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.)

 

[coercitiv]

I don’t even recall AMD making a ton of proclamations about Zen 1 pre official release.

Here's a video from Aug 2016, AMD launched Zen 1 in Feb 2017

Consumers need not know much beforehand about how good the next gen Intel products are, but stockholders and investors should. Intel needs as much trust capital as they can borrow. When they have something good we will know, they will tell enough for analysts to raise an eyebrow.

 

[dullard]

Intel would look pretty foolish talking about process leadership and coming up with this elaborate plan, executing on it for years, and then turning around and saying “just kidding guys, IFS sucks, we are using TSMC”.

You would sooner convince me that AMD is using IFS.

Is this the hill you are going to fight for? Evidence for your side: you think that is the way it would be. Heck, your own argument fails on its own merits (TSMC already is producing some E cores on Meteor Lake).

Evidence that Intel will use TSMC:
1) From Intel:
https://www.intel.com/content/www/u...technology-roadmaps-milestones.html#gs.2tquli

Lunar Lake and Beyond – Fueled by its IDM 2.0 strategy, Intel will be using both internal and external process nodes to deliver leadership products.

2) From Intel: https://download.intel.com/newsroom/2022/corporate/2022-Intel-Investor-Meeting-Client.pdf

Combine that with a 2024 Lunar Lake launch and 18A not launching until 2025 makes a pretty strong case for External.

3) Supposedly leaked from Intel: https://www.tomshardware.com/news/intel-lunar-lake-mx-to-use-tsmc-n3b


4) From Intel (Gelsinger): https://www.anandtech.com/show/1657...-on-intel-ip-blocks-for-foundry-cores-on-tsmc

Intel’s x86 Designs No Longer Limited to Intel on Intel: IP Blocks for Foundry, Cores on TSMC

5) From Intel (Gelsinger): https://www.tomshardware.com/news/intel-pat-gelsinger-clients-want-custom-x86-socs

We are going to make selective use of foundries. […] If I need to use a foundry to have a unique product or a halo product and certain portions, absolutely.

I could go on, but they are just variations of the above.

 

[dullard]

ARL desktop is getting a massive node jump. From Intel 7 to 20A. Thats more than a 2X node jump. ARL is gonna be a lot more power efficient than the RPL generations. 50% to 60% efficiency gains just from the node jump alone clock-for-clock. Architectural efficiency gains are extra.

Seems like you didn't get the memo. When Intel is on an older, less efficient node, we are to ignore single thread performance and complain about power & efficiency. When Intel moves to a newer, more efficient node, we are to ignore power & efficiency and complain that it might not match previous single thread performance.

 

[mikk]

Intel's Lunar Lake will have multiple SKUs for different segments (based on power) it seems based on both the early and recent leaks. But what's really interesting is these clocks don't look great, and we know LNC is on N3B, uses memory on package and has some low power tech, and isn't going to be a humongous IPC upgrade given how long it's been. It will get them right in line with like Firestorm or the X3/4 at best on IPC and probably not.

The detected max frequency on this page is a pretty much meaningless value there. Have a look to Rapor Lake, Alder Lake, Arrow Lake-S, Tigerlake and so on, they are mostly below 3000 Mhz. Dit it mean they only boosted to 3 Ghz in real products? Certainly not. We don't know what it means, is it a PL1 or PL2 max speed and what power. Or is it even an average speed from P+E cores combined. Pretty much meaningless there. You read too much into it.

 

[DavidC1]

For "Process Leadership" there sure are a lot of "External" showing up.

The idea itself of using whatever process necessary is a good thing.

Let's look at what they themselves have outlined as a problem within their company:
-The process development team got arrogant and treated equipment vendors badly.
-The foundry service never worked out because the management saw it merely as an icing on the cake and potential customers were treated as 2nd class citizens, rather than equal.
-Many years of decisive process advantage meant their design team was wasteful with resources and time. The only team within Intel more arrogant than the process team was design. The old adage "crap CPUs made up by excellent process" was true.
-Semianalysis went into detail about such waste: https://www.semianalysis.com/p/rebuilding-intel-foundry-vs-idm-decades

Hot Lots -

On the other hand, a leading-edge wafer can also be processed in a few weeks if no care is given to cost, but this leads to tremendous inefficiency with tools constantly

They could do as many hot lots as they want, they can do as many samples as they want.

David Zinsner, Intel CFO

Now, they are going to charge their internal design team for the hot lots.

Steppings -

They can do as many steppings as they want.

David Zinsner, Intel CFO

It took Intel 12 steppings to bring Sapphire Rapids to market, whereas AMD usually takes only 2 to 3 steppings for competing chips such as Bergamo and Genoa.

We even heard rumors from former employees that there was a time when some Intel design teams would rather send a design to the fabs, get a hot lot sample back, and test it for bugs rather than complete more through simulations and verification.

Intel plans to fix this issue by reducing the number of samples and steppings by charging the design and product business units a fair price for these operations rather than allowing as many steppings as wanted.

Test, Sort, Bin -

Intel believes charging business units more directly for test, sort, and bin will make the business units more conscious about what test strategies they use within designs and lead to savings of ~$500M a year.

Design and Ramp -

Intel design teams also tend to ignore the reticle conundrum and lithography tool throughput issues for internal manufacturing, which is something TSMC charges customers for. For a deeper explanation, see here.

This essentially tells us that Intel's problems were widespread and it wasn't simply a matter of "replace the CEO" or "split the company". These were long-standing issues arising from a company that ironically touts "Only the Paranoid win" but not living up to that standard themselves, and were comfortable with their massive financial stockpile and believing they were invincible.

Look at Golden Cove, and even the suspected performance for Lion Cove. They are disappointing. Because they lost the process leadership, they are fully exposed. Rather than thinking Intel was a decent CPU designer with phenomenal process technology, their process leadership was masking growing internal problems just as if a mother would shield her adult son from all troubles, even if going through that trouble would mean her son would benefit from it.

They are basically telling Intel Foundry that you can't be just making chips for Intel anymore, and it has to do all things for all customers, and they are telling Intel design teams that we won't give you training wheels and have to stand on merit, on equal ground.

We should have cycles of TSMC-fabbed Intel chips prompting Intel Foundry to do better, and Intel-fabbed AMD and Nvidia designs prompting Intel's design to do better.

 

[dullard]

MLID is a very good source, most of his leaks have come true or very close to. The times he has been incorrect has been due biased sources

It would be interesting if anyone had the time to actually make a scoring sheet with track records of their past predictions for products out now.

 

[aigomorla]

@Abwx You are getting VERY close to inciting and flame baiting. I have to agree with other a few other posters that you are cherry picking and then trying to pass AMD propaganda, instead of technical. Do it again and I will exit you from future participation of this thread permanently.

@SiliconFly The only people that are allowed to tell someone to leave the thread is us moderators, do it again, and you will be given an infraction. Everyone is entitled to there opinion, don't like it, then put them on ignore. If we find they are distructive to chat, or inciting, we will force a exodus on them and ban them from future participation in the thread.

Moderator Aigo

 

[Geddagod]

Abwx is competent and smart, but stubborn as a mule and convinced that most other people are less than well prepared (to use an euphemism). Add a language barrier and the fact that most of the time we work with indirect data sources and napkin math, and we have a recipe for disaster in communication. Most of the time it ends up with people talking at each other with different premises and reasoning in mind. One such example is him expecting to derive GLC data from that Intel 7/Intel 4 v/f graph while you were just trying to discuss node properties. The resulting exchange is pointless, like most of the discussion on power efficiency in the other thread (or in this one, I can't tell anymore).

Moving away from this debate, one thing that stuck in my head after seeing that graph again last night, is the frequcny scaling for Intel 4. It's interesing that in the sub-4Ghz range the new node performs better than Intel 7 (as one would expect), and yet we know MTL is subject to lower fmax in the 5Ghz range than RPL. The two facts are not necessarily ot odds with eachother, but it makes me curious to see whether this ends up being the result of the new P core design (kinda unlikely), node frequency scaling in the higher region, or maybe even a deliberate choice by Intel to opt for density optimisation in their mobile lineup.

Lower Fmax vs RPL could be attributed to a combination of these factors:
lower fin counts (using HP instead of UHP)
Low volume so binning has to be more flexible
pretty likely DTCO improvements from RPC>GLC didn't track into RWC (considering MTL and RPL had pretty close to parallel development timelines)
Slight architecture changes (L1D edit: L1I mb in particular could make clocking high harder)
RPL uses Intel 7+ rather than Intel 7 (as shown in the graph)

In comparison to ADL in mobile, MTL actually has a 200MHz Fmax advantage.
A pretty common problem, IMO, for Intel, is that since they have been delaying their nodes for so long, the node behind it gets extremely mature, the architecture behind it gets insanely optimized, and the binning for it becomes extremely well tuned. Kinda handicaps how high they can say their "next gen" product improves upon the previous one in some aspects.

 

[H433x0n]

It's closer to 15-20%, which is cutting it really close. I think it will tie or maybe even slightly lose to Phoenix tbh. Also, in single core perf/watt (for RWC), I believe it will be worse there as well.
In iGPU, it appears to be better
In battery life, I mean Intel claims 2x, but that's highly dependent on how they test battery life. I expect a massive improvement, but there's a lot of ground to cover too.
And in ST peak performance, I expect them to be roughly tied.

This isn't an overall win over Phoenix, since they have some areas where they might lose (albeit slightly) or be tied in. What's even worse is how much later it's coming compared to Phoenix as well, and how much more expensive it looks to be too. At best one could say this is a slightly buffed Phoenix, depending on what metric you are comparing it for.

I'd take that outcome tbh. Phoenix is on N4P, arguably the best TSMC has to offer at the moment. It's also going against Zen 4, probably the best overall core design in the past 10 years. If it matches Phoenix efficiency (+ or - 5%) I would say that's the best position Intel has had in a while especially since they can actually provide volume and support.

 

[Hulk]

I don't have any of the old performance claim charts as they are immediately old news once reviews are out but I do have a lot of examples of how Intel used to give us actual information pre-launch. Here are some I have on hand.

 

[tamz_msc]

Geekbench 6 is calibrated to give 2500 score on a desktop 12700 in the ST tests.

Since it is performing essentially the same as a desktop 12700, considering the variability of Geekbench runs on a laptop, I would say it's performing as expected.

The MT score is really impressive, considering 8+4 for the 12700 vs 6+8 for the Meteor Lake chip, with the obvious caveat that it's not known what the LP Island E-cores do when the CPU is run at full tilt. Lower boost frequency of the E-cores in general may also be a factor.

 

[gdansk]

Intel Ultra 5 125H and Ryzen 7 7840HS tested at 65W: Intel with higher iGPU score but falls short in CPU tests - VideoCardz.com

Intel Core Ultra 5 125H at 65W shows lower CPU score and higher iGPU score than AMD Ryzen 7 7840HS Intel’s upcoming Core Ultra 100H series is just around the corner. Intel is going to show off its new laptop series called Meteor Lake in two days. This will introduce a new name for the […]

videocardz.com

Seems Intel is back in the game, at least for the iGPU.

One primary reason for getting a Radeon laptop just went whoosh!

My condolences, @adroc_thurston

View attachment 90294

If you scroll up a bit, he already posted this and says it means MTL is a flop.
In either case it's probably best to wait for testing at MTL's target power limits.
And if it's like other Intel GPUs it'll kill in benchmarks then flail around in actual games.

 

[CouncilorIrissa]

Intel's mature high clocking nodes really have a tendency to screw them over, don't they? It's what, third time when their new lineups lose in ST compared to predecessors (Haswell -> Broadwell, RPL -> MTL, Whiskey Lake -> Ice Lake)?

 

[maddie]

Me hyping? I think you are very confused. MTL has met all my expectations. I very clearly said MTL's performance will be similar to that of RPL (even a bit less in certain workloads) & it's efficiency will be a lot higher compared to RPL in very light loads due to LP E cores. Everything what I said has come true. Kindly read my words clearly before accusing me of hyping.

In your own words: With your AMD bias, I can tell you whats going to happen to your favourite brand starting December 14th.

Intel, for the last 5 years, have been releasing outdated products built on outdated nodes based on outdated and power-hungry core designs. Something that even I find a bit despicable. But in spite of all these severe shortcomings, they managed 88% desktop & laptop CPU market share consistently leaving AMD in dirt picking up scraps. And the credit goes to Intel marketing. They're one of the best in the industry.

Now Intel has given these sharks with an amazing product they've been waiting for all these years. A hyper power-efficient CPU thats far more advanced and a true engineering marvel compared to AMD's old and outdated chiplet design. Not only they're just going to simply outsell AMD like always, but this time it's going to a massacre and it's going to be very brutal.

AMD's time has come. Live with it. And I think it's time to start praying so that Intel has some decent competition in the future for our own good. 😧

Did you write this a few days ago?

Did you write this a few days ago?

 

[Hulk]

Such as?

Moving their mainstream chips from monolithic to tiled and using outside foundries. The biggest change in direction for them since their inception.

 

[controlflow]

Power comsumption with the OLED screen off is 8W, for the 7840U Pavillion it s about 6.5W on average, numbers are here, it s just a matter of spoting them...

So it appear that when idling MTL is not that efficient, its idle power is 1.5-2W higher than the Pavillion and the OLED screen has nothing to do with it.

Idle and light load power is probably the strongest feature of MTL.

There are some issues currently with FW and OS scheduling which result in sub-optimal use of the LP E cores but this test does show the potential in idle and low use situations. This system here has a 120hz OLED BTW.

 

[TESKATLIPOKA]

Looks like your post is written by some cheap and deceptive marketing department. Reeks of desperation about a failed product like 7840. (Recommend not to engage in personal attacks. As recommended by the mods, attack the post not the person who posts.)

7840 is a failed product? Can you stop posting lies? Thank you.

 

[Goop_reformed]

So is MTL good now after all the shenanigans ?

 

[TESKATLIPOKA]

This is not how it works, you have to test the efficiency or performance at the same power.

Who said you have to test It like you said?
You either test It at ISO performance, which @Abwx did, or ISO power like you did.
You want to use ISO power, because the difference is smaller.
@Abwx wants to use ISO performance because the difference is bigger.
You want to make MTL look better, he wants It to look worse, yet both of you are correct.

 

[TESKATLIPOKA]

That s my point, by using the same throughput we use the same time, that s the only way to compare intrinsical perf/Watt.

That's not perf/W but W/perf.

If you increase time you have flawed results, yes, but if you reduce time by increasing frequency you ll have flawed results as well, just in the other direction, hence a same throughput as basis wich mean a same time to perform a same workload.

The flawed result was not only because you increased/decreased the time(performance) needed to finish the job(work, task), but also because the power consumed was not the same. So of course the result was BS.

You can play with both ISO performance and ISO power comparisons, so I find It incomprehensible that you consider only comparing at ISO performance as the correct one, despite It having glaring flaws.

1.flaw) you can set a performance level to one CPU which is not achievable for the other one regardless of how much W It consumes.
2.flaw) If you want to measure efficiency in games, then there you clearly don't measure how fast you finish the game, but how much FPS the CPU can produce per W. So here you are really measuring perf/W and not W/perf.

Curious that you said that a longer time for RPL (compared to MTL) is flawed but didnt catch that a shorter time for the 7840U (compared to MTL) is also flawed...

Just because I didn't comment it doesn't mean I didn't see It or thought It was correct.
At that time I only replied to your first sentence, despite quoting your whole post, that's all.

 

[Markfw]

LoL now you're doing the same thing - comparing laptops with different panels - that you complained I was doing in my earlier post.

Your commentary reminds me of this quote from Shakespeare:



This forum is an echo-chamber full of Gratianos supporting a particular "team".

The fact that your particular "team" is not doing well at the moment and you keep defending them speaks volumes.

Please try to stay away from being a cheerleader.

 

[coercitiv]

I worry that you fall into that category.

You know what, you're right. I should be more open minded. I'll just open his video about Meteor Lake.

 

[Thibsie]

Clickbait video. ML may be disappointing, but it is hardly "Bulldozer".

Well, that's MLID isn't it ?
Well...

 

[ondma]

Intel slides during launch said it's NOC. Even anandtech says NOC. Many other say it's NOC. If you believe otherwise, it's perfectly okay. It's a free world and we all have our right to believe what we want. Just for you though...

Link1
Link2
Link3
Link4
Link5

Sometimes we just have to accept the fact that ARL is gonna be superior to Zen5 in some aspects.

I'll repeat it once again esp. for you though: As of now, only LNC appears to be the weak link in ARL. As far as packaging is concerned, ARL is far superior than Zen5.

Give it a rest with the packaging hype. Dont get me wrong, I would love to see ARL be very competitive or even beat Zen5. It is the final performance and power consumption that counts though. That is the sum total of all parts of the chip. So what if Foveros is superior to AMD's solution? If the core is not as good, that is like putting higher speed rated tires on a car with a less powerful engine. Yes, it may be better, but it is not the determining factor in final performance.

 

[FlameTail]

If Intel wants to beat Apple's power efficiency, they need to let go of their addiction to HP node libraries.