Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads

[Tigerick]

As Hot Chips 34 starting this week, Intel will unveil technical information of upcoming Meteor Lake (MTL) and Arrow Lake (ARL), new generation platform after Raptor Lake. Both MTL and ARL represent new direction which Intel will move to multiple chiplets and combine as one SoC platform.

MTL also represents new compute tile that based on Intel 4 process which is based on EUV lithography, a first from Intel. Intel expects to ship MTL mobile SoC in 2023.

ARL will come after MTL so Intel should be shipping it in 2024, that is what Intel roadmap is telling us. ARL compute tile will be manufactured by Intel 20A process, a first from Intel to use GAA transistors called RibbonFET.



Comparison of upcoming Intel's U-series CPU: Core Ultra 100U, Lunar Lake and Panther Lake

Model	Code-Name	Date	TDP	Node	Tiles	Main Tile	CPU	LP E-Core	LLC	GPU	Xe-cores
Core Ultra 100U	Meteor Lake	Q4 2023	15 - 57 W	Intel 4 + N5 + N6	4	tCPU	2P + 8E	2	12 MB	Intel Graphics	4
?	Lunar Lake	Q4 2024	17 - 30 W	N3B + N6	2	CPU + GPU & IMC	4P + 4E	0	12 MB	Arc	8
?	Panther Lake	Q1 2026 ?	?	Intel 18A + N3E	3	CPU + MC	4P + 8E	4	?	Arc	12

Comparison of die size of Each Tile of Meteor Lake, Arrow Lake, Lunar Lake and Panther Lake

	Meteor Lake	Arrow Lake (N3B)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop & Mobile H&HX	Mobile U Only	Mobile H
Process Node	Intel 4	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	8P + 16E	4P + 4E	4P + 8E
LLC	24 MB	36 MB ?	12 MB	?
tCPU	66.48
tGPU	44.45
SoC	96.77
IOE	44.45
Total	252.15

Intel Core Ultra 100 - Meteor Lake



As mentioned by Tomshardware, TSMC will manufacture the I/O, SoC, and GPU tiles. That means Intel will manufacture only the CPU and Foveros tiles. (Notably, Intel calls the I/O tile an 'I/O Expander,' hence the IOE moniker.)

 

[MoistOintment]

I don't know why people still watch that butthead. He must make decent money pulling things out of his rear while dumbtards hang on his every word.

I feel like I have to be up to date on what he's saying so that I can recognize when people repeat his claims in forums

 

[MoistOintment]

This is it for x86 i guess first Zen5% now core ultra 2% 💀

I think more so a shift towards focusing on IPC instead of clockspeeds will be beneficial for mobile and server, even if that means enthusiast desktop will suffer in the transition.

Zen 5 certainly saw great improvements in server. Strix is great in performance mobile. LNL is great in ultra mobile, ARL-H will also likely be good in performance mobile where it'll see a bigger uplift vs MTL-H than ARL-S sees vs RPL-S.

I'm not concerned about the death of x86. I see x86 responding to the threat of ARM chips in laptop and datacenter, while enthusiast desktop, their strongest and safest market, will be temporarily stagnant.

 

[511]

I think more so a shift towards focusing on IPC instead of clockspeeds will be beneficial for mobile and server, even if that means enthusiast desktop will suffer in the transition.

Zen 5 certainly saw great improvements in server. Strix is great in performance mobile. LNL is great in ultra mobile, ARL-H will also likely be good in performance mobile where it'll see a bigger uplift vs MTL-H than ARL-S sees vs RPL-S.

I'm not concerned about the death of x86. I see x86 responding to the threat of ARM chips in laptop and datacenter, while enthusiast desktop, their strongest and safest market, will be temporarily stagnant.

Yeah i mean Arm will not bother for enthusiast market for a while it will be the last pie they want cause it is a nightmare vs laptop for sure

 

[Hitman928]

Well, that kinda sucks. There were some really bold predictions here recently that just turned sour real fast.

 

[KompuKare]

Well, that kinda sucks. There were some really bold predictions here recently that just turned sour real fast.

While a post mortem is premature until actual reviews, I notice some of the big hype train drivers seem to have made themselves scares now.

And that after going after anyone more cautious with great zeal only a short while ago!

 

[Josh128]

This is it for x86 i guess first Zen5% now core ultra 2% 💀

Negative 2%. 🤷‍♂️

 

[Wolverine2349]

If this is the best Intel can show in their own benchmarks, the 285k is DOA for gamers

View attachment 108889

Yes that would be disaster. I sold my 7800X3D, a backup system and downgrading a backup system and downgraded to a basic laptop from a gaming laptop to get more cores in hopes Arrow Lake and breaking even on money is on par gaming of 7800X3D. I sure hope those slides are not true. Or if they are its severely gimped RAM and clock speed.

I have not purchased anything but in a position to make a decision awaiting Arrow Lake release.

 

[AcrosTinus]

I might have fallen for the doom and gloom side prematurely.
The best case scenario is that this is a severely nerved power profile, like 177 or 180W and with a high performance profile that is easily accessible the, full potential can be unleashed.

This might just be a pipe dream though due to being on a foreign efficiency tuned node... with the feared tile latency that I hoped would not be a thing.

 

[Saylick]

btw i find it really deafening there aint a single word about this leak over at wccftech.. wanted to check the breakdown in the comments section
guess some PR money found its way over there 😅

WTFtech is in cahoots with Intel… they were one of the publications that got exclusive access to a Meteorlake briefing before everyone else. I suspect they have an ARL review sample already and don’t want to sour their relationship with Intel by commenting on Arrowlake leaks prior to launch while presumably under NDA. But yeah, obvious PR is obvious.

 

[Wolverine2349]

I was trying to warn people, but did not want to push it on Intel thread.

Lunar Lake vs. Meteor Lake -> 1x latency improvement
Arrow Lake vs. Raptor Lake -> 1x latency deterioration

So, any conclusions about Arrow Lake based on Lunar Lake performance improvements were going to lead to a big disappointment.

Are you saying warning us Arrow Lake will be disappointing based on Lunar Lake results, or that we cannot base Lunar Lake results on how Arrow Lake will perform?

 

[AcrosTinus]

Yes that would be disaster. I sold my 7800X3D, a backup system and downgrading a backup system and downgraded to a basic laptop from a gaming laptop to get more cores in hopes Arrow Lake and breaking even on money is on par gaming of 7800X3D. I sure hope those slides are not true. Or if they are its severely gimped RAM and clock speed.

I have not purchased anything but in a position to make a decision awaiting Arrow Lake release.

I sold my 13700k and 12900K systems for this as well, hence my panic. I just don't like regressions in any aspect when the time is moving forward, maybe I'm just spoiled. But that is why I could not buy Rocket Lake.

 

[igor_kavinski]

How can I tell at what settings each system is?

9950X with 6400 MT/s 32GB RAM while 285K has denser 48GB RAM @ 7200 MT/s.

The 9950X is almost handicapped there. Happy?

 

[Wolverine2349]

I think more so a shift towards focusing on IPC instead of clockspeeds will be beneficial for mobile and server, even if that means enthusiast desktop will suffer in the transition.

Zen 5 certainly saw great improvements in server. Strix is great in performance mobile. LNL is great in ultra mobile, ARL-H will also likely be good in performance mobile where it'll see a bigger uplift vs MTL-H than ARL-S sees vs RPL-S.

I'm not concerned about the death of x86. I see x86 responding to the threat of ARM chips in laptop and datacenter, while enthusiast desktop, their strongest and safest market, will be temporarily stagnant.

IPC improvements will be just as good as clock speed improvements and why would enthusiast desktop suffer?

Conroe had like almost double IPC of Netburst worse clocks at first and it was a marvell on enthusiast desktop.

 

[OneEng2]

If this is the best Intel can show in their own benchmarks, the 285k is DOA for gamers

View attachment 108889

AMD has moved up their 9XXX X3D launch. IMO, this is what is going to be Intel's biggest headache for gamers.

Considering that they've got rid of HT to eke out more Mhz in ST and MT this is not very likely. Probably the same OC potential as with AMDs offering on TSMC processes

I am surprised at the ST performance as I assumed that by going to a higher performance process and removing SMT in order to use the transistor budget more effectively on improving IPC that Arrow Lake would indeed have very good single thread performance. My GUESS is that moving to tiles has killed their latency figures and this is hurting the performance.... but just a guess.

I've said this before, but this outcome shouldn’t be a surprise. RPL was essentially a swan song of Intel's approach to desktop CPUs: a monolithic chip that does not suffer from penalties typical to disaggregated architectures (higher mem and C2C latencies), setup with no care for power consumption on a node that clocks sky high like no other, pushed to the absolute limits and beyond of its silicon. It’s a perfect setup for gaming.

Where this really hurt them was in the data center. When you pack a ton of cores into a big data center die, you quickly become thermally limited on a design like this.

I think more so a shift towards focusing on IPC instead of clockspeeds will be beneficial for mobile and server, even if that means enthusiast desktop will suffer in the transition.

Zen 5 certainly saw great improvements in server. Strix is great in performance mobile. LNL is great in ultra mobile, ARL-H will also likely be good in performance mobile where it'll see a bigger uplift vs MTL-H than ARL-S sees vs RPL-S.

I'm not concerned about the death of x86. I see x86 responding to the threat of ARM chips in laptop and datacenter, while enthusiast desktop, their strongest and safest market, will be temporarily stagnant.

Exactly this. Everyone here is disappointed in the single threaded performance; however, it is in the server markets where all the profits are (and growth). It is in this market where Intel will need to be making up that lost cash. If the server variants of the new core perform well, then the bet will have worked. My repeated concern here is that the lack of SMT in a data center CPU looks an awful lot like a bad design choice. I believe that the die space and thermal efficiency trade offs for SMT are a big win in highly threaded work loads.

We will see fairly soon I guess.

 

[511]

SPIE Monterey- ASML, INTC – High NA Readiness- Bigger Masks/Smaller Features - Semiwiki

- SPIE Photomask -all about High NA and bigger masks - High NA will ramp very fast & is ramping fast - two already at Intel - Doubling size of photomask will offset High NA exposure size problem - Assembling litho tools at customer saves time/money-

semiwiki.com

At least Intel foundry is not struggling it's all Intel design now

 

[Kocicak]

9950X with 6400 MT/s 32GB RAM while 285K has denser 48GB RAM @ 7200 MT/s.

The 9950X is almost handicapped there. Happy?

Where did you find these RAM speeds? Without knowing the power profiles these CPUs run at this comparison still makes no sense.

BTW I commented in August here that Arrow lake will be more about features than raw computing performance, so the fact 285K will be "only on par" with 9950X does not bother me at all.

Cheaper Arrow lake CPUs with good integrated graphics will be a killer product for lower end PCs.

Also comparison with 14900K does not look good mainly because of the self-destructively high frequencies these CPUs have been shipping with, if they had been set more reasonably, the new CPUs would look better.

 

[igor_kavinski]

Where did you find these RAM speeds?

ASUS System Product Name - Geekbench

Benchmark results for an ASUS System Product Name with an Intel Core Ultra 9 285K processor.

browser.geekbench.com

Gigabyte Technology Co., Ltd. X670E AORUS PRO X - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. X670E AORUS PRO X with an AMD Ryzen 9 9950X processor.

browser.geekbench.com

 

[511]

The more i think about the iGPU the 4 Xe core gpu is on same level as TGL iGPU and the Media Support is outstanding with QSV

 

[MoistOintment]

Exactly this. Everyone here is disappointed in the single threaded performance; however, it is in the server markets where all the profits are (and growth). It is in this market where Intel will need to be making up that lost cash. If the server variants of the new core perform well, then the bet will have worked. My repeated concern here is that the lack of SMT in a data center CPU looks an awful lot like a bad design choice. I believe that the die space and thermal efficiency trade offs for SMT are a big win in highly threaded work loads.

We will see fairly soon I guess.

Depends. Datacenter will be skipping Lion Cove. Diamon Rapids will be on Panther / Coyote Cove.

Intel made a big deal in their original LNL presentation back in June about how they can have an SMT variant of their P core for datacenter while leaving SMT completely out of client at a hardware level. They didn't, however, mention any specific architectures where we would see this split.

 

[Wolverine2349]

Where did you find these RAM speeds? Without knowing the power profiles these CPUs run at this comparison still makes no sense.

BTW I commented in August here that Arrow lake will be more about features than raw computing performance, so the fact 285K will be "only on par" with 9950X does not bother me at all.

Cheaper Arrow lake CPUs with good integrated graphics will be a killer product for lower end PCs.

Also comparison with 14900K does not look good mainly because the self-destructively high frequencies these CPUs have been shipping with, if they had been set more reasonably, the new CPUs would look better.

On par in gaming with with 9950X is a disaster. The 9950X is barely better than 7950X in gaming (that is assuming dual CCD scheduling fix for 9950X comes out) and quite a bit behind 13th and 14th Gen. And with its dual CCD scheduling crap, 9950X way behind even 9700X. If its on par with that Arrow Lake desktop is worse than dead on arrival if its true. I hope its not.

If Arrow Lake is on par with 13h and 14th Gen in gaming with more features and lower heat output then ok its fine and actually good.

13th and 14th Gen already trade blows with 7800X3D or a little behind and well tuned they jump ahead. SO enthusiasts will buy it if on par with 13th and 14th Gen with the stability fixes and lower heat. Now if its on par gaming wise with vanilla Zen 5, ouch that s very bad.

 

[gdansk]

Looks like it comes down to pricing. Which could be good for consumers but how much room does Intel have on pricing when using interesting construction on newer, expensive node?

Bartlett Lake S might actually be more interesting than some people expected.

 

[coercitiv]

The 9950X is barely better than 7950Xin gaming and quite a bit behind 13th and 14th Gen. And with its dual CCD scheduling crap, 9950X way behind even 9700X.

Yes, waaaaay behind the 9700X.

 

[Kocicak]

I am really curious what will be the difference in performance between the two power limits - 250W is 40% more than 177, so it could mean something like 7 or 8% higher performance?

 

[511]

This video is damm funny

 

[igor_kavinski]

I am really curious what will be the difference in performance between the two power limits - 250W is 40% more than 177, so it could mean something like 7 or 8% higher performance?

May not be much in the intial reviews coz I don't think anyone will be using a CUDIMM RAM kit. Highest we may see is DDR5-8600 if we are lucky. I have hopes that it will be formidable in MT workloads with DDR5-9600 CUDIMM kits.