this meteor will have no impact on desktop enthusiasts

We need a badum-tss smiley.

 

[Assimilator]

The only news about this was that people were stupid enough to take a rhetorical question and interpret it as some sort of product announcement. There was never any possibility of MTL coming to desktop because it's a test for a node shrink, just like CNL, ICL and TGL were.

Yes, Intel gets to make people pay for its products AND then you get to beta test those products too. The company truly is an amazing triumph of capitalism.

 

[Space Lynx]

arrow lake is really only one i prob will read reviews on, arrow lake vs 8800x3d will be a fun battle to watch. i think amd will win it, even if arrow lake is increasing its 3d cache somewhat, i just think 8800x3d is going to be AMD's crown jewel, but we will see heh

 

[bug]

guess this meteor will have no impact on desktop enthusiasts

If I had to guess, I'd say this is just Intel catering to system builders that want to tout a new CPU every year.
But I'm good, my 12600k is not slow by any means, I can keep it for a few more years without missing out on anything.

 

[Dirt Chip]

Pulling the trigger on that 13900k of mine seems like the better choice by the day.
Big improvement over 12900k (almost choose this one..), just about as any upcoming 14900k and no real performance uplift from Intel in the near future.

Shame it underclock like cr*!& and do bad with my memory config, but you can`t have it all I guess...

 

[Wirko]

Yes, Intel gets to make people pay for its products AND then you get to beta test those products too.

... AND, even before that, Intel gives us the opportunity to salivate over their beta roadmaps.

Oh no, look at this dry crater of disappointment, we're missing out on a Lake!

My conclusion after 45 seconds of long and deep thinking: when you put a notebook on top of your desk, it's a desktop PC!

 

[dj-electric]

The only news about this was that people were stupid enough to take a rhetorical question and interpret it as some sort of product announcement. There was never any possibility of MTL coming to desktop because it's a test for a node shrink, just like CNL, ICL and TGL were.

Yes, Intel gets to make people pay for its products AND then you get to beta test those products too. The company truly is an amazing triumph of capitalism.

idk why people call it a test, like a small group of 10 engineers decided to try something whacky. Like this wasn't a company wide, 5 year long effort to bring into market.
This is about Intel's node shrink, but no less than it is about connecting several silicon parts, with large representation from TSMC too. This "test" CPU is going to be Intel's de-facto next gen mobile CPUs. This isn't some one off, but something they are going to iterate in years to come.

Regarding desktop performance, maybe its a good reminder to not have too high of expectations since showing any kind of significance in gaming performance over 12th gen is probably going to take a few healthy generations. From both sides.

 

[AusWolf]

So what's with LGA-1851, then?

As far as I recall, Ice Lake and Tiger Lake were meant to be mobile-only architectures right from the drawing board, but Meteor Lake must scale pretty badly if Intel turned its back to the desktop variant.

 

[bug]

So what's with LGA-1851, then?

As far as I recall, Ice Lake and Tiger Lake were meant to be mobile-only architectures right from the drawing board, but Meteor Lake must scale pretty badly if Intel turned its back to the desktop variant.

You don't know why they stopped at 6P cores. It could be production issues, not scaling per se. It could be fab capacity (i.e. if you put more P cores in a CPU, you get less CPU per wafer, so you need more wafers).

 

[AusWolf]

You don't know why they stopped at 6P cores. It could be production issues, not scaling per se. It could be fab capacity (i.e. if you put more P cores in a CPU, you get less CPU per wafer, so you need more wafers).

True, but they cancelled the desktop chips, nonetheless, on contrary to Ice/Tiger Lake which were never meant to come to desktop in the first place (as far as I know). It means to me that the architecture isn't suited for desktop, be it for performance, scaling, or economic reasons.

 

[bug]

True, but they cancelled the desktop chips, nonetheless, on contrary to Ice/Tiger Lake which were never meant to come to desktop in the first place.

Well, they saw they couldn't deliver a healthy performance increase, and they said "screw that". What were they supposed to do? I mean, look at all the flak they're catching for RPL refresh (and that's not even out yet).

As an engineer, I can see how this is a delicate dance: you want a single architecture that can scale across the entire stack, but at the same time you have to make sure you don't put more effort into that than the effort required to build separate designs. Because you're building something that hasn't been built before, you'll get that wrong more than you'll get it right.

Iirc, Ice Lake was meant for the desktop initially, it wasn't until the second node slip-up or so that they had to revise that. But that's really just academic at this point.

 

[AusWolf]

Well, they saw they couldn't deliver a healthy performance increase, and they said "screw that". What were they supposed to do? I mean, look at all the flak they're catching for RPL refresh (and that's not even out yet).

As an engineer, I can see how this is a delicate dance: you want a single architecture that can scale across the entire stack, but at the same time you have to make sure you don't put more effort into that than the effort required to build separate designs. Because you're building something that hasn't been built before, you'll get that wrong more than you'll get it right.

Iirc, Ice Lake was meant for the desktop initially, it wasn't until the second node slip-up or so that they had to revise that. But that's really just academic at this point.

Fair enough. As a home user, I'd much rather see a significant leap less often than a small step every year anyway.

 

[Wirko]

It means to me that the architecture isn't suited for desktop, be it for performance, scaling, or economic reasons.

Or reasons of reputation. Everyone behold, here are the most advanced processors we can sell to you: the i3 and the i5 the 3 Ultra and the 5 Ultra!

 

[Chrispy_]

 

[_Flare]

And as it looks to me Lunar Lake will be the same no-LGA topic.

 

[Wirko]

And these two have some good company just next to the right:

 

[Daven]

Fair enough. As a home user, I'd much rather see a significant leap less often than a small step every year anyway.

I thought it was pretty well reported that Meteor Lake desktop was meant to be released this year with the full 8x16x2 design but it was delayed due to production issues. So they renamed the full chip to Arrow Lake for release next year and are releasing Raptor Lake refresh this year to fill the desktop space void until then.

View attachment 315355

Lololololol! I guess the two states aren’t mutually exclusive. Quantum mechanics is right!

 

[Assimilator]

idk why people call it a test, like a small group of 10 engineers decided to try something whacky. Like this wasn't a company wide, 5 year long effort to bring into market.
This is about Intel's node shrink, but no less than it is about connecting several silicon parts, with large representation from TSMC too. This "test" CPU is going to be Intel's de-facto next gen mobile CPUs. This isn't some one off, but something they are going to iterate in years to come.

It's a processor range built using a new process/technology/whatever that has shown performance deficiencies that would be unacceptable in desktop applications, but that Intel has decided to launch anyway. Essentially a prototype pushed into production because their investors demand a new product range every year. You can call that what you like, but there's no getting around the fact that these are beta-quality products being sold as final ones, and in my books that makes its buyers beta testers

And you can bet that if MTL fails horribly, Intel will quietly sweep it under the rug, pretend the failure never happened, and take a different direction - exactly what happened with CNL. Now if MTL was as bad as CNL the marketing hype for the former would never have been allowed to begin, but it doesn't change the fact that Intel has sold and continues to be willing to sell processors that it knows are not up to snuff, and ultimately consumers are the ones who get saddled with its "production quality" lemons.

As I said, from the viewpoint of end-stage capitalism, it's an amazing feat. But from the viewpoint of honesty and common decency, it's an absolutely vile practice, and the reason I will probably never buy Intel CPUs again in my lifetime.

 

[Squared]

lol

So what's with LGA-1851, then?

As far as I recall, Ice Lake and Tiger Lake were meant to be mobile-only architectures right from the drawing board, but Meteor Lake must scale pretty badly if Intel turned its back to the desktop variant.

Rocket Lake was either Ice Lake or Tiger Lake remade on the 14nm+++++ node, which I think shows that Intel wanted those architectures for desktop, but either not enough chips could be produced or the early 10nm chips couldn't reach high enough clock speeds. I would guess the latter for Ice Lake and the former for Tiger Lake. I think Tiger Lake was competing for foundry space while server chips were in really high demand and Ice Lake for servers was about to be released.

My guess is Intel 4 has poor yields or is just very late, otherwise it would be released before December. In either case it'd be hard to bring it to market on desktop very long before Arrow Lake which will be a much bigger upgrade and is claimed by Intel to be on track. Moreover the first chips to come out of a new node tend to have low clock speeds. Arrow Lake might avoid these concerns by being available on a good schedule and by including a new microarchitecture which should be a bit faster even without a clock speed boost.

Lastly, there is good news probably for Meteor Lake: some ultraportable laptops may have removeable LPDDR memory: https://www.techpowerup.com/314093/...rst-lpcamm-ushers-in-future-of-memory-modules

 

[bug]

Fair enough. As a home user, I'd much rather see a significant leap less often than a small step every year anyway.

You and me both.
But there are those that want a new shiny every year. System builders strongly push for that. And since Intel is ultimately in the business of selling things, they have to cater to everyone.
Personally, I have no problem with frequent, incremental releases. I just ignore them till there's an actual significant performance gap between what I have and what I can buy.

 

[HD64G]

Intel Confirms Meteor Lake Desktop CPUs Coming in 2024

PC World's senior editor, Mark Hachman, managed to corner an important Team Blue staffer during this week's Innovation 2023 event, where Core Ultra "Meteor Lake" processors were presented in an official manner. This CPU family is set to arrive on December 14, but starting out only in mobile...

www.techpowerup.com