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.)

 

[Abwx]

Literally on the leak you were quoting, it says Granite Rapids AP, targeting ~2X performance of EMR-64C.
And also I'm pretty sure, at the very least, 2X SPR is enough to beat Genoa.
SPR 56C scores ~65K on CBR23 MT while Genoa scores 115K apparently.
In GB5, 2X the 56C SPR would be 108% of top end Genoa as well.
These are the only MT scores I can find comparing the two. If you can find some that show otherwise, I would be happy to hear them....

At wich TDPs thoses expectations..?.

 

[msj10]

At wich TDPs thoses expectations..?.

500W but these leaked documents are old from early 2021 before the node of GNR was upgraded from Intel 4 to Intel 3

 

[Geddagod]

500W but these leaked documents are old from early 2021 before the node of GNR was upgraded from Intel 4 to Intel 3

This might be more a limit of the platform rather than the node used.
I think GNR is supposed to stay on the same platform, even after redefinition, so I think it might still be limited to 500W. My Speculation though.

 

[Exist50]

How old does a forum poster need to be before you understand this endless quarrel is completely detrimental to the health of the forum? I'm tired of the same noise every day, I read less and interact less because people actively choose to inject their opinions everywhere between the few actual good leaks we get on new products. For me the highlight of the Intel threads was those Granite Rapids and Sierra Forest slides, I finally got to read things without constant interjections from overzealous forum members.

Constant arguments drive the typical users away, and you will be left with partizans willing to stomach the trolling.

It's simple. He will not be happy until he drives everyone away that doesn't hate Intel as much as he does, and he doesn't care how many forum rules he needs to break to do it. Nor how many lies he needs to tell.

 

[nicalandia]

Explain that estimation please?

Agregate performance is IPC + Speed regardless of core count and uArch. The x2 is exactly 100% performance boost when compared to Emerald Rapids 64C. Being 120 cores gives about 13% boost per core and that is the combination of IPC and Speed. Beign optimistic thats about 10% IPC Boost and 5% Speed

 

[H433x0n]

Constant arguments drive the typical users away, and you will be left with partizans willing to stomach the trolling.

As somebody new to this forum, I completely agree. In my view, this place is supposed to be a more technical and grown up version of r/Hardware. It would be a shame for it to devolve to just partisan bickering.

 

[Exist50]

Agregate performance is IPC + Speed regardless of core count and uArch. The x2 is exactly 100% performance boost when compared to Emerald Rapids 64C. Being 120 cores gives about 13% boost per core and that is the combination of IPC and Speed. Beign optimistic thats about 10% IPC Boost and 5% Speed

With a node shrink, speed should see >5% improvements at a server operating point. It's hard to even speculate with neither Intel 4/3 data nor EMR data, but only 5% would be a failure. Especially since they're unlikely to get 10% IPC.

And I think "aggregate performance" is just peak throughput, which basically scales with cores, clocks, and memory bandwidth.

 

[Geddagod]

Agregate performance is IPC + Speed regardless of core count and uArch. The x2 is exactly 100% performance boost when compared to Emerald Rapids 64C. Being 120 cores gives about 13% boost per core and that is the combination of IPC and Speed. Beign optimistic thats about 10% IPC Boost and 5% Speed

I used the +40% performance claim between GNR SP and EMR for my math:
SPR/EMR and GNR SP both have 350W TDP no?
~40% better aggregate performance. Nice.
80/64 = 25%, the remaining ~10% performance looks to be coming from better clocks+IPC. Impressive, considering GNR SP still has more cores and yet clocks better as well compared to EMR.
RWC IPC gains should be very low, I'm guessing most of the gains come from clocks.
In comparison, going from 84 cores vs 64 cores on the same node looks to be a ~10% loss in clocks (for Genoa and similar TDP). Same with going from 60 cores vs 48 cores for SPR (25% gain in cores just like 80/64 cores of GNR-AP vs EMR).
So the gain in frequency/watt from RWC is ~20%?
Hmmm sounds like that slide from Adored TV
(insert slide showing 20% better core performance from power-perf improvements)
I don't think IPC gain will be that large for RWC, which was the core they said they were using in that 2X performance slide.
I mean I would love to be wrong, since that means better products for RWC products, but I just don't think it's happening....
Unless the way better bandwidth per core for memory and also prob L3 could boost it by that much, (I don't think it will be a whole 10% though), but that also seems to be unique to RWC on server vs GLC on server, not RWC client vs GLC client.

 

[nicalandia]

But I mean granite rapids does not seem to compete with Genoa on any level. Or I am misreading this leak ??

Only the AP Variant(I don't remember but I read that the AP variant have been cancel or perhaps its Sierra Forest AP have been cancelled), The AP variant of Granite Rapids will have 120 Cores with about 15% aggregate performance per core and 12 Channel of DDR5, So in Paper that's quite competitive... But the SP variant will be restricted to 80 Cores 8 Channel DDR5 and that is not competitive of curse

 

[ondma]

It's simple. He will not be happy until he drives everyone away that doesn't hate Intel as much as he does, and he doesn't care how many forum rules he needs to break to do it. Nor how many lies he needs to tell.

Yea, I dont understand such intense hate for a product or corporation. Is it really worth the energy? I simply put him on ignore.

 

[nicalandia]

I don't think IPC gain will be that large for RWC

So MTL would be a side grade at best when you have true Raptor Cove Cores on a Laptop?

 

[Exist50]

So MTL would be a side grade at best when you have true Raptor Cove Cores on a Laptop?

MTL will probably benefit from better graphics, accelerators, power efficiency, and MT performance (i.e. power-limited performance). But peak ST to peak ST, seems like it's going to wind up very close to RPL mobile. Hence why it's not showing up on the desktop.

 

[mderbarimdiger]

And why is it that you felt the need to post this ? I don't believe that a FUTURE chip is what they are saying. That is ON TOPIC. You are a relatively new poster and you felt the need to insult my intelligence with this post ?

I'm sorry if that came across as insulting your intelligence, Discussing Genoa VS SPR, or analysis of benchmarks is not something that belongs for dumb folks, and i dont doubt your knowledge, hands on experience of tuning servers, or your understanding of energy consumption and performance analysis, etc.

However, "the reason i felt the need to post this" is because i felt:
1. you are not in sync with the conversation here.
2. you are way to much attached to the facts that "Intel sucks, intel sucks, genoa is super, others are biased towards intel, intel sucks, i am not biased, i am based on fact, i am not for example claiming that genoa is 5X better than SPR, intel sucks, i need facts!"

I do agree that Genoa is significantly ahead of SPR, and i agree that many people are biased, and i agree that intel servers are in bad shape now, yet i still feel disturbed when those sentiments are somehow finding its place wherever the word Intel is mentioned.

to me it feels like reading posts in a TV screen thread "you guys should better buy Samsung curved TVs, they are the best! the fact is that earth is not flat, those flat earth believers are living in denial, i love Samsung curved OLED TVs, earth is not flat, flat earth believers are hurting"

wouldn't it be disturbing? when reading TV screen discussions and someone is constantly repeating the "truth" about earths shape seemingly to indicate that flat TV screens is somehow related to flat earth is definitely annoying.

To be more specific, your initial comments indicated you have not been reading and/or understanding the leak, you claimed "this is not touching Genoa (which is out) what is intel thinking?" than saying you you have not seen any leaks about GNR, than arguing "the best they came up with is a target" "intel is hurting" "this is a fact!" "its only a claim" "lets stick close to reality" --- while All of those comments would be appropriate as a rebuttal if Intel PR would claim that they are providing superior servers vs AMD servers, they are totally out of place in the conversation yesterday.

The leaked slide is a internal note, written in 2021 purely as a target, purely for internal discussions, and yet you seemingly indicate that this is something intel PR department 'came up with'

And if someone is assessing that GNR have a chance to compete with genoa according to those leaked targets from 2021, you first deny his assessment, than you seemingly indicate that the speculator missed his own written word "targeting", or that he is denying the fact that intel server business is hurting, or that he miss to understand the difference between claims and facts, all the while he said from the beginning this is a target, all the while no one here was arguing intel server business is in great shape, all the while intel did not came up with any claims here, all the while recommending to buy flat TV screens is not related to flat earth believers.

listen to a note from someone with good intention: the frequency of repeating the "intel hurts, Xeon sucks power, genoa is super, facts, biased" slogan is reaching the level of annoying, please dont assume that whoever is saying the slightest positive thing about intel is by default living in denial or is having intention and is getting paid from intel, thats not the case.

maybe my sense is wrong, i might have misunderstood your expressions,

 

[mderbarimdiger]

you might just disagree and ignore my post.

sorry, i am having issues to connect my last line to the previous note

 

[Markfw]

I'm sorry if that came across as insulting your intelligence, Discussing Genoa VS SPR, or analysis of benchmarks is not something that belongs for dumb folks, and i dont doubt your knowledge, hands on experience of tuning servers, or your understanding of energy consumption and performance analysis, etc.

However, "the reason i felt the need to post this" is because i felt:
1. you are not in sync with the conversation here.
2. you are way to much attached to the facts that "Intel sucks, intel sucks, genoa is super, others are biased towards intel, intel sucks, i am not biased, i am based on fact, i am not for example claiming that genoa is 5X better than SPR, intel sucks, i need facts!"

I do agree that Genoa is significantly ahead of SPR, and i agree that many people are biased, and i agree that intel servers are in bad shape now, yet i still feel disturbed when those sentiments are somehow finding its place wherever the word Intel is mentioned.

to me it feels like reading posts in a TV screen thread "you guys should better buy Samsung curved TVs, they are the best! the fact is that earth is not flat, those flat earth believers are living in denial, i love Samsung curved OLED TVs, earth is not flat, flat earth believers are hurting"

wouldn't it be disturbing? when reading TV screen discussions and someone is constantly repeating the "truth" about earths shape seemingly to indicate that flat TV screens is somehow related to flat earth is definitely annoying.

To be more specific, your initial comments indicated you have not been reading and/or understanding the leak, you claimed "this is not touching Genoa (which is out) what is intel thinking?" than saying you you have not seen any leaks about GNR, than arguing "the best they came up with is a target" "intel is hurting" "this is a fact!" "its only a claim" "lets stick close to reality" --- while All of those comments would be appropriate as a rebuttal if Intel PR would claim that they are providing superior servers vs AMD servers, they are totally out of place in the conversation yesterday.

The leaked slide is a internal note, written in 2021 purely as a target, purely for internal discussions, and yet you seemingly indicate that this is something intel PR department 'came up with'

And if someone is assessing that GNR have a chance to compete with genoa according to those leaked targets from 2021, you first deny his assessment, than you seemingly indicate that the speculator missed his own written word "targeting", or that he is denying the fact that intel server business is hurting, or that he miss to understand the difference between claims and facts, all the while he said from the beginning this is a target, all the while no one here was arguing intel server business is in great shape, all the while intel did not came up with any claims here, all the while recommending to buy flat TV screens is not related to flat earth believers.

listen to a note from someone with good intention: the frequency of repeating the "intel hurts, Xeon sucks power, genoa is super, facts, biased" slogan is reaching the level of annoying, please dont assume that whoever is saying the slightest positive thing about intel is by default living in denial or is having intention and is getting paid from intel, thats not the case.

maybe my sense is wrong, i might have misunderstood your expressions,

First, thanks for explaining. And for the rest here, I will answer this and move on.

I really hate Intel PR, like saying "AMD is in the rear view mirror, and we won't see them again" is just insulting to everyone, even those that still think Intel is doing well. Also, I have worked in a data center, one that has square MILES of servers, and power is a big deal, but many here seem to ignore or minimize that fact. And I really am getting sick of hearing about new Intel servers, year after year of products that should not even be sold. The 13900k is faster than one of the new Intel HEDT chips ? is that not insane ?

And maybe lastly.... I am not biased, I WANT Intel to come up with a good product, otherwise without competition, things will be worse, not better. Its just that everyone keeps trying to pump up the current server products, many of which fail to have independent benchmarks (like serverthehome and such), and then I keep hearing about future products that are going to be great. Well how many years has Intel been saying "good things are coming" ?? fool me once shame on me, fool me twice shame on you (Intel). THAT is what I am sick of hearing.

One last comment. I became a moderator circa 2006 right after Conroe came out, and I was a HUGE fan, and was overclocking them like no tomorrow. I was ELECTED by the users here, probably largely due to that. And I was an Intel supporter until 2017 When Ryzen came along. People have a short memory, but I really like to own the best, and I have swung back and forth like a see-saw for years.

 

[Exist50]

And maybe lastly.... I am not biased

You call people trolls for referencing data that disagrees with whatever you happen to be lying about, come into threads like this just to insult everyone, and openly and repeatedly falsify claims about Intel. The vast majority of your comments in this Intel thread alone are just saying how much better AMD is and you believe will be.

No reasonable person could view your history on this forum and consider you unbiased. Your definition of a "good product" seems to constantly shift to whatever AMD does differently, without any regard to consistency or objective merits.

Seriously, is this a joke that's going over my head?

 

[reqq]

Hi, can someone smart speculate how good the ram latency will be on arrow lake compared to raptor lake. They say arrow lake will support 6400 mhz ram, which is higher then raptor lake. But all this new tile design makes me wonder if the ram latency advantage intel has will remain?

 

[esquared]

This thread is locked for awhile.

Get back on topic or infractions will be given.


esquared
Anandtech Forum Director.

 

[esquared]

Ok folks, keep on topic.

 

[A///]

imo meteor lake this fall for mobile because it's cancelled for desktop. raptor refresh this summer for desktop delivering more cache and whatever crap intel has devised, arrow lake release in q1 2024, lunar lake a mobile only platform in q3 or q4 2024, panther lake in q4 2024 or q1 2025.

it wouldn't be the first time intel released 2 major desktop generations in a 365 day period. to quote the general theme being said by many is intel is plowing through the generations in volume or not in volume to show shareholders they mean business. realistically the take rate for mtl will be high but not high for arrow lake since it'll be a first gen chiplet product from intel in its most modern sense. 12th gen was a new design approach and problematic whether people want to admit it or not. 13th gen does not have those same issues.

it doesn't help that windows 11 is junk, but both gens display normal behavior under windows 10 enviros even if they're not 100pc efficiently using their core context switching capabilities due to the scheduler.

 

[DrMrLordX]

It would be pretty surprising if Intel managed a desktop Arrow Lake launch in Q1 2024.

 

[A///]

It would be pretty surprising if Intel managed a desktop Arrow Lake launch in Q1 2024.

the smack talk online is pointing towards to them being on schedule or slightly ahead. whether that's bull will be realized later but if they can manage two major launches in a year like they have in the past I would be surprised too.

I'm not sure how it'll work with mtl-s taken out of the mix and whether or not arrow and panther are on the same platform socket or not. arrow is supposed to be impresive but it's a 1st gen chiplet product from intel. Panther Lake will remedy those issues if any and deliver a new core design for both p and e. there was a nova lake on an old slide from 2020 or 2021 but nothing since then. Nova would land in 2025 or 2026 and I suspect it'll be much a ddr4 and ddr5 split window period. ddr6 is supposed to come out in 2026 according to samsung.

 

[mikk]

imo meteor lake this fall for mobile because it's cancelled for desktop. raptor refresh this summer for desktop delivering more cache and whatever crap intel has devised, arrow lake release in q1 2024, lunar lake a mobile only platform in q3 or q4 2024, panther lake in q4 2024 or q1 2025.

it wouldn't be the first time intel released 2 major desktop generations in a 365 day period. to quote the general theme being said by many is intel is plowing through the generations in volume or not in volume to show shareholders they mean business. realistically the take rate for mtl will be high but not high for arrow lake since it'll be a first gen chiplet product from intel in its most modern sense. 12th gen was a new design approach and problematic whether people want to admit it or not. 13th gen does not have those same issues.

it doesn't help that windows 11 is junk, but both gens display normal behavior under windows 10 enviros even if they're not 100pc efficiently using their core context switching capabilities due to the scheduler.

MTL-S 6-16 seems to be cancelled, it's not clear what happens with the 6+8 tile which they need for mobile anyways. There are some rumors that MTL-S 6+8 and lower will be used for i3 and i5 SKUs whereas ARL-S gets the i7 and i9 spot with its 8+16 tile. ARL-S early 2024 in what process? 20A no chance and even TSMC 3nm looks slim at the moment for an early 2024 approach. They wouldn't even need Raptor refresh if ARL-S is coming early 2024. Most realistic target is more likely mid 2024. Panther Lake too early, it comes with a Xe3 GPU tile. This can't be ready till late 2025. Lunar Lake might come earlier than most expected, maybe before ARL-P. It's production ready in 2024 says Intel, 18A makes no sense unless Intel once again wants to delay a ready CPU because of waiting for 18A.

 

[DrMrLordX]

the smack talk online is pointing towards to them being on schedule or slightly ahead. whether that's bull will be realized later but if they can manage two major launches in a year like they have in the past I would be surprised too.

I'm not sure how it'll work with mtl-s taken out of the mix and whether or not arrow and panther are on the same platform socket or not. arrow is supposed to be impresive but it's a 1st gen chiplet product from intel. Panther Lake will remedy those issues if any and deliver a new core design for both p and e. there was a nova lake on an old slide from 2020 or 2021 but nothing since then. Nova would land in 2025 or 2026 and I suspect it'll be much a ddr4 and ddr5 split window period. ddr6 is supposed to come out in 2026 according to samsung.

The only way Arrow Lake happens that quickly is if Intel has been porting it to N3/N3e the entire time as a failsafe in case their own nodes aren't ready. Intel 4 still isn't on the market, does anyone believe that Intel will have Intel 3 or 20a or whatever available by Q1 2024? I certainly don't.

 

[A///]

The only way Arrow Lake happens that quickly is if Intel has been porting it to N3/N3e the entire time as a failsafe in case their own nodes aren't ready. Intel 4 still isn't on the market, does anyone believe that Intel will have Intel 3 or 20a or whatever available by Q1 2024? I certainly don't.

GR has been sampling to key customers for a while now.Or do you mean general availability to anyone and all DC?