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

Page 622 - 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
700
616
106






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 + 4E012 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 (N3B)Lunar LakePanther Lake
PlatformMobile H/U OnlyDesktop & Mobile H&HXMobile U OnlyMobile H
Process NodeIntel 4TSMC N3BTSMC N3BIntel 18A
DateQ4 2023Desktop-Q4-2024
H&HX-Q1-2025
Q4 2024Q1 2026 ?
Full Die6P + 8P8P + 16E4P + 4E4P + 8E
LLC24 MB36 MB ?12 MB?
tCPU66.48
tGPU44.45
SoC96.77
IOE44.45
Total252.15



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.)



 

Attachments

  • PantherLake.png
    283.5 KB · Views: 24,009
  • LNL.png
    881.8 KB · Views: 25,496
Last edited:

DavidC1

Golden Member
Dec 29, 2023
1,086
1,739
96
I haven't seen any interesting Lunar Lake laptops that can compete with MBA. They're all too expensive in comparison. I guess I'll have to wait for discounts but it seems weird that now Apple is the king of $700-1000 laptops...
Some of us have been asking Lunarlake for years. Now the manufacturers have been wasting using OLED and 1440p screens, necessitating a ridiculously sized battery to do so.

-11.6-inch screen 2-in-1
-Max 2.2lbs
-Good FHD touchscreen with active digitizer.

$1000 is totally acceptable. Unlike the Core M chips it's super fast in single thread, and has great graphics performance. Oh, and for most of us that wants it, MT performance is really not bad at all.
 

itsmydamnation

Platinum Member
Feb 6, 2011
2,947
3,625
136
Can someone expand on which factors may contribute the 9950X and 285K having such similar IPC in 1T when it was shown Zen 5 operates as a 4-wide decode design outside of SMT and Lion Cove is apparently up to 8-wide decode?

Too many mispredictions? There isn't more ILP to exploit? Lion Cove is bottlenecked at execution? Then why does Lion Cove have 8 wide decode?
most workloads execute out of the OP cache for both , so more decode is more around worst case situations, so then you need to look at branch / fetch / decode cache etc performance as a whole.

i would really love to see intel or AMD do IBM style virtual L3 , so that way cores can have massive fast private L2's. If we look at the high IPC fat cores and the lower IPC fat cores, L2 cache is one thing that stands out. Looking at some early Zen5 x3d geekbench etc when every workload is faster with extra cache your probably struggling to feed the core.
 

H433x0n

Golden Member
Mar 15, 2023
1,222
1,599
96
Maxon s historical relation with Intel is documented, so why are you doing like it s a conspiration theory while it s a fact.?.

You know better than Maxon s senior software developper.?

Do you really expect Zen 5 to win 100% of the benchmarks otherwise there’s a conspiracy going on?

I watched the video and it was created in 2011. Do you remember the state of the x86 CPU market in 2011? It was a totally different world back then - ARM was non existent in mobile and datacenter, Apple was using Intel chips and AMD was totally irrelevant and flirting with bankruptcy. Intel was the dominant player at the time in design and fabrication. Odds are that most of the employees from 2011 don’t even work there anymore.

I could make similar tinfoil hat conspiracies with Phoronix but hanlan razor is a thing.
 

controlflow

Member
Feb 17, 2015
186
315
136
Pretty detailed ARL review from TechPowerup



They used DDR5 6000. Memory latency was a pretty crappy 88.4 ns.

I found these comments interesting:

"Intel has been praising their new Lion Cove architecture for months now, and we finally have our own results. Overall application performance of the Core Ultra 9 285K is improved by 1.2% over the Core i9-14900K—not much. AMD's flagship, the Ryzen 9 9950X still wins, by a small 3.4% margin. Once we take a look at individual benchmarks we can see why Arrow Lake isn't winning more battles. In several of our tests, the new processors lag seriously behind the competition, sometimes by significant amounts. I'm not exactly sure yet what causes it, but my first guess would be mis-scheduling of threads, either to E-Cores or bouncing between cores too often. Interestingly, this happens mostly in light and medium workloads. Heavy workloads run very well and show how mighty these new P-Cores are. "


"
When pairing Windows 24H2 with Arrow Lake, performance will be terrible—we've seen games running at 50% the FPS vs 23H2. One solution is to turn off Thread Director or disable the "Balanced" power profile, which is why we decided to use 23H2 for the time being. Last but not least, there are some driver issues and bluescreens when both a dGPU and iGPU are active at the same time.

To me, this experience felt a bit like AMD Ryzen 1st generation, but I'm confident that Intel can solve all these issues, like AMD did. "
 

poke01

Platinum Member
Mar 8, 2022
2,352
3,081
106
Anyway to prove that? Otherwise we’ll be dealing with conspiracy theories about Maxon and Intel for the next 2 years.
So I have no absolute proof but the leaked M4 ST Cinebench 2024 is any indication, it likes cores that are well fed and obviously cores that are fast as well.
 

Hulk

Diamond Member
Oct 9, 1999
4,613
2,645
136
Here are some things I need to know, among others, before I can come to a conclusion regarding ARL.
1. What Vcore is when frequency is at 5.7GHz and 5.4GHz.
2. Confirmation that the E cores aren't engaging when they shouldn't be, resulting in some of the abnormally low performance we're seeing in things like MS Office.
3. As above, but is there something going on with Windows and ARL that is hindering performance?
4. Isolate the E cores in some benches just because I'm curious. For example. Run some MT benchmarks on both the 285K and 14900K with 1 P core @ 800MHz (basically taking them out of the mix) and setting the E's at 4.4 for both Skymont and Gracemont. Report power as well. Is Skymont the star of the show as we are thinking? If Skymont is really carrying the load then perhaps Lion Cove is lacking more than many of us would have thought.
 
Reactions: Tlh97 and carancho

poke01

Platinum Member
Mar 8, 2022
2,352
3,081
106
Maxon s historical relation with Intel is documented, so why are you doing like it s a conspiration theory while it s a fact.?.
Because that video is so old now?

I will reserve my “CB2024 is Intel sabotaged” till next week to confirm the ST score of M4. If it’s real then, we can safely say that CB2024 just likes fast and well fed cores.
 

Abwx

Lifer
Apr 2, 2011
11,592
4,410
136
Do you really expect Zen 5 to win 100% of the benchmarks otherwise there’s a conspiracy going on?

I watched the video and it was created in 2011. Do you remember the state of the x86 CPU market in 2011? It was a totally different world back then - ARM was non existent in mobile and datacenter, Apple was using Intel chips and AMD was totally irrelevant and flirting with bankruptcy. Intel was the dominant player at the time in design and fabrication. Odds are that most of the employees from 2011 don’t even work there anymore.

I could make similar tinfoil hat conspiracies with Phoronix but hanlan razor is a thing.

Agree on the context at the time, but Philip Losch is now even higher in Maxon s hierarchy since he s currently chief technology architect, this is a german company and in Europe there s not the turnover that could exist in the US or even China, people generaly make most of their career in a single entreprise if possible and he was at the origin of Maxon creation back in 1989.

Recently they did put the focus on Apple as there s lot of creatives that use this brand s computers, and in a significantly higher proportion than Apple s general marketshare, that s why they optimised Cinema 4D for their hardware and also in their bench since R23, Qualcomm s good numbers in CB is just a by product of this ARM dedicated optimisation directed primarly toward Apple.
 
Last edited:
Reactions: lightmanek

OneEng2

Member
Sep 19, 2022
160
210
86
Power efficiency still pretty meh even with 3nm and all these "efficiency" cores. Multi threaded performance also not that amazing, basically ties zen 5 average.

People seem to be severely over estimating e-cores. The only thing they seem good for is optimizing the die area for Cinebench r24.
I agree. N3B should have provided fertile ground for better power efficiency than we are seeing. If all AMD did was shrink 9950x down to N3B we would expect much better results than this. In fact, we would expect that the additional transistor budget would have been used in the desktop to much better effect IMO.

The e Cores on Arrow Lake seem pretty good to me (especially without SMT). What exactly do you use as evidence for this statement?
Intel is fine? I wholeheartedly disagree.
Foundry is a big problem. Just last year foundry had a $7B loss. That's why their operating income was only $93M despite them having $54.2B revenue!
This year and next year will be another loss in foundries, who knows how big, I wouldn't be surprised If It was even bigger with all the money paid to TSMC.
People shouldn't forget that the money they pay to TSMC is the money, which should have gone to their foundries.
Check out my post from a different thread. Link
I believe @511 mentioned this earlier, but Intel design team has relied WAY too much over the years on the foundry making miracles for them. As the cost of new Lithography processes and machines has grown exponentially, continuing on this strategy is a death sentence. Intel must be able to sell its foundry services to spread the obscene cost of the development to others as TSMC does. They ALSO need to get their design teams aligned with the idea that they must design within the limits of the process generation they will release at and not to expect a 1-2 node advantage over the entire industry for their designs to be competitive.
Intel really needed to keep hyperthreading. It's easy to see why Intel is behind without it.
I agree. Study after study has shown that both the power efficiency and space efficiency of SMT is hugely worth it. I am baffled by this decision as well.
AGREED

Their biggest lie: "We axed HT for single threaded performance"

WHAT single thread performance????

You are literally losing in almost 99% of games!!!!

Like, you don't wanna hear my shrieking screaming voice right now, Intel!

Lion Cove's expanded structures probably only work to the max with HT enabled!
First, I believe that the ring bus implementation is hurting Arrow Lake greatly. I am pretty sure that Intel will figure it out in the next iteration and the Lion Cove follow on core will reap huge benefits from it in ST performance (and everyone's blessed games for the love of God).

Second, I have also heard a lot about Lion Coves architecture being created such that the resources are all kept pretty busy and that as a result, SMT would not provide much of an uplift. If that is correct, then it isn't enough for Intel to just slap on SMT, it would require a pretty big overhaul.
With SMT they would had gained the equivalent of 2P assuming a paltry 25% SMT gain, that s not negligible, eventualy they ll get back to this approach in their next designs as the bigger a core the more the SMT relevance.
Read my reply above. I think it would require an overhaul of the design to make SMT an effective part of the design.
This. AMD is utterly dominating the Cloud and server hosting market. If you need a VPS/VDS or a dedicated server it's all Ryzen9 for frequency or Epyc for compute. SMT allows hosts to cheaply sell more vCPUs per node, and that allows more VM nodes per machine. As cool as Granite Ridge is, it has nothing against Turin dense for the cloud.
... and THIS is one of the biggest problems Intel has. AMD's Zen 5 design works beautifully in DC applications. It is interesting that only Turin dense is on N3E while Turin is on N4P. I will be interested to see how Clearwater Forest on 18A does.
Maybe the disappointment with both Zen 5 and Arrowlake is due to unwarranted hype.

How many believed the 30-40% per clock gains by MLID? If people here didn't believe MLID, many outside this forum definitely did.

Zen 5 was hyped because there seemed to be structures that enlarged enough that such high numbers for gains seemed reasonable. It is a lesson that high level disclosures are not worth much.
Certainly we all love a good ole Core 2 release like back in the day. Better in everything by a long shot, less power ..... and by God, it is even a nicer looking processor!

I think that with the death of "Tock" due to the exponential cost of new process nodes ..... and .... you know ... physics, we will be seeing more "Tick's" in the future and that 5% to 10% is about all we should be expecting IMHO.
......
So Lion Cove's 10% gain should have got it's branch predictor little better too, not worse. Branch predictor is very, very important as less misses mean less flushed pipelines, meaning more performance and less wasted power. And it improves ILP. So Lion Cove is wasting it's resources elsewhere, like the overly big decode for one. Where else is LNC wasting it?
One has to wonder had the spent a little bit more transistor budget on a better branch predictor, and a little bit more on SMT, and wouldn't have botched the ring bus, that we wouldn't have seen a pretty darned fantastic CPU today.
ahh I know why it performs well in CB2024, it’s not cause the P core is awesome, it’s because DRAM bandwidth effects ST by a noticeable margin.

The older Cinebench versions are not dependent on memory for ST scaling.
Well ..... inevitably as you process more information (and rendering is a VERY information heavy workload), you must be able to feed the engine. So I would argue that the P core is, in fact, "awesome" in that it is able to fully utilize all that considerable bandwidth. It also seems to do well in CB2024 MT.
This is one of those times when I really wish Anand was still doing this so he could tell me what to think right now.
Awww yes, Anand was definitely a gem. I remember debating with him in the very early day's of AnandTech. Good guy.
 

poke01

Platinum Member
Mar 8, 2022
2,352
3,081
106
Recently they did put the focus on Apple as there s lot of creatives that use this brand s computers, and in a significantly higher proportion than Apple s general marketshare, that s why they optimised Cinema 4D for their hardware and also in their bench since R23, Qualcomm s good numbers in CB is just a by product of this ARM dedicated optimisation directed primarly toward Apple.
If by optimisation you mean adding better NEON support in CB2024, then yes?

R23 “supported” NEON but it was pitiful. The point is Apples cores do well in other ST benchmarks like Geekbench 5/6, web, photoshop, affinity etc…


Whereas Intel does does only well in ST in Cinebench 2024. I really hope it’s just firmware issues. Hopefully will get sorted in a month. For example that Speedometer 3.0 score is horrible and that’s on Linux.
 
Reactions: okoroezenwa

Hans Gruber

Platinum Member
Dec 23, 2006
2,329
1,233
136
If by optimisation you mean adding better NEON support in CB2024, then yes?

R23 “supported” NEON but it was pitiful. The point is Apples cores do well in other ST benchmarks like Geekbench 5/6, web, photoshop, affinity etc…


Whereas Intel does does only well in ST in Cinebench 2024. I really hope it’s just firmware issues. Hopefully will get sorted in a month. For example that Speedometer 3.0 score is horrible and that’s on Linux.
It is a complete waste of time trying to compare Apple CPU's to X86 processors. Apple stuff is geared towards power efficiency and multimedia performance. They kick ass in arcade style games that you play on smart phones. Not AAA graphics intensive multiplayer games. Apple is great for Netflix and battery life is beyond 10 hours. Apple processors should be compared to Snapdragon processors and the new MediaTek Dimensity 9400 CPU.
 
Reactions: Nothingness

DavidC1

Golden Member
Dec 29, 2023
1,086
1,739
96
But yes that triangle between the 9950X and 285K shows what looks to be the advantages of more cores and more efficient glue.
Some of that maybe, but mostly it looks like Intel rejigged the optimization so it benefits lower power more than the other end. The slope of the curve is different - that is a legitimate optimization criteria.

Also, if the curve stays true, then at 55W it should still be 28K or more.
 

OneEng2

Member
Sep 19, 2022
160
210
86
I actually doubt this. I think it'd end up about the same as the 4nm 9950X but smaller. And perhaps a bit more efficient at lower TDPs which aren't relevant for the SKU.
I would expect 9950X on N3B, to yield the same power efficiency as it has now .... which is already better than Arrow Lake and be able to do so with about a 10-15% clock increase.

Why do you think differently?
 

gdansk

Diamond Member
Feb 8, 2011
3,114
4,844
136
I don't want to be an entirely negative Nancy. So I will say that Python and nginx performance look good in Phoronix tests.

But on to the negative side: Intel refused to say anything about LGA1851's future even when asked by tech media. Is there even still a plan for Arrow Lake Refresh? It's Intel so I had previously assumed it was inevitable but that seems weird.
 

gdansk

Diamond Member
Feb 8, 2011
3,114
4,844
136
I would expect 9950X on N3B, to yield the same power efficiency as it has now .... which is already better than Arrow Lake and be able to do so with about a 10-15% clock increase.

Why do you think differently?
I mean I don't expect a hypothetical 9950X 3nm edition to be much more efficient except in low TDPs. Check the chart above where 285K out performs the 9950X (under 150W). And there it is the IOd which is wasting power and here Intel has an advantage due to interconnect.
 

DavidC1

Golden Member
Dec 29, 2023
1,086
1,739
96
4. Isolate the E cores in some benches just because I'm curious. For example. Run some MT benchmarks on both the 285K and 14900K with 1 P core @ 800MHz (basically taking them out of the mix) and setting the E's at 4.4 for both Skymont and Gracemont. Report power as well. Is Skymont the star of the show as we are thinking? If Skymont is really carrying the load then perhaps Lion Cove is lacking more than many of us would have thought.
Computerbase has that tested on CB2024.

If you divide 285K 8P, it's 96, but scaling isn't perfect so let's say it's 100. 1P+16E = 1444, so without the P, it's 1340.

1340 divide by two is 670, but scaling isn't perfect. Cinebench is about 90% scaling but let's assume few different points:
-95%: 687
-90%: 705

So roughly Golden Cove levels. By the way, in Cinebench Lion Cove is only 7% faster than Golden Cove. It's not a perfect way of comparison but gives us a rough idea.

Skymont comparison shows 31%, but it's a bit skewed as the 1P has HT disabled on 285K but enabled on 14900K. So, E-to-E it's higher than 31%. If you assume 117 points for the 14900K's HT enabled P core, then it's 1340 vs 983, or 36%.
 
Last edited:

cannedlake240

Member
Jul 4, 2024
163
94
61
Computerbase has that tested on CB2024.
View attachment 110244
If you divide 285K 8P, it's 96, but scaling isn't perfect so let's say it's 100. 1P+16E = 1444, so without the P, it's 1340.

1340 divide by two is 670, but scaling isn't perfect. Cinebench is about 90% scaling but let's assume few different points:
-95%: 687
-90%: 705

So roughly Golden Cove levels. By the way, in Cinebench Lion Cove is only 7% faster than Golden Cove. It's not a perfect way of comparison but gives us a rough idea.

Skymont comparison shows 31%, but it's a bit skewed as the 1P has HT disabled on 285K but enabled on 14900K. So, E-to-E it's higher than 31%. If you assume 117 points for the 14900K's HT enabled P core, then it's 1340 vs 983, or 36%.
Clearwater with its stronger L3 will be a beast if 18A doesn't disappoint
 
Reactions: Tlh97 and OneEng2
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |