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

Page 83 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Tigerick

Senior member
Apr 1, 2022
696
602
106
PPT1.jpg
PPT2.jpg
PPT3.jpg



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

ModelCode-NameDateTDPNodeTilesMain TileCPULP E-CoreLLCGPUXe-cores
Core Ultra 100UMeteor LakeQ4 202315 - 57 WIntel 4 + N5 + N64tCPU2P + 8E212 MBIntel Graphics4
?Lunar LakeQ4 202417 - 30 WN3B + N62CPU + GPU & IMC4P + 4E08 MBArc8
?Panther LakeQ1 2026 ??Intel 18A + N3E3CPU + MC4P + 8E4?Arc12



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

Meteor LakeArrow Lake (20A)Arrow Lake (N3B)Lunar LakePanther Lake
PlatformMobile H/U OnlyDesktop OnlyDesktop & Mobile H&HXMobile U OnlyMobile H
Process NodeIntel 4Intel 20ATSMC N3BTSMC N3BIntel 18A
DateQ4 2023Q1 2025 ?Desktop-Q4-2024
H&HX-Q1-2025
Q4 2024Q1 2026 ?
Full Die6P + 8P6P + 8E ?8P + 16E4P + 4E4P + 8E
LLC24 MB24 MB ?36 MB ?12 MB?
tCPU66.48
tGPU44.45
SoC96.77
IOE44.45
Total252.15

LNL-MX.png

Intel Core Ultra 100 - Meteor Lake

INTEL-CORE-100-ULTRA-METEOR-LAKE-OFFCIAL-SLIDE-2.jpg

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.)



Clockspeed.png
 

Attachments

  • PantherLake.png
    PantherLake.png
    283.5 KB · Views: 24,006
  • LNL.png
    LNL.png
    881.8 KB · Views: 25,490
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
22,065
11,693
136
I don't think tsmc would bump off amd.

I don't either, but anyone who becomes too dependent on one foundry partner to rule them all and in the darkness bind them is mmm well let's just say that monopolies in Taiwan are unstable and leave it at that.

Yes it's a potential problem for Intel too.

Samsung saved AMD once (GF14nm), but I don't think they can do it again. Would be interesting to see what would happen if Samsung made an alliance with Intel, but honestly if that was gonna work, Gelsinger would have had to do it as soon as he took office for it to matter.
 
  • Like
Reactions: A///

NostaSeronx

Diamond Member
Sep 18, 2011
3,706
1,233
136
Samsung saved AMD once (GF14nm), but I don't think they can do it again.
Samsung didn't save AMD. AMD was already working on Zen on TSMC's 16nm, since 2014. 14nm was suppose to save GlobalFoundries, but ran a huge debt across the board. It then caused AMD to delay from a 1H2016 Zen(16nm node) general availability to a 1H2017 Zen(14nm node) general availability. Doing irreversible harm to revenue till AMD finally returned to Zen's original home foundry TSMC for 7nm.

Getting forced in 14nm by GF via WSA. Lead them to also going into a sub-par node which cut LV and HV performance. Leading to lower than ideal frequency in both low-voltage and restrictive scaling for high-voltage. TSMC's 16nm was much better for scaling high-speed across low voltage/low-TDP and high voltage/high-TDP.

The issue Intel is hitting right now is loosely similar to Samsung's 14nm and the Zen-port on to it. The process is tuned for low-power-orientated designs but Intel is pushing high-speed-orientated designs on them.

Where as TSMC has knobs for tuning the process from a low-power-orientated one into a high-speed-orientated process. Intel is simply not doing the work for these knobs to actually be exploited on their node at architecture release. GlobalFoundries excuse was they didn't have any additional money or know-how to fix their stuff.
 
Last edited:

Hulk

Diamond Member
Oct 9, 1999
4,525
2,518
136
TMSC has all of the x86 information from AMD and now Intel required to manufacture CPU's. When are they going to realize they can just design (reverse engineer) their own CPU's and rule the semiconductor world?
 
  • Haha
Reactions: Thunder 57

mikk

Diamond Member
May 15, 2012
4,247
2,311
136
"I'll believe it when I see it". They may WANT to speed up their processes, but based on the immediate past it's hard to believe that they'll reach those targets. Intel has been blowing this smoke for awhile, but what are the results? Best-case scenario, we see market readiness for Intel 4 in late 2023, Intel 3 in late 2024, and Intel 20a in late 2025. Best case.

Realstically-speaking, Intel probably couldn't get 20a ready before 2027. At least based on 14nm and 10nm-family nodes and their rates of progression, along with Intel 4 which is going to be three years late (was supposed to debut in Ponte Vecchio some time ago).


This is not the best case lol. You are way off. Best case would be H1 2024 Intel 3, H2 2024 Intel 20A and H2 2025 Intel 18A. A more realistic timeline would have been 2H 2024 Intel 3, H1 2025 Intel 20A and H2 2025 - H1 2026 Intel 18A. ARL-P could come with Intel 20A but realistically not before H1 2025. ARL-S is only doable in 2024 if they switch to TSCM which some rumors suggest.
 

bwhitty

Junior Member
Oct 21, 2021
5
7
51

A///

Diamond Member
Feb 24, 2017
4,351
3,158
136
straight from wiki. either you got your dates wrong or you're talking about something else.


Cannon Lake was initially expected to be released in 2015[6]/2016, but the release was pushed back to 2018.[7] Intel demonstrated a laptop with an unknown Cannon Lake CPU at CES 2017[8][9] and announced that Cannon Lake based products would be available in 2018 at the earliest.

At CES 2018 Intel announced that it had started shipping mobile Cannon Lake CPUs at the end of 2017 and would ramp up production in 2018.[10][11][12]

On April 26, 2018 in its report on first-quarter 2018 financial results, Intel stated it was currently shipping low-volume 10 nm product and expects 10 nm volume production to shift to 2019.[13] In July 2018, Intel announced that volume production of Cannon Lake would be delayed yet again, to late Q2 2019.[14]
 

DrMrLordX

Lifer
Apr 27, 2000
22,065
11,693
136
Wait you are saying intel 3 is in the same shape as 10nm back in 2016.

I'm saying Intel 4 is in almost the same shape as 10nm back in 2016. Intel 3 is a derivative node. It took Intel maybe two years to get 10nm+ out the door in commercial products. Even if Intel 4 shows up in good order by the end of 2023 in Meteor Lake, it's completely unreasonable to expect Intel 3 to show up less than 6 months later in viable quantities. Unless we believe "it's not the process it's the design" is the reason for every Intel 4 product to be scotched up to this point, which seems ridiculous.

Maybe Intel can set wafers on fire for huge losses and push out a few Sierra Forest units to key customers by March 2024 (end of Q1). Anything more than that? Nah, not buying it.
 
  • Like
Reactions: moinmoin

DrMrLordX

Lifer
Apr 27, 2000
22,065
11,693
136
I'm saying Intel 4 is in almost the same shape as 10nm back in 2016. Intel 3 is a derivative node. It took Intel maybe two years to get 10nm+ out the door in commercial products. Even if Intel 4 shows up in good order by the end of 2023 in Meteor Lake, it's completely unreasonable to expect Intel 3 to show up less than 6 months later in viable quantities. Unless we believe "it's not the process it's the design" is the reason for every Intel 4 product to be scotched up to this point, which seems ridiculous.

Maybe Intel can set wafers on fire for huge losses and push out a few Sierra Forest units to key customers by March 2024 (end of Q1). Anything more than that? Nah, not buying it.

@A///

You're splitting hairs. You've been registered here long enough to know what BK was doing in 2017 with Cannonlake to keep investors off his back. "Cannonlake when?" achieved meme-tier status back then. It was really supposed to be a 2016 product but got pushed back to "it's done when it's done", then got launched in some Chinese laptops in 2c i3 format with disabled iGPU in Q4 2017 and was dead-and-buried. 2018 was meant to be the general date range for volume shipping which Intel never achieved.

BK limped it out the door in Q4 2017 to hit his ship target and keep investors at bay.
 
  • Like
Reactions: Ajay

techinvestor1

Junior Member
Apr 3, 2023
2
1
41
I'm saying Intel 4 is in almost the same shape as 10nm back in 2016. Intel 3 is a derivative node. It took Intel maybe two years to get 10nm+ out the door in commercial products. Even if Intel 4 shows up in good order by the end of 2023 in Meteor Lake, it's completely unreasonable to expect Intel 3 to show up less than 6 months later in viable quantities. Unless we believe "it's not the process it's the design" is the reason for every Intel 4 product to be scotched up to this point, which seems ridiculous.

Maybe Intel can set wafers on fire for huge losses and push out a few Sierra Forest units to key customers by March 2024 (end of Q1). Anything more than that? Nah, not buying it.
As far as I understood, each of these nodes is done in parallel to avoid the issues of the past. Much more likely to scrap one and just skip to the next node if there are issues as it is separate teams now as far as I understand it.

The problem with AMD and TSMC's relationship is that without real competition (Samsung is NOT reliable enough to be competition, ask Apple how come they switched), TSMC will squeeze AMD's margins because it's common sense/great business. So while TSMC is superior to Intel in manufacturing, that is not an infinite "moat" for TSMC. Intel can always have about 20-30% worse yields than TSMC and still be competitive in prices with AMD, because that is TSMC's current margin.

Intel's mistake was underinvesting, something it's clearly fixing now. I am amazed at how many people here act as Intel's current operations are "business as usual" (same as the last 10 years). In the last 3 years Intel has spent $60b on new nodes, which is the same amount spent over the 5 years before that! And before you say "TSMC spent $100b" or similar, keep in mind that getting new nodes online is not a "more is better" ad infinitum. TSMC manufactures much more quantity so needs more capacity, but getting new process technology is ~$20b and Intel is investing everything needed to make its roadmap.
 
  • Like
Reactions: controlflow

DrMrLordX

Lifer
Apr 27, 2000
22,065
11,693
136
As far as I understood, each of these nodes is done in parallel to avoid the issues of the past. Much more likely to scrap one and just skip to the next node if there are issues as it is separate teams now as far as I understand it.

I could see 20a being developed in parallel. Intel 4 and Intel 3? Probably not, since Intel 3 is an iterative step.
 
  • Like
Reactions: Lodix

coercitiv

Diamond Member
Jan 24, 2014
6,677
14,274
136
As far as I understood, each of these nodes is done in parallel to avoid the issues of the past. Much more likely to scrap one and just skip to the next node if there are issues as it is separate teams now as far as I understand it.
We've been over this before in the forums: volume production in a node is an important step in preparing a healthy transition to the next node, field data from volume production is needed for R&D. Scrapping a node and skipping to the next shrink is risky.
 
  • Like
Reactions: moinmoin

jpiniero

Lifer
Oct 1, 2010
15,223
5,768
136
We've been over this before in the forums: volume production in a node is an important step in preparing a healthy transition to the next node, field data from volume production is needed for R&D. Scrapping a node and skipping to the next shrink is risky.

Intel 3 isn't really a new node. It's more like 7+.
 

techinvestor1

Junior Member
Apr 3, 2023
2
1
41
Intel 3 isn't really a new node. It's more like 7+.
Compare it on density and metrics.

We've been over this before in the forums: volume production in a node is an important step in preparing a healthy transition to the next node, field data from volume production is needed for R&D. Scrapping a node and skipping to the next shrink is risky.
Sure, but that doesn't factor in much on the first year of running it. Volume helps perfect the yield for sure.

I could see 20a being developed in parallel. Intel 4 and Intel 3? Probably not, since Intel 3 is an iterative step.
I am not familiar enough with the process of developing the chips to be able to address this, but from how they presented it, it sounded like they aren't dependent on one another. I might be wrong on this.

The yields are going to be way way worse than 20-30% than TSMC. Meteor Lake and Sierra Forrest might be the guinea pigs as to whether the "f it, we'll launch it anyway!" strategy can be viable if the tiles are salvageable enough.
I don't think they'll have an issue footing the bill to stay competitive, especially if it's for external Foundry customers. Keep in mind the subsidies Intel will receive will probably end up at around $40b+, so they have a lot of short term leeway to catch up, poor yields or not.

The only big question is if they will have another big stumble on delivery, since they've eaten up all the margin of error they can with SPR. If they can delivery on 20A and 18A on time, they will do very well. If not, bleak future.
 

mikk

Diamond Member
May 15, 2012
4,247
2,311
136
I'm saying Intel 4 is in almost the same shape as 10nm back in 2016. Intel 3 is a derivative node. It took Intel maybe two years to get 10nm+ out the door in commercial products. Even if Intel 4 shows up in good order by the end of 2023 in Meteor Lake, it's completely unreasonable to expect Intel 3 to show up less than 6 months later in viable quantities. Unless we believe "it's not the process it's the design" is the reason for every Intel 4 product to be scotched up to this point, which seems ridiculous.

Maybe Intel can set wafers on fire for huge losses and push out a few Sierra Forest units to key customers by March 2024 (end of Q1). Anything more than that? Nah, not buying it.


Noboody with some deeper inside knowledge is saying this, we hear quite the opposite. MTL-P ES1 hits 3.6-4Ghz already. 10nm back in 2016 could only dream of this. It's just trash talk what you are posting.
 
Nov 8, 2022
43
77
51
No question that A18 is in great shape, so is I3, of course the recent intel presentation is a best case scenario, but its based on the facts on factory floor, the core design is possibly less clear than the transistor nodes, and QUALCOMM is a costumer of 18a, and they are the one who are sampling it with PDKs, it might be delayed with 2 quarters, but its definitely in great shape.

- Says everyone with honest inside knowledge.
 
  • Like
Reactions: techinvestor1

DrMrLordX

Lifer
Apr 27, 2000
22,065
11,693
136
Story time, because most of you all (I won't say ya'll or whatever despite being raised in a southeastern state) need a reality check. I was born into poverty. Back when Cannon Lake was introduced I was bankrupt. Oh and I almost died. Literally. The day after my bankruptcy proceeding I landed in the hospital in the ICU for unrelated causes. My family was called in to say goodbye. They were told I wasn't going to live. Family flew in from all across the US.

Today, I am in the top 10% of earners. I did not die and I am killing it every single year. I am also in the best physical shape I've been in since a teenager.

Never underestimate anyone. That includes any company as well. If past performance were indicative of the future, I should be long dead, buried, and my family, broke. Skepticism is good. Healthy. Allowing it to jade your views is not, especially when a company (Intel) still manages to be competitive with another (AMD) despite the latter company having access to superior technology.

Note that Apple was once very close to bankruptcy as well.

Glad you aren't dead or broke. Yeah, things can turn around, but it's a lot easier to turn around a single life than the Titanic. Intel has a lot of momentum, for good or for ill.
 
  • Like
Reactions: Mopetar