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 (20A)	Arrow Lake (N3B)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop Only	Desktop & Mobile H&HX	Mobile U Only	Mobile H
Process Node	Intel 4	Intel 20A	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Q1 2025 ?	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	6P + 8E ?	8P + 16E	4P + 4E	4P + 8E
LLC	24 MB	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.)

 

[Thunder 57]

Hear me out... Lunar Lake is mopping the floor with Zen 5!

I'd be cautious about stating anything until there are 3rd party benchmarks. We all know how well that worked out for AMD last time...

 

[SiliconFly]

I currently use an Apple M1, but I think I might actually buy an LNL laptop. Need to see 3rd party benchmarks but it looks like it could be the Apple M moment of the x86 world and that iGPU looks like it can actually put out real 1080p gaming performance with reasonable settings and at 15-20 watts.

I'm confident that idle or low usage battery life will be great but still need to see what moderate use power really looks like. One concern is that Windows is fairly terrible for battery life and just not as well sorted as Mac OS. Hopefully that doesn't ruin LNLs potential.

One of my main concern is LNL is a one off product. There's not even a LNL refresh next year!

 

[igor_kavinski]

One of my main concern is LNL is a one off product. There's not even a LNL refresh next year!

Who cares? It will be a collector's item, well, at least a few million of them

 

[Hail The Brain Slug]

Who cares? It will be a collector's item, well, at least a few million of them

Does that mean the ones with particularly egregious defects are actually more desirable, like for coins and notes?

 

[SiliconFly]

You re talking as if we were back in 2001 at the time of single core era, what about the MT scores.?.
So far at 30W the HX 370 should outperform LNL by 70%, sorry for the few.

MT is pretty much useless for thin & light laptops. If you believe MT is more important than ST, then I think we should consider buying a laptop with a 288 core Sierra Forest CPU. That should make a difference.

 

[SiliconFly]

I don't know if this is as big of a factor as it's made out to be - Look at the Snapdragon Windows laptops and their battery life. In addition, we even have an HX370 Windows laptop that achieved an insane 16+ hours of wifi websurfing (Vivobook 14, notebookcheck review).

Windows on Arm is very different from Windows x86/x64. WoA kernel is relatively new, built from ground up with very less baggage. And more importantly, the kernel level stuff is pretty much controlled and locked down by M$ compared to it's x86/x64 counterparts. This may make it difficult for some software to work properly, but this helps Windows run exactly like how M$ intends... like macOS. ;-)

Even in Windows x86/x64, a fresh install of Windows with latest OEM drivers and without much or any tweaking of the OS helps it run better than more mature installs that have tons of software which sometimes do strange things that are hard to predict. Windows is good, but not that good.

 

[Abwx]

MT is pretty much useless for thin & light laptops. If you believe MT is more important than ST, then I think we should consider using 288 core Sierra Forest CPUs in our laptops. That should make a difference.

Technicaly it s much more important for laptops than for DT, when you are power limited the only way to increase perf and battery life is to inflate the core count and reduce frequency accordingly.

And Intel did present flawed number for battery life because they stated the duration without specifying the quantity of work that was processed.

FI with a 75Wh battery if a fast CPU manage to do 100 Cinebench runs in say 2 hours it will have better battery life than a slower CPU that does 50 runs within 3 hours, yet you can present the latter as having better battery life, that s what Intel did with their comparison for battery life, they didnt point that a CPU was doing the work more rapidly within its rated battery duration.

 

[poke01]

R24 number from core ultra 9

Love love the fact they probably told reviewers to exclude M3. Fastest mobile cores “period”, yeah right…

 

[poke01]

Don't worry. Good thing about Intel/AMD laptops is that they get discounted.

All laptops yes even Apple laptops get discounted.. but early adopters get shafted lol

 

[SiliconFly]

... 100 Cinebench runs ...

Still hung up on multi threading. Thats too Cinebench on a thin & light laptop? Yikes!

when you are power limited the only way to increase perf and battery life is to inflate the core count and reduce frequency accordingly.

No. When you are power limited the only way to increase perf is to use very performant and highly efficient CPU like Lunar Lake instead of slower CPUs like Zen 5 HX 370, etc.

 

[controlflow]

I don't know if this is as big of a factor as it's made out to be - Look at the Snapdragon Windows laptops and their battery life. In addition, we even have an HX370 Windows laptop that achieved an insane 16+ hours of wifi websurfing (Vivobook 14, notebookcheck review).

I'm probably being unfair to Microsoft since I haven't been using Windows laptops much recently but in the past I had an Intel based Macbook and a similar Intel CPU in a work windows laptop and the difference in battery life was stark. Additionally, the windows machine would drain massive amounts of power in sleep with the lid closed. The Apple Intel machine didn't suffer from those oddities.

Perhaps Microsoft has fixed those issues since then.

 

[controlflow]

Asus Zenbook S 14 OLED UX5406S impressions (2024 series, Intel Lunar Lake)

This article gathers my thoughts on the Asus Zenbook S 14 OLED UX5406S series, the mid-2024 addition to the Zenbook lineup of ultrabooks. This is not a

www.ultrabookreview.com

"Having used this early sample over the last few weeks, I have no complaints. This unit felt snappy with light use on battery power, lasted for a long while on a charge, and didn’t lose battery while in sleep mode even for a few days. I haven’t noticed any wifi issues while resuming from sleep either."

This looks promising, maybe Microsoft did fix that mess with connected standby.

 

[Abwx]

Still hung up on multi threading. Thats too Cinebench on a thin & light laptop? Yikes!


No. When you are power limited the only way to increase perf is to use very performant and highly efficient CPU like Lunar Lake instead of slower CPUs like Zen 5 HX 370, etc.

Not at all, one core using one thread at 5GHz can consume say 15W, two cores working at 2.5Ghz with one thread each will have the same throughput but at 6W total power, that s 2.5x better efficency at same processed work quantity.

And we can go beyond and double down the core count at 4 wich will reduce power to 2.4W for the same work at 6x better efficency.

Indeed badly multithreaded softwares is limiting efficency and battery life since ages, and unless there s a an effort from developpers improving CPUs efficency is a moot point since there s much more to gain with better softs.

As an exemple if you do a 7 Zip compression, wich is a common office task, a 8C/8T will be trounced by a 12C/24T, or even a 10C/20T, both in perf and in battery life impact.

 

[SiliconFly]

Not at all, one core using one thread at 5GHz can consume say 15W, two cores working at 2.5Ghz with one thread each will have the same throughput but at 6W total power, that s 2.5x better efficency at same processed work quantity.

Nope. Cos the primary thread gets real slow at 2.5GHz which hurts ST performance in thin and light laptops. And thats very significant. When in doubt, kindly refer to Apple M series silicon. ST is key in thin & lights. MT not so much. Go Lunar!

And we can go beyond and double down the core count at 4 wich will reduce power to 2.4W for the same work at 6x better efficency.

Indeed badly multithreaded softwares is limiting efficency and battery life since ages, and unless there s a an effort from developpers improving CPUs efficency is a moot point since there s much more to gain with better softs.

As an exemple if you do a 7 Zip compression, wich is a common office task, a 8C/8T will be trounced by a 12C/24T, or even a 10C/20T, both in perf and in battery life impact.

If you're worried about multi-tasking in laptops, I recommend you wait for ARL-H which again is expected to outperform competition in all aspects. Go Arrow!

 

[Abwx]

Nope. Cos the primary thread gets real slow at 2.5GHz which hurts ST performance in thin and light laptops. And thats very significant. When in doubt, kindly refer to Apple M series silicon. ST is key in thin & lights. MT not so much. Go Lunar!


If you're worried about multi-tasking, I recommend you wait for ARL-H which again expected to outperform competition in all aspects. Go Arrow!

That s the only way to improve battery life significantly, cap the single core frequency
at say 3GHz and the all cores at something like 2GHz and see the difference.

And btw, there was much more that was brought by fast SSD than by CPUs this last 15 years, so it s not mobile CPUs are that slow, remember the time of HDDs, i once swapped a 5" HDD in my C2D laptop for a SSD back in the days, and it was like the CPU became 3x faster.

You have surely noticed that after several CPUs gen battery life didnt improve as much as could have been hoped for, that s because pure perf has always been privileged for competitive reasons, till Apple came with quite better numbers for battery life, at wich point the X86 players started to react.

 

[SiliconFly]

... the windows machine would drain massive amounts of power in sleep with the lid closed. ...

I've had the same problem too with many Intel laptops. And I think it's more of a Intel+OEM issue rather than just Microsoft alone.

"Having used this early sample over the last few weeks, I have no complaints. This unit felt snappy with light use on battery power, lasted for a long while on a charge, and didn’t lose battery while in sleep mode even for a few days. I haven’t noticed any wifi issues while resuming from sleep either."

This looks promising, maybe Microsoft did fix that mess with connected standby.

Again, I think it's more of Intel+OEM issue rather than M$ by itself. Most of the power used during sleep by a Windows+Intel Laptop, goes to the RAM modules rather than any other components afaik.

Also, I dunno much about RAM power draw. But I think it may have something to do with switching from DDR memory to LPDDR in LNL. Here I'm assuming LPDDR has very less power draw compared to regular DDR during laptop standby. I maybe wrong. Can someone shed some light? Is LPDDR the key?

 

[SiliconFly]

That s the only way to improve battery life significantly, cap the single core frequency
at say 3GHz and the all cores at something like 2GHz and see the difference.

Why would anybody want to buy a high performance part and then increase battery life by decreasing its performance? Strange.

And btw, there was much more that was brought by fast SSD than by CPUs this last 15 years, so it s not mobile CPUs are that slow, remember the time of HDDs, i once swapped a 5" HDD in my C2D laptop for a SSD back in the days, and it was like the CPU became 3x faster.

System performance is very different from CPU performance. We're squarely focused on the latter. I would still prefer to have a very performant CPU like Lunar Lake in my laptop whenever possible.

... till Apple came with quite better numbers for battery life, at wich point the X86 players started to react.

Intel has reacted. Hence, Lunar Lake!

I hope someday competition reacts too (preferably with Zen 6).

 

[ondma]

Why would anybody want to buy a high performance part and then increase battery life by decreasing its performance? Strange.


System performance is very different from CPU performance. We're squarely focused on the latter. I would still prefer to have a very performant CPU like Lunar Lake in my laptop whenever possible.


Intel has reacted. Hence, Lunar Lake! I hope someday competition reacts too (preferably with Zen 6).

Yea, well lets just hope LL is not to little, to limited in scope, and too late. It is a niche product without any apparent follow up and certainly will not turn the company around.

 

[adroc_thurston]

Intel has reacted. Hence, Lunar Lake! I hope someday competition reacts too (preferably with Zen 6).

AMD has had tablet chips since 2021(-ish) in Van Gogh.
It's just a niche-niche market so not worth the ROI usually.

 

[Abwx]

Why would anybody want to buy a high performance part and then increase battery life by decreasing its performance? Strange.

Because CPU vendors and OEMs are reluctant to do so, as bigger perf numbers ultimately are way more marketable than battery life, athough there s silent/whisper options in current laptops wich should yield much better numbers than the official ones.

System performance is very different from CPU performance. We're squarely focused on the latter. I would still prefer to have a very performant CPU like Lunar Lake in my laptop whenever possible.


Intel has reacted. Hence, Lunar Lake! I hope someday competition reacts too (preferably with Zen 6).

It s Intel that is late, AMD has already reacted, btw Intel used an Asus 14" 3k LNL laptop to prop up their own numbers, granted the AMD variant has only a 60Hz 1080p OLED screen that help a lot, but still, we ll see in reviews if Intel can boast such numbers, and here it s a HX370 that can score 2x LNL CB score if required :

 

[511]

I've had the same problem too with many Intel laptops. And I think it's more of a Intel+OEM issue rather than just Microsoft alone.

Works fine on linux also AMD has this issue as well

Again, I think it's more of Intel+OEM issue rather than M$ by itself. Most of the power used during sleep by a Windows+Intel Laptop, goes to the RAM modules rather than any other components afaik.

Also, I dunno much about RAM power draw. But I think it may have something to do with switching from DDR memory to LPDDR in LNL. Here I'm assuming LPDDR has very less power draw compared to regular DDR during laptop standby. I maybe wrong. Can someone shed some light? Is LPDDR the key?

Nope MSFT is to blame why did it work for Apple and on Linux than ?

 

[511]

Btw for battery life geekerwan is best as his simulation is real life

 

[Magio]

Works fine on linux also AMD has this issue as well

Nope MSFT is to blame why did it work for Apple and on Linux than ?

I've been using Linux for more than a decade on a bunch of laptops, modern standby is consistently a pain in the ass, sometimes broken and often drains battery like mad. You may have gotten lucky on one of the few models that fares better but the state of standby on Linux is absolutely not all sunshine and rainbows.

 

[SiliconFly]

Nope MSFT is to blame why did it work for Apple and on Linux than ?

Apple uses their own custom hardware designs and their choice of select/premium components when compared to OEMs like HP, Dell, Lenovo, etc. All vertically integrated. Intel+OEM laptops can never match Apple's perfection.

But you getting Linux to sleep without any power drain on an Intel laptop is a bit intriguing. Not sure, but your Linux install may have defaulted to hybrid sleep or hibernate. And coupled with a SSD, it may not even be noticeable at times. Windows I think defaults to active standby. Just a guess.

(Not sure as I always shutdown my laptop every time to reduce the negative impact active standby has on overall battery life.)

 

[majord]

MT is pretty much useless for thin & light laptops. If you believe MT is more important than ST, then I think we should consider buying a laptop with a 288 core Sierra Forest CPU. That should make a difference.

This really makes no sense.

If I could do all the work I can do my desktop, on a thin and light laptop , why wouldn't I want that..

ST is important, and LNL has that in droves it seems, but this strength will drop off pretty quickly beyond 4 threads , and be completely DOA byond 8 obviously compared to other offerings. That's a bit lacklustre in 2024. And it means it gets flogged by other U class chips from a generation ago . Matched by some 2 gen's ago. That's a problem that's hard to shake. I'm certainly not forking out for something that can't do more real work than my 2+yr old machine, just because it can chomp through a ST workload faster.

IF ST is strong enough to also be performant in MT with a low core count, PLUS it's efficient enough to still have good perf/watt whilde doing it, then your point is valid, but none of these are true.. the ST is basically in line with everything else give or take. and it comes at a consumption we don't know yet.. Intel certainly ain't mentioning it in their slides.

I also can't help notice your inconsistencies and irony.. you're singing the praises of of C24 ST performance, but mocking Cinebench as a whole the next minute. Even though technically, rendering is a thing, and it's the ST portion of Cinebench that's the irrelevant one , since no rendering program in existence is limited to 1T