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.

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.)

Timmah! · Wednesday at 2:49 PM

So this rumored 16P + 32E + 4LP CPU is happening when? I stopped following Intel releases since i got AMD platform.

Thunder 57 · Wednesday at 3:02 PM

Timmah! said:
So this rumored 16P + 32E + 4LP CPU is happening when? I stopped following Intel releases since i got AMD platform.

At least a year away.

Fjodor2001 · Wednesday at 4:43 PM

adroc_thurston said:
I dunno try looking at AMD DT channel sell thru. It ain't hard.

If it ain’t hard for you, please share it.

adroc_thurston · Wednesday at 4:43 PM

Fjodor2001 said:
If it ain’t hard for you, please share it.

I ain't sharing stuff I pay Actual Real Money for.

Fjodor2001 · Wednesday at 4:45 PM

adroc_thurston said:
I ain't sharing stuff I pay Actual Real Money for.

As usual you make dead certain claims but provide no evidence to back it up.

adroc_thurston · Wednesday at 4:47 PM

Fjodor2001 said:
As usual you make dead certain claims but provide no evidence to back it up.

You're free to pay for MSS reports with detailed share breakdowns yourself.
Canalys et al are your friends.

Fjodor2001 · Wednesday at 4:52 PM

adroc_thurston said:
You're free to pay for MSS reports with detailed share breakdowns yourself.
Canalys et al are your friends.

No need to. I got detailed sales share breakdowns personally sent to me from the Intel and AMD CEOs that contradict your claim, but unfortunately I cannot share that info as you surely understand. Just trust me on this.

Det0x · Wednesday at 5:16 PM

Dont know if this have been posted already

Intel’s upcoming Nova Lake processor introduces a different approach to its cache architecture by having two performance cores share a 4MB L2 cache. This change is a departure from the current Arrow Lake processors, where each large performance core gets its own dedicated 3MB of L2 cache. This shared cache design, confirmed by leaks from source HXL, aims to improve cache resource utilization and scalability across the cores. The core layout of the Nova Lake CPU shows two distinct performance cores, often referred to internally as "Coyote Cove," arranged together in a cluster. Unlike the previous design where cores had independent caches, these two cores will share a larger L2 cache pool. This could allow for a more flexible allocation of cache resources, potentially increasing the effective cache size accessible to each core during high-demand scenarios.

However, this architectural adjustment raises questions about how the shared cache will perform when both cores try to access it simultaneously. The potential for cache contention or latency increases could impact performance, especially in workloads that heavily rely on L2 cache access. Intel’s current Arrow Lake design avoids this by giving each core a separate 3MB cache, minimizing such conflicts but limiting the maximum cache available per core.

The actual impact of the shared 4MB L2 cache on processing speed, efficiency, and power consumption will become clear only after the processor hits the market and undergoes thorough benchmarking. It remains to be seen whether this design provides a net positive effect or introduces trade-offs that affect certain types of applications more than others.

DavidC1 · Wednesday at 5:31 PM

OneEng2 said:
20-30% of x86 sales are desktop processors.

What?

Intel's 2023 revenue in client was $10 billion for desktop and $17 billion for laptop. Last time I heard the volume was 40/60 and based on the revenue that might still be true.

I am pretty sure Intel cares about 25% of that $10 billion dollars being sold as high margin, high revenue chips.

I don't consider any laptop to be a "gaming" system, but we could consider some small percentage of laptops to be purchased for the purpose of mobile gaming.

Really. So a $3000 laptop that weights 8lbs and has a 19 inch screen, lasts 2 hours on battery is bought by people who doesn't game at all?

OneEng2 said:
Games are likely one of the least used applications on x86.

Perhaps I am looking at this all wrong, but it seems like gaming isn't that big of a consideration for the x86 market.

Are you relying on ChatGPT for your answers?

Or actually you know less than you think you do.

jpiniero · Wednesday at 6:08 PM

DavidC1 said:
Intel's 2023 revenue in client was $10 billion for desktop and $17 billion for laptop. Last time I heard the volume was 40/60 and based on the revenue that might still be true.

For 2024, it was 19.1B for Notebook and 9.7B for Desktop.

DavidC1 said:
Are you relying on ChatGPT for your answers?

Ha, I could see that.

dullard · Wednesday at 6:14 PM

DavidC1 said:
What?

Intel's 2023 revenue in client was $10 billion for desktop and $17 billion for laptop. Last time I heard the volume was 40/60 and based on the revenue that might still be true.

I am pretty sure Intel cares about 25% of that $10 billion dollars being sold as high margin, high revenue chips.

1) The topic was laptop vs desktop, yet you only quote Intel numbers as if you forget AMD exists.
2) What Intel cares about is not what is being discussed.
3) What Intel revenue ratios might be are not what is being discussed.

What is being discussed in the quote you linked is percent of desktop vs laptop computers (I'll ignore mobile for this discussion). It is pretty widely reported that laptops are about 70% of the market and desktops are about 30%.

Sales of Client CPUs Soared in Q4 2023: Jon Peddie Research

www.anandtech.com

"70% of client PC CPUs sold in Q4 2023 were aimed at notebooks, which is up significantly from 63% represented by laptop CPUs in Q4 2022. Indeed, notebook PCs have been outselling desktop computers for years, so, unsurprisingly, the industry shipped more laptop-bound processors than desktop-bound CPUs."

Here is Q1 2024 data: laptops were 73%.

https://www.pcgamer.com/hardware/processors/laptops-are-selling-like-hot-cakes-compared-to-desktops-but-generally-cpu-shipments-are-sluggish/

Back to the original debate. As for the claim that "Gaming isn't that big a deal to most consumers of x86 chips.", I'll rate that a true. Most x86 chips are business-type work (office, home office, or general tasks like internet browsing, etc). As for the claim that "Games are likely one of the least used applications on x86", I'll rate that as laughably false.

desrever · Wednesday at 7:13 PM

Why is there so much stuff coming about Nova Lake, where is Panther Lake is so quiet?

poke01 · Wednesday at 7:47 PM

desrever said:
Why is there so much stuff coming about Nova Lake, where is Panther Lake is so quiet?

Because let’s just say it only exists to get laptop off N3B and to show Intel can do 18A other than that it’s not impressive CPU wise.

Thunder 57 · Wednesday at 7:50 PM

poke01 said:
Because let’s just say it only exists to get laptop off N3B and to show Intel can do 18A other than that it’s not impressive CPU wise.

That's my best guess as well unfortunately.

OneEng2 · Wednesday at 8:09 PM

Thibsie said:
Link ?

LOL. Fair. See the responses below . dullard beat me to it.

Fjodor2001 said:
No need to. I got detailed sales share breakdowns personally sent to me from the Intel and AMD CEOs that contradict your claim, but unfortunately I cannot share that info as you surely understand. Just trust me on this.

LMFAO. I was having a completely shitty day, but you gave me a great laugh. Thanks!

dullard said:
1) The topic was laptop vs desktop, yet you only quote Intel numbers as if you forget AMD exists.
2) What Intel cares about is not what is being discussed.
3) What Intel revenue ratios might be are not what is being discussed.

What is being discussed in the quote you linked is percent of desktop vs laptop computers (I'll ignore mobile for this discussion). It is pretty widely reported that laptops are about 70% of the market and desktops are about 30%.
View attachment 125883

Sales of Client CPUs Soared in Q4 2023: Jon Peddie Research

www.anandtech.com

"70% of client PC CPUs sold in Q4 2023 were aimed at notebooks, which is up significantly from 63% represented by laptop CPUs in Q4 2022. Indeed, notebook PCs have been outselling desktop computers for years, so, unsurprisingly, the industry shipped more laptop-bound processors than desktop-bound CPUs."

Here is Q1 2024 data: laptops were 73%.

https://www.pcgamer.com/hardware/processors/laptops-are-selling-like-hot-cakes-compared-to-desktops-but-generally-cpu-shipments-are-sluggish/

Back to the original debate. As for the claim that "Gaming isn't that big a deal to most consumers of x86 chips.", I'll rate that a true. Most x86 chips are business-type work (office, home office, or general tasks like internet browsing, etc). As for the claim that "Games are likely one of the least used applications on x86", I'll rate that as laughably false.

Thanks for posting the links. I was going through the tread and had a very very similar response .

But since someone asked if I got my information from ChatGPT, I decided to go and ask AI what it thought. Here is the response:

Question: What percentage of x86 global sales are for desktop vs laptop vs server??

The answer:

Based on the latest data available, specifically from Q1 2025:

Desktop: The desktop/mobile unit sales ratio for the overall x86 market is 1:2.5. This means that for every 1 desktop CPU sold, 2.5 mobile (laptop) CPUs are sold. This indicates that laptop sales represent a significantly larger portion of x86 units compared to desktops.

Laptop: Notebooks outsell desktops 3:1.

Server: The client/server unit sales ratio for the overall x86 market is 9.3:1. This means that for every 9.3 client (desktop/laptop) CPUs sold, 1 server CPU is sold. This indicates that server sales, in terms of units, are a much smaller segment compared to client PCs. However, the server segment boasts a significantly higher average selling price.

In summary, in terms of the number of units sold, the x86 market is dominated by laptop sales, followed by desktop sales, and then server sales. Server CPUs, despite being a smaller unit volume, represent a larger revenue portion due to their higher price points.

The links provided that backup the data are:

https://www.reddit.com/r/hardware/comments/1kocozp/q1_2025_market_share_for_x86_processors

AMD gained consumer desktop and laptop CPU market share in 2024, server passes 25 percent

But AMD's server march has started to slow down.

www.tomshardware.com

AMD records its highest server market share in decades — Intel fights back in client PCs

Another strong quarter for AMD.

www.tomshardware.com

So it seems that a number of sources consider desktop sales to be dwarfed by laptop sales (about 3:1).

If you want to argue the merits of the much larger margins that AMD enjoys by selling so much into the high end gaming and DIY market, these are quite valid arguments. Since AMD has a much more constrained supply, I am certain they would strongly prefer to sell AS MUCH as possible at the highest margins.

IMO, Intel's biggest hurt has NOTHING to do with gaming and EVERYTHING to do with DC. It also has to do with their money pit called foundry (at least for now).

If Clearwater Forest and Diamond Rapids do not manage to catch AND eclipse Venice D and Venice Standard, Intel is going to continue to have issues making money.

Seems to me like Intel's SMT implementation was never as good as AMD's. Where AMD gets ~1.4x performance from SMT in highly threaded loads, Intel gets 15-20%. Where AMD has a full 512bit AVX data path, Intel appears to be limiting theirs to 256bit in P cores and 128bit in E cores (someone correct me if this is wrong. I got it from a good source in SemiWiki, but it could be wrong).

I am not saying that Desktop is unimportant, only that it shouldn't be (and isn't for AMD as specifically stated from the company) since that isn't where the big profit comes from.

While everyone will easily agree that the highest profit margins are in DC, I wonder what the profit margin is in Desktop vs Laptop. I am guessing that Desktop has a higher margin since more laptops will be for business use (I am guessing that Desktop will either be for gaming or rendering or something else that requires power beyond a Laptop (but shy of HPC).

Still all those that have gloom and doom for Intel based on ARL's horrible gaming performance are missing the point. If Intel can create an apex laptop or server processor, the stock market (and banks) will applaud them .... although they are likely to get another thrashing here on the forum

OneEng2 · Wednesday at 8:23 PM

poke01 said:
Because let’s just say it only exists to get laptop off N3B and to show Intel can do 18A other than that it’s not impressive CPU wise.

One issue that has been reported by some pretty reputable people over at SemiWiki is that BSPDN suffers from hot spots in the design (more so than FSPD used in current designs). For lower power, lower clocked applications (laptops and DC) 18A may well be a match made in heaven. For desktop it might well be a nightmare.

While I think that N3B was quite expensive for Intel, I suspect it is a bargain compared to the total cost of ownership of 18A. Intel has sank over 20bn into 18A development. That is more than a Ford class US aircraft carrier!

I have hope for Nova Lake though. ARL was so obviously constrained by the horrendous latency issues Intel had in its first "tile design" that it seems like there may be a breath of fresh air on the horizon for the design.

I believe that AMD had some setbacks with Zen 2 for the same reason but Zen 3 is considered a breakthrough design when in fact it was likely just some adjustments and improvements to overcome the original latency problems of their chiplets.

Imagine if ARL->NVL is the same jump as Zen 2 to Zen 3.

Io Magnesso · Wednesday at 8:33 PM

poke01 said:
Because let’s just say it only exists to get laptop off N3B and to show Intel can do 18A other than that it’s not impressive CPU wise.

However, at least XE3 and Celestal are installed, so I think it's an important milestone.

Io Magnesso · Wednesday at 8:42 PM

OneEng2 said:
LOL. Fair. See the responses below . dullard beat me to it.

LMFAO. I was having a completely shitty day, but you gave me a great laugh. Thanks!

Thanks for posting the links. I was going through the tread and had a very very similar response .

But since someone asked if I got my information from ChatGPT, I decided to go and ask AI what it thought. Here is the response:

The answer:

The links provided that backup the data are:

https://www.reddit.com/r/hardware/comments/1kocozp/q1_2025_market_share_for_x86_processors

AMD gained consumer desktop and laptop CPU market share in 2024, server passes 25 percent

But AMD's server march has started to slow down.

www.tomshardware.com

AMD records its highest server market share in decades — Intel fights back in client PCs

Another strong quarter for AMD.

www.tomshardware.com

So it seems that a number of sources consider desktop sales to be dwarfed by laptop sales (about 3:1).

If you want to argue the merits of the much larger margins that AMD enjoys by selling so much into the high end gaming and DIY market, these are quite valid arguments. Since AMD has a much more constrained supply, I am certain they would strongly prefer to sell AS MUCH as possible at the highest margins.

IMO, Intel's biggest hurt has NOTHING to do with gaming and EVERYTHING to do with DC. It also has to do with their money pit called foundry (at least for now).

If Clearwater Forest and Diamond Rapids do not manage to catch AND eclipse Venice D and Venice Standard, Intel is going to continue to have issues making money.

Seems to me like Intel's SMT implementation was never as good as AMD's. Where AMD gets ~1.4x performance from SMT in highly threaded loads, Intel gets 15-20%. Where AMD has a full 512bit AVX data path, Intel appears to be limiting theirs to 256bit in P cores and 128bit in E cores (someone correct me if this is wrong. I got it from a good source in SemiWiki, but it could be wrong).

I am not saying that Desktop is unimportant, only that it shouldn't be (and isn't for AMD as specifically stated from the company) since that isn't where the big profit comes from.

While everyone will easily agree that the highest profit margins are in DC, I wonder what the profit margin is in Desktop vs Laptop. I am guessing that Desktop has a higher margin since more laptops will be for business use (I am guessing that Desktop will either be for gaming or rendering or something else that requires power beyond a Laptop (but shy of HPC).

Still all those that have gloom and doom for Intel based on ARL's horrible gaming performance are missing the point. If Intel can create an apex laptop or server processor, the stock market (and banks) will applaud them .... although they are likely to get another thrashing here on the forum

Arrow Lake's games are weak, but overall they aren't bad CPUs I can't say that it's really good if you ask me However At least it's much better than Bulldozer or Pentium D

511 · Wednesday at 10:10 PM

poke01 said:
Because let’s just say it only exists to get laptop off N3B and to show Intel can do 18A other than that it’s not impressive CPU wise.

Well the P core is based on slightly Shrunk down Lion Cove xDd.
The darkmont cores should be impressive though.

dullard · Wednesday at 10:17 PM

desrever said:
Why is there so much stuff coming about Nova Lake, where is Panther Lake is so quiet?

?

Panther Lake was all the leaks in Jan through mid-May. Intel discussed Panther Lake at CES, Intel Vision, and showed running at Computex. https://newsroom.intel.com/press-kit/intel-at-computex-2025
https://newsroom.intel.com/wp-content/uploads/2025/05/Intel-Computex-2025-Panther-Lake-2-scaled.jpg

Intel demos running Panther Lake systems, touts performance and power efficiency improvements

Panther Lake comes screaming to market in early 2026.

www.tomshardware.com

OneEng2 · Wednesday at 10:29 PM

Io Magnesso said:
Arrow Lake's games are weak, but overall they aren't bad CPUs I can't say that it's really good if you ask me However At least it's much better than Bulldozer or Pentium D

ARL is certainly not Bulldozer or Pentium D (or Itanic )! It appears to me that all of ARL weaknesses can be traced back to the high latency. As AMD has shown, the latency penalty typically attributed to going off-die to L3 and the memory controller, etc, can be elevated with the proper design choices. I suspect that it is still better to have everything on one big (and God awful expensive) die, but it is completely possible to have outstanding performance with chiplets/tiles. I am hopeful that Intel will show this with Nova Lake.

When Bulldozer launched, it was bad at MOST things. Ironically, Bulldozer suffered badly in single threaded workloads. It is ironic that Arrow Lake has reversed the roles.

Still, Bulldozer had issues in its architecture that bottlenecked its ST performance. I am still GUESSING that Arrow Lake ST performance (when it IS lacking .... which isn't always) is ONLY held up by its latency issues.

dullard · Wednesday at 10:40 PM

Io Magnesso said:
Arrow Lake's games are weak, but overall they aren't bad CPUs I can't say that it's really good if you ask me However At least it's much better than Bulldozer or Pentium D

Weak at gaming, but only when compared to the X3D chips. When compared to any other chip, Arrow Lake is on par. Not great, but not really bad either. It is so much more GPU dependent at that point.

Here are the updated benchmarks after the whole patch fiasco.

Games @ CPU limit

14600K

14700K

14900K

245K

265K

285K

9700X

9900X

9950X

6P+8E RPL

8P+12E RPL

8P+16E RPL

6P+8E ARL

8P+12E ARL

8P+16E ARL

8C Zen5

12C Zen5

16C Zen5

average gaming performance

94.6%

101.8%

103.6%

91.0%

96.6%

100%

97.6%

95.0%

99.7%

average game consumption

102W

133W

136W

76W

89W

100W

~86W

114W

120W

https://www.3dcenter.org/artikel/ei...ormance/neubetrachtung-arl-performance-seite2

MS_AT · Thursday at 4:20 AM

dullard said:
Weak at gaming, but only when compared to the X3D chips.

I always thought the problem was it was weak compared to RaptorLake; in gaming

DrMrLordX · Thursday at 5:01 AM

MS_AT said:
I always thought the problem was it was weak compared to RaptorLake; in gaming

That was almost inevitable, barring some breakthrough the likes of which Intel hasn't had in years. Raptor Lake flew too close to the sun. Arrow Lake isn't as suicidal.

511 · Thursday at 5:08 AM

MS_AT said:
I always thought the problem was it was weak compared to RaptorLake; in gaming

Ofc the mainstream Media is all about gaming

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

Senior member

Attachments

Golden Member

Diamond Member

Diamond Member

Diamond Member

Diamond Member

Diamond Member

Diamond Member

Golden Member

Golden Member

Lifer

Elite Member

Senior member

Diamond Member

Diamond Member

Senior member

Senior member

Member

Member

Platinum Member

Elite Member

Senior member

Elite Member

Senior member

Lifer

Platinum Member