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

Tigerick · Aug 22, 2022

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.)

Abwx · Apr 11, 2024

On the same page we can read that :

As we can see from our rendering results, the Intel Core Ultra 7 155H is very competitive in single-threaded performance and is ahead of AMD's Zen 4 mobile Phoenix-based Ryzen 9 7940HS

And the related graph about this aheadness :

Must be some pavlovian reflex...

But that s not all, also :

Looking at performance in our web and office-based testing, in the UL Procyon Office-based tests using Microsoft Office, the Core Ultra 7 155H is actually outperforming the AMD Ryzen 9 7940HS, which is a good win in itself.

So that is, if less is more that s indeed quite a good win...

FlameTail · Apr 11, 2024

Lunar Lake being on N3B is great...

Because it finally allows us to do an isometric node comparison of Apple vs Intel.

M3 vs Lunar Lake

Both on N3B

We'll be able to judge how good Intel's architecture is (vs Apple's), how much power efficient it is, how much performant it is, and how much area efficient it is (how much bloat?).

Abwx · Apr 11, 2024

Looking at the review there s even more counter truths, here another exemple with the commented graph :

it remains competitive in the SPECint2017 section of our single-thread testing against the Ryzen 9 7940HS. The AMD chip performs better in two of the tests (525.x264_r and 548.exchange2_r);

The 7940HS is ahead in 7 out 10 tests...

And then about SpecFP with the related graph:

, we again see a very competitive showing in SPECfp2017 between the Intel Core Ultra 7 155H and the AMD Ryzen 9 7940HS. The only test we see a major gain for the Ryzen 9 7940HS is in the 503.bwaves_r test

Here the 7940HS is ahead in 9 out of 12 tests, seems that the reviewer is looking at the bars absolute lengths rather than the relative difference in each test to state where there s a large advance or not.

In the MT test he somewhat cooled down his flawed assumptions since he surely noticed the bars lengths more accurately, anyway that look more like some paid review than anything else.

Ghostsonplanets · Apr 11, 2024

Yeah, the more I read, the more errors can be noticed.

Hulk · Apr 11, 2024

SiliconFly said:
Well, the jury is still out on that one. But it appears the newer models are working well as advertised. Something has changed apparently. Wish someone can shed more light on that.

Well, battery life is one thing. But performance wise, it's barely on par with previous gen. If its efficiency is as good as advertised, it's definitely a decent buy. But not "top notch" imho.

Battery life is important for me, especially when performance is good. Furthermore, ST performance is really good for MTL, which implies two things. First, multicore performance is more limited by form factor/thermals, which is bad. Second, the architecture is good so process improvements might lead to better MT performance.

Hitman928 · Apr 11, 2024

Ghostsonplanets said:
Yeah, the more I read, the more errors can be noticed.

You mean like this result where the multi-threaded results are lower than the single threaded (and the 14600k seems not to have completed at all)?

Hitman928 · Apr 11, 2024

Hulk said:
Battery life is important for me, especially when performance is good. Furthermore, ST performance is really good for MTL, which implies two things. First, multicore performance is more limited by form factor/thermals, which is bad. Second, the architecture is good so process improvements might lead to better MT performance.

Thermals are fine. ST is lower than RPL and equal to slower than Zen4 (but still uses a lot of power to get there). This particular laptop just is configured for shorter turbo windows and restricts the power consumption into actual "U" model territory which MTL performs poorly in (but is good for battery life). Other MTL models let the chip use higher power and sustain boosts longer.

Panino Manino · Apr 11, 2024

Hitman928 said:
They are not really comparable models. Anandtech is terrible about this in laptop reviews to the point that they use desktop CPUs in their benchmark comparisons (probably because they do so few so their comparison options are limited). The AMD unit has an RTX 4070 in it and is meant as a gaming laptop. The Zenbook with MTL is meant as an ultraportable and geared for battery life. That's why the MTL gets crushed in the performance benchmarks but wins convincingly in the battery life ones.

MTL should not lose this bad in performance, but the configurations of the laptops are basically at opposite ends of the spectrum. Using a more comparable model with an AMD chip would see the performance be much closer, but battery life would be roughly equal as well.

I don't thing this is the only reason. Those two low power isolated cores does make the difference here.
At least in light use like this.

Hitman928 · Apr 11, 2024

Panino Manino said:
I don't thing this is the only reason. Those two low power isolated cores does make the difference here.
At least in light use like this.

Light use like what? Offline video, sure. But even then, the difference isn't as big as you'd expect.

Hitman928 · Apr 11, 2024

Abwx said:
snip

Looks like they've quietly updated the office and rendering comments. The Spec comments are still the same.

Looking at performance in our web and office-based testing, in the UL Procyon Office-based tests using Microsoft Office, the Core Ultra 7 155H is competitive but lags behind the AMD chips slightly.

As we can see from our rendering results, the Intel Core Ultra 7 155H is quite competitive in single-threaded performance and is a little behind AMD's Zen 4 mobile Phoenix-based Ryzen 9 7940HS.

reggie_fils_aime · Apr 11, 2024

Hitman928 said:
They are not really comparable models. Anandtech is terrible about this in laptop reviews to the point that they use desktop CPUs in their benchmark comparisons (probably because they do so few so their comparison options are limited). The AMD unit has an RTX 4070 in it and is meant as a gaming laptop. The Zenbook with MTL is meant as an ultraportable and geared for battery life. That's why the MTL gets crushed in the performance benchmarks but wins convincingly in the battery life ones.

MTL should not lose this bad in performance, but the configurations of the laptops are basically at opposite ends of the spectrum. Using a more comparable model with an AMD chip would see the performance be much closer, but battery life would be roughly equal as well.

This has been a hallmark of every MTL review I've seen - weird comparison points and a general lack of context outside of "it's faster than what we had lying around". It smacks of not wanting to rain on OEM parades, very strange stuff.

dullard · Apr 11, 2024

FlameTail said:
Lunar Lake being on N3B is great...

Because it finally allows us to do an isometric node comparison of Apple vs Intel.

M3 vs Lunar Lake

Both on N3B

We'll be able to judge how good Intel's architecture is (vs Apple's), how much power efficient it is, how much performant it is, and how much area efficient it is (how much bloat?).

Only partly. Intel has high density libraries and high performance libraries. Sometimes a third library. These are optimized for different types of uses for different chips (optimized for performance, optimized for cost, optimized for power, etc). M3 vs Lunar Lake would only give a comparison of one of those libraries.

Ghostsonplanets · Apr 11, 2024

Can't recommend enough Geekerwan review:

Hitman928 · Apr 11, 2024

reggie_fils_aime said:
This has been a hallmark of every MTL review I've seen - weird comparison points and a general lack of context outside of "it's faster than what we had lying around". It smacks of not wanting to rain on OEM parades, very strange stuff.

Closest equivalent model I could find on the AMD side is the Zenbook S13 with a Zen 3 based 6800u. Here are the battery life test results(reviews here and here):

The Zenbook 14 does have higher performance (as expected). Even this isn't an exact comparison because (for the who knows how many times now) a battery life test is a whole system comparison. Everything from the screen, memory, Wi-Fi module, and even voltage regulators come into play, not to mention how the OEM has configured the firmware for the device. MTL, when configured for it, can give great battery life, no doubt, it's just not really a step above what we've seen previously (that distinction solely belongs to Apple for the time being).

SiliconFly · Apr 11, 2024

FlameTail said:
Lunar Lake being on N3B is great...

Because it finally allows us to do an isometric node comparison of Apple vs Intel.

M3 vs Lunar Lake

Both on N3B

We'll be able to judge how good Intel's architecture is (vs Apple's), how much power efficient it is, how much performant it is, and how much area efficient it is (how much bloat?).

Both M3 & LNL are SoC. Too many variables which makes direct comparison impossible. Isometric or otherwise.

reggie_fils_aime · Apr 11, 2024

Hitman928 said:
Closest equivalent model I could find on the AMD side is the Zenbook S13 with a Zen 3 based 6800u. Here are the battery life test results(reviews here and here):

View attachment 96775

The Zenbook 14 does have higher performance (as expected). Even this isn't an exact comparison because (for the who knows how many times now) a battery life test is a whole system comparison. Everything from the screen, memory, Wi-Fi module, and even voltage regulators come into play, not to mention how the OEM has configured the firmware for the device. MTL, when configured for it, can give great battery life, no doubt, it's just not really a step above what we've seen previously (that distinction solely belongs to Apple for the time being).

Ah, I understand perfectly now. We'll never see a straight comparison review for MTL because it gets washed by years-old chips in significantly cheaper machines. Nobody in their right mind (especially those in charge of procurements) would choose MTL over, well, anything.

This isn't quite Bulldozer-levels of underwhelming, but it's damn close to Netburst in terms of reality-bending equivocation from professional outlets.

SiliconFly · Apr 11, 2024

dullard said:
...Sometimes a third library.

Don't remember exactly. But I think they're referred to as high performance HD cells (apart from HD, HP & UHP). Has to be in between HP & HD.

Saylick · Apr 11, 2024

DavidC1 said:
Only guy worse than MLID for "leaks" is that guy. Somebody said he had over 10K posts or something?

If you talk bad about Intel he will badmouth you. If you post any Intel info he begs you for info.

Yep, over 100k posts. Actually, almost 120k now, and the bulk of them are reposting pro-Intel marketing or PR statements. After the Intel AI event, he probably tacked on another 20 or so posts, all of which were reposting other people's reporting on the event.

Tigerick said:
Hoho, you guys are too fast. I have verified with my source, most of the info are correct: That's mean some of old info are no longer valid:-

PTL-H comes with 3 tiles, not 2 as previously leaked: CPU, GPU and PCD tiles.

CPU Tile: This round CPU tile is a main tile with 2 D2D connectors: CPU tile is connecting with PCD and GPU tiles. CPU tile consists of 4P+8E+4LPe core with NPU, MC, media/display engine and IPU. 4 LPe can function as MTL; they are being used to shut down main CPU. NPU should have ~48TOPS.

GPU Tile: Xe3 iGPU has 120 TOPS with 12Xe cores. 40-50% faster is believable cause Xe3 comes with extra 50% core.

PCD is the name Intel used to replace SoC tile. Consists of PCIE, USB, TB, sound, display, imaging, wifi/bt, security complex, sensor hub and some other minor stuff. And indeed, is made by N6.

"Up to 40% lower power and 3 hours more battery life." Sadly, Intel don't specify which one. But I believe it is compared to ARL-H, not LNL due to 3 tiles design...

WAIT, so you're actually saying this Prakhar guy actually has legit info??

SiliconFly · Apr 11, 2024

reggie_fils_aime said:
Ah, I understand perfectly now...

I don't think you do. MTL is far ahead than previous gen in terms of battery usage. Far far ahead. But not much performance gains though. And it doesn't have significant competition in the x86 laptop space. It stands on its own. A little late to the market, but not as bad as you think.

reggie_fils_aime · Apr 11, 2024

SiliconFly said:
I don't think you do. MTL is far ahead than previous gen in terms of battery usage. Far far ahead. But not much performance gains though. And it doesn't have significant competition in the x86 laptop space. It stands on its own. A little late to the market, but not as bad as you think.

Okay, so that's one facet of the computer which has improved. Everything else is... stagnant? A regression? Much like Bulldozer and Netburst?

quick edit: And if it were so far ahead of previous generations, why does it require cherry picked, byzantine professional review heuristics to show that? If it's that good, it should be obvious - and it would be marketed as such. For now, we're seeing how difficult it is to sell a new thing that's oftentimes slower than the old thing (unless you look at it from one VERY particular angle).

SiliconFly · Apr 11, 2024

Saylick said:
WAIT, so you're actually saying this Prakhar guy actually has legit info??

When a person makes 100s of posts, some of them will tend to be right. Even a a broken clock is right twice a day.

SiliconFly · Apr 11, 2024

reggie_fils_aime said:
Okay, so that's one facet of the computer which has improved. Everything else is... stagnant? A regression? Much like Bulldozer and Netburst?

quick edit: And if it were so far ahead of previous generations, why does it require cherry picked, byzantine professional review heuristics to show that? If it's that good, it should be obvious - and it would be marketed as such. For now, we're seeing how difficult it is to sell a new thing that's oftentimes slower than the old thing (unless you look at it from one VERY particular angle).

True. It was launched late by more than 2 months when compared to their usual schedule. Then, it's been 4 months already since launch and only now we're seeing the products hit shelves in limited numbers. Initially there were too many driver/bios/pcode issues (which I believe have been fixed now). MTL is the first ultra-complex SoC Intel has released for client ever! Tiles, Foveros, New Node, tGPU, NPU, etc. So, it's taking it's own sweet time to ramp up and reach market.

Now that it's mature, the results have started to trickle in. And almost all the newer (battery life) results are looking very promising. It's doing a lot better than previous gen. Performance wise not much improvement over previous gen, but excellent efficiency. No doubts there.

SpudLobby · Apr 11, 2024

dullard said:
Only partly. Intel has high density libraries and high performance libraries. Sometimes a third library. These are optimized for different types of uses for different chips (optimized for performance, optimized for cost, optimized for power, etc). M3 vs Lunar Lake would only give a comparison of one of those libraries.

Lunar Lake is explicitly targeted at low power market — ultra-low power, even, and we heard forever about how finally Intel would have some node parity or an advantage with this and then show Apple, QC etc that obviously Intel design and engineering is on top when they “try” — Intel was mostly failed by manufacturing. This is paraphrased but essentially has been the line.

What Intel does with that and their die area on N3B is on them. It’s not literally an academically perfect iso-node comparison, correct in that sense, but directionally it’s great and if Intel messes up the implementation because they don’t have the IPC to target denser cores or prioritized something different, that’s on them and an indictment.

SiliconFly said:
Both M3 & LNL are SoC. Too many variables which makes direct comparison impossible. Isometric or otherwise.

This is ultimately preemptive obfuscation.

Intel (AMD too but they’re usually more fair honestly) fans loved talking up how they’d achieve node parity with TSMC (or 18A if not better) and win with architectural advantages they supposedly have, but the closer we get and in 2024 I think we all know why some are antsy about this argument now, lol.

Even the leaked documents basically point to the iGPU matching the 2020 N5 vanilla M1 iGPU iso-power (at 12W), boosting higher than M2 GPU peak to match it’s performance.

That’s actually fine, the CPU is more important, but it gives us an indication of where they’re actually at. And this is a 140mm^2 die on N3B lmfao.

I don’t want to hear it about how Intel didn’t have xyz or needed more time or die $ or whatever when this thing is just B- to B+.

SiliconFly · Apr 11, 2024

SpudLobby said:
This is ultimately preemptive obfuscation.

Intel (AMD too but they’re usually more fair honestly) fans loved talking up how they’d achieve node parity with TSMC (or 18A if not better) and win with architectural advantages they supposedly have, but the closer we get and in 2024 I think we all know why some are antsy about this argument now, lol.

You can't compare nodes based on their products. Thats foundry 101.

You can compare nodes based on their products only if both are exact same (even the libraries have to be very similar). Otherwise it makes no sense.

Hitman928 · Apr 11, 2024

reggie_fils_aime said:
Ah, I understand perfectly now. We'll never see a straight comparison review for MTL because it gets washed by years-old chips in significantly cheaper machines. Nobody in their right mind (especially those in charge of procurements) would choose MTL over, well, anything.

This isn't quite Bulldozer-levels of underwhelming, but it's damn close to Netburst in terms of reality-bending equivocation from professional outlets.

It's not a bad chip and should not be compared to Bulldozer or Netburst. It is an improvement over Raptorlake in most of the ways that count. It is a regression in max performance (but Raptorlake blows out power to get there) and mem latency but is an improvement in efficiency and in performance for most of the usable (for a laptop) power curve. Battery life is more dependent on the entire system but MTL does bring some improvement there as well, though it really depends on the configuration. The problem is, the improvement wasn't all that significant overall and the competitive landscape around Intel has changed from 5 - 10 years ago. This leaves MTL as being an improved, but overall somewhat mediocre product when compared to the competition. Still a good product, but not able to take back their seat as the performance/technological leader.

Now, Zen 5 and Qualcomm's chips are right around the corner and it seems like the competition is only getting more fierce. Will Intel's next gen be able to compete blow for blow? I don't know, but MTL was supposed to be their return to grace but it didn't really live up to expectations. We'll see how well their next gen can do.

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

Senior member

Attachments

Lifer

Platinum Member

Lifer

Senior member

Diamond Member

Diamond Member

Diamond Member

Senior member

Diamond Member

Diamond Member

Member

Elite Member

Senior member

Diamond Member

Golden Member

Member

Golden Member

Diamond Member

Golden Member

Member

Golden Member

Golden Member

Senior member

Golden Member

Diamond Member