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	8 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)	Arrow Lake Refresh (N3B)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop Only	Desktop & Mobile H&HX	Desktop Only	Mobile U Only	Mobile H
Process Node	Intel 4	Intel 20A	TSMC N3B	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Q1 2025 ?	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2025 ?	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	6P + 8E ?	8P + 16E	8P + 32E	4P + 4E	4P + 8E
LLC	24 MB	24 MB ?	36 MB ?	?	8 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.)

 

[gdansk]

It is still Xe 12.7 like Arc. But competing with RDNA3 in ppw.
Is the GPU using more area than 780M? Is RDNA3 that handicapped without last level cache? Other contributing factors?

 

[Abwx]

View attachment 90365
View attachment 90367

1370P at same FPS than the 7840U..?..

 

[Khato]

1370P at same FPS than the 7840U..?..

While it certainly is a game, League of Legends isn't particularly demanding of the GPU. The FPS is more representative of CPU + driver performance.

The most amusing of the slides in my opinion was the failure of the non-Intel AI solutions to seamlessly work as promoted. Having more TOPS available doesn't do much good if software isn't making use of them.

 

[Hitman928]

I doubt Intel shares anything more of significance about MTL tomorrow, but we'll see. Today's presentation obviously painted MTL in a much better light on the CPU side than the leaks (GPU seems about in line from the leaks) so we'll have to wait and see from 3rd party reviewers which numbers end up being more accurate.

 

[Abwx]

Having more TOPS available doesn't do much good if software isn't making use of them.

Moreover when it s your competitor that is doing the "test", you know, the same one that once talked of snake oil..

Anyway they state that both MTL and 7840U are 30W CPUs, for the AMD part that s likely but for Intel we know what is their conception of TDP when it comes to their own products, we ll know when real reviews drop...

 

[Khato]

Anyway they state that both MTL and 7840U are 30W CPUs, for the AMD part that s likely but for Intel we know what is their conception of TDP when it comes to their own products, we ll know when real reviews drop...

Agreed, it's unfortunate that they didn't specify PL1=PL2=30W for the MTL system. Certainly would be nice if systems were provided to some reviewers in advance of launch to provide launch day reviews. Been awhile since we've had that.

 

[Abwx]

Agreed, it's unfortunate that they didn't specify PL1=PL2=30W for the MTL system. Certainly would be nice if systems were provided to some reviewers in advance of launch to provide launch day reviews. Been awhile since we've had that.

If you think that it was running at 30W you are good for a hard landing, guess that s why there was nothing sent to reviewers, better to control the infos as long as possible if things are not as rosy as in marketing parades..

 

[Hitman928]

Comparing to the HXL Timespy leak, either the HXL leak was a complete fabrication, or the live demo Timespy we saw today with the 155H was run at like 70W.

 

[controlflow]

If you think that it was running at 30W you are good for a hard landing, guess that s why there was nothing sent to reviewers, better to control the infos as long as possible if things are not as rosy as in marketing parades..

I have no idea what power MTL was using in those demos, but why is it so hard for you to even pretend to be objective? You have no idea what power MTL was using and you have no data to suggest that MTL will be much more inefficient than how it was represented in the demo today. You also don't know that laptops weren't sent to reviewers, it could be that samples were sent to reviewers and they are under embargo until tomorrow's reveal.

The details will become clear soon enough but making statements in confidence when you have nothing to back up that confidence is pretty silly.

 

[H433x0n]

guess that s why there was nothing sent to reviewers, better to control the infos as long as possible if things are not as rosy as in marketing parades..

What makes you think nothing was sent to reviewers?

 

[jpiniero]

What makes you think nothing was sent to reviewers?

Yah, If it's going to be available for sale tomorrow I'd expect reviews tomorrow.

 

[Geddagod]

Did I just hear that the NPU is 10 TOPS?

IIRC AMD says Phoenix is also 10 TOPS, and Hawkpoint is like 15 TOPS or higher, idk.

Hallock claiming 30-35% longer battery life with video playback over RPL.

Also:

Anyone know a good battery life test chart for laptops?

View attachment 90360

8% better ST, noice. I expect ST in general to be basically a draw though overall.

Steam quality is too poor to keep taking screenshots. 155H scored about 25% higher than the 7840u in CB24 multi. No power numbers given outside of saying they are both ~30W CPUs.

I am very wary of boost power being different for the two CPUs benched.

I will wait for benchmarks from reputable sites before I comment.

And yet you can't help yourself from commenting when @adroc_thurston leaks something. Oh the irony lol

View attachment 90361

No idea how Intel got this graph tbh.

View attachment 90365
View attachment 90367

Nice they didn't just cherry pick a couple games.

What makes you think nothing was sent to reviewers?

Cuz Intel bad and mean. Snake Oil marketing grrrrr
Also @Abwx Intel said MTL was the most efficient x86 laptop CPU for thin and lights. Obviously this isn't conclusive of anything, but ik you were talking about how MTL must be a dud because Intel said MTL was their most efficient laptop CPU rather than just the most efficient laptop CPU or something, idk.

View attachment 90361

Lastly, while I did say idk how they got this graph, is the best PPW they can get out of MTL vs RPL on the PPW curve is 8%, that is pretty bad. It's not like, wtf are you doing Intel levels of bad, but AMD managed a ~15% perf/watt uplift IIRC from Zen 3 to Zen 4 from node alone, and I strongly believe Intel has had much more room to improve perf/watt between RPC and RWC than AMD had between Zen 3 and Zen 4.

 

[H433x0n]

Lastly, while I did say idk how they got this graph, is the best PPW they can get out of MTL vs RPL on the PPW curve is 8%, that is pretty bad. It's not like, wtf are you doing Intel levels of bad, but AMD managed a ~15% perf/watt uplift IIRC from Zen 3 to Zen 4 from node alone, and I strongly believe Intel has had much more room to improve perf/watt between RPC and RWC than AMD had between Zen 3 and Zen 4.

If you didn’t watch the video it was SIR2017 at 30W(?) isopower in an Intel test lab on a test socketable(?) motherboard.

Just rewatched it and it didn’t say specific power limit, so who knows.

 

[Markfw]

And yet you can't help yourself from commenting when @adroc_thurston leaks something. Oh the irony lol

Trying to not assume about anything Intel lately with their almost zero delivery of things. AMD on the other hand have over-delivered for YEARS.

 

[Geddagod]

Trying to not assume about anything Intel lately with their almost zero delivery of things. AMD on the other hand have over-delivered for YEARS.

So lemme get this straight. Your order of believability is 3rd party benches > Anandtech Leaker with no past credibility > Info from Intel? That's wild.
Also, over delivered... hmm. AMD didn't seem to overdeliver with RDNA 3 did they?

 

[adroc_thurston]

AMD didn't seem to overdeliver with RDNA 3 did they?

Esoteric copium hit.
That's not a CPU.

That's wild.

Yeah, at least Shrout is no longer with Intel.

 

[Abwx]

I have no idea what power MTL was using in those demos, but why is it so hard for you to even pretend to be objective? You have no idea what power MTL was using and you have no data to suggest that MTL will be much more inefficient than how it was represented in the demo today. You also don't know that laptops weren't sent to reviewers, it could be that samples were sent to reviewers and they are under embargo until tomorrow's reveal.

The details will become clear soon enough but making statements in confidence when you have nothing to back up that confidence is pretty silly.

They gave some numbers wich allow to know the extent of the power used for all CPUs, this is right on the slides.

They state that MTL perform 8% better than the 1370P at same power, so you get the idea of the perf/watt improvement, this mean about 17-18% lower power at same perf, but that s not all...

They also state that the 1370P has 2% better MT perf than the 7840U, wich only happen if the 1370P pull roughly 45W while the 7840U is at 28W or so.

Their statement that MTL perform 11% better than the 7840U is with MTL at 45W and the 7840U at 28W.

As usual with Intel they bench their CPU at 45W FI then compare it to the competition at 28W all while claiming that their CPU is also a 28W one...

 

[Geddagod]

Esoteric copium hit.
That's not a CPU.

I'm not talking about just CPUs tho. Even AMD manages to blunder, and besides, even that's not even relevant in comparing the relevancy to a leaker with no history vs marketing material. Even if AMD always hits perf targets, the perf targets or ES scores you mention might not even be true.
The true copium is willing to believe a leaker of no past history over info from Intel because he wants MTL to fail so bad lol.
Oh, and don't take this personally. I'll eat crow if Zen 5 does manage to be as amazing as you make it out to be, just like I did with @Exist50

Yeah, at least Shrout is no longer with Intel.

Was Ryan Shrout that bad? Lol I never really paid attention

 

[adroc_thurston]

Even AMD manages to blunder

RDNA3 is a very specific thing that caused plenty people to get sacked.
You're not allowed to blunder at AMD.

because he wants MTL to fail so bad lol.

There's nothing interesting about MTL.
Middling perf, high costs.
Useless.
If you want to shill Intel client parts, focus on LNL-onwards, these aren't marred with insane design decisions (they're all sane and very straightforward).
Server, well, Intel is dead.

I'll eat crow if Zen 5 does manage to be as amazing as you make it out to be

I'm actually underselling it, real numbers are higher.

Was Ryan Shrout that bad? Lol I never really paid attention

Arguably the worst era of Intel marketing, full of very unhealthy obsession with AMD.
That TGL slideware deck was memey enough to get a whole gamersnexus video.

 

[H433x0n]

There's nothing interesting about MTL.
Middling perf, high costs.
Useless.

That's just an opinion man. It's the first time Intel has used this new node and introduced a disaggregated architecture on foveros in volume. The performance of the cores themselves are probably underwhelming but it doesn't mean it's not new and interesting.

If you want to shill Intel client parts, focus on LNL-onwards, these aren't marred with insane design decisions (they're all sane and very straightforward).

@Geddagod doesn't shill anything. It's possible to not actively hate either company.

 

[adroc_thurston]

the first time Intel has used this new node and introduced a disaggregated architecture on foveros in volume

See, those things are just means to an end.
Plus passive 2.5D slabs aren't anything interesting or impressive, quite literally 2012 tech.

The performance of the cores themselves are probably underwhelming

That's the issue.
Cool novelty tech is only worth it when the end product smacks hard.
See PVC vs MI300 family of gizmos.

doesn't shill anything. It's possible to not actively hate either company.

I don't hate Intel, they're just utterly inept in markets most relevant to me (server/edge/networking).
Once again, if you want of focus on Intel parts, do LNL-onwards, that stuff is sane.

 

[Geddagod]

Cool novelty tech is only worth it when the end product smacks hard.

Things can be interesting even if they are bad. CNL is a insanely bad product but I'm extremely curious abt it. Someone give me a techinsights subscription >:c

 

[adroc_thurston]

Things can be interesting even if they are bad

Trainwrecks are only entertaining for a tiny short while, but a kickass product is eternal.

 

[Markfw]

So lemme get this straight. Your order of believability is 3rd party benches > Anandtech Leaker with no past credibility > Info from Intel? That's wild.
Also, over delivered... hmm. AMD didn't seem to overdeliver with RDNA 3 did they?

I did not even comment on it, and you are all over my case.

Chill dude.

 

[Geddagod]

I did not even comment on it, and you are all over my case.

Chill dude.

Ma fault, but you do realize how... uhh... wild , let's call it that, it sounds for someone to say, I won't comment on stuff from Intel without confirmation, but I will comment on stuff from random leakers, right? Like I'm not even gonna say anything else abt it tho if you want me to chill lol

Trainwrecks are only entertaining for a tiny short while, but a kickass product is eternal.

If Zen 5 is as kickass as you say, no one is gonna be able to afford it
Esp when ARL leaks are so unimpressive lol