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.)

 

[Henry swagger]

It is a alright benchmark that at least scales compared to GB6. CPU-Z is a good indicator if compared against other Intel CPUs.
Chips and Cheese analyzed the bench and came to the conclusion that the instruction mix does not accurately resemble consumer workloads.
I call half BS on that, which bench really resembles consumer workloads SPEC - FP or INT, Cinebench, all the other crazy SIMD benches ?
I think the bench got overcriticized because AMD is slow in it, showing that the Zen4 core is less wide AND SLOWER, which it is.

I cannot make this sh... up, wait for GB6 benches and don't trust CPU-Z.....

The leaks seem possible, E-Cores are buffed beyond 50% and the P core are min +15% faster with additional news of the new layout.
Current System (13700K@1,34V 2P@5,8 6P@5,7 8P@5,5 8E@4,4 = CPU-Z : ST=938, MT= 13100), so quite realistic.

Cpu z line up perfectly with cinebench r23..

 

[Henry swagger]

We should ban CPU-Z results, what a waste of time.

Like what does that benchmark even tell you?

And geekbench is better lol ?

 

[Thunder 57]

And geekbench is better lol ?

It sucks, but it is still more useful than CPU-z.

 

[Kepler_L2]

Probably fake score: https://tieba.baidu.com/p/9073941826?pid=150518902691

 

[ondma]

Nobody is making a 50% IPC gain in a single generation, that's simply not possible.

It also doesn't make sense they'd use TSMC N2 when 18A is supposed to be ready before it (if not 14A!) and they keep claiming they are going to achieve process leadership. That article reads like someone's wet dream about what they want to happen, much like how people here were building up all this hype for Zen 5 to gain 40% or more IPC.

I wouldn't call it "not possible". Zen did it over Bulldozer, and Skymont is in the ballpark of 50% increase. Granted, both are coming vs a much weaker baseline than intel's big cores. And as other posters said, it was probably based on a 2 or 3 generation leap. I am not saying it will happen, admittedly, it is very unlikely.

 

[coercitiv]

So in a desktop motherboard, there is the "Processor" and the "Chipset", as shown in the above diagram.

Is this true for laptop motherboards too? My understanding is that in laptops, the Chipset and Processor are integrated into one SoC. Is that so?

Most laptop designs that aim to have decent battery life will have the chipset on package or on-die, because it affects idle power consumption. Speaking specifically about Intel, historically they preferred to have the chipset on package for low power SKUs and kept it separate for 45W+ products. Here's a screenshot with Skylake packages for different form factors:



Obviously as we move towards the modern implementations like MTL and LNL, integration became tighter. The "discrete" chipset is probably on it's way out even for higher performance products from Intel.

 

[ondma]

Until CPU-z corrects the design of its benchmark to respect the CPPC scheduling of Ryzen, the results will be a wildcard because it always forces to core 0, which could be any random quality core boosting to any unknown frequency below or up to max singlethread boost.

Comparing any two samples of Ryzen will give you unpredictable results due to this.

Comparing Ryzen ST results to intel is also disingenuous because it will schedule properly on Intel to the fastest core.

Not saying CPU-Z is a valid benchmark or not, but to be fair, Acros did state that it was useful for Intel vs Intel.

 

[TwistedAndy]

It's pretty weird to see 20-30% IPC claims. Intel has given us some first-party numbers:



It makes sense to expect similar numbers for P-cores in Arrow Lake vs. Raptor Lake.

Also, if we take a closer look at Lion Cove's performance/power curve, we will see that it's steeper at the beginning and "saturates" faster. There are fewer reasons to push the power limits higher. Probably, the PL2 for Arrow Lake will be decreased to a more reasonable 180-200W. PL1 is expected to be 125W for ARL-S.

 

[Doug S]

It's pretty weird to see 20-30% IPC claims. Intel has given us some first-party numbers:

Dunno, it is pretty much a repeat of AMD telling us that Zen 5 would be a 15% IPC gain and people clutching onto the flimsiest of straws and crazy conspiracy theories to justify hyping it to 40% gain and beyond.

 

[Tup3x]

It's pretty weird to see 20-30% IPC claims. Intel has given us some first-party numbers:

View attachment 102235

It makes sense to expect similar numbers for P-cores in Arrow Lake vs. Raptor Lake.

Also, if we take a closer look at Lion Cove's performance/power curve, we will see that it's steeper at the beginning and "saturates" faster. There are fewer reasons to push the power limits higher. Probably, the PL2 for Arrow Lake will be decreased to a more reasonable 180-200W. PL1 is expected to be 125W for ARL-S.

Well, some test can very well show 20-30% improvement as Intel showed but average is not going to be like that.

 

[DAPUNISHER]

Holy cow. This thread turned into a mess really fast and devolved into the usual conspiracy drivel.

The performance of ARL-S is more or less known. There are Intel NDA overclocking guides that have already been given to partners and tons of ES samples in the wild. There’s no data to suggest a +20% 1T performance increase. Hype it at your own peril.

I agree. I don't know why some think this is a versus thread. Or are disregarding the CPU forum rule about vendor threads being safe spaces but the next post throwing shade at Intel or pimping AMD is getting double ban bingo spaces filled.

If you (collective you) don't like what is posted in this thread, don't read it. You can't come in here and start accosting the Intel users about any claims made. And keep AMD out of it, they have their own threads. As I have posted before and The Offspring rocked - You gotta keep 'em separated.

Mod DAPUNISHER

 

[KompuKare]

There are fewer reasons to push the power limits higher. Probably, the PL2 for Arrow Lake will be decreased to a more reasonable 180-200W. PL1 is expected to be 125W for ARL-S.

And that's the best thing IMO.
What I would call a return to sanity after all these pushed to the limit (or beyond) parts Intel released the last few years.
IPC 15% ± 5% or whatever we end up at the end is great but what had made Intel an automatic no for me the last few years was the power usage.

 

[TwistedAndy]

Well, some test can very well show 20-30% improvement as Intel showed but average is not going to be like that.

Intel Lion Cove has 33% more peak throughput than Raptor Cove (8-wide vs. 6-wide), but in the real apps, the IPC increase will be closer to the values presented by Intel (10-20%).

Obviously, some apps will be able to utilize all the throughput and show ~30% increase in performance (as we see on the Intel's chart), but it will be rather an exception than a rule.

What I would call a return to sanity after all these pushed to the limit (or beyond) parts Intel released the last few years.

Yes, leaks were suggesting that Arrow Lake desktop CPUs will have 125W PL1 and 177W PL2. It looks pretty reasonable.

As for the mobile ones, ARL-HX will have the same 55W PL1 and 115W PL2.

 

[DavidC1]

Obviously as we move towards the modern implementations like MTL and LNL, integration became tighter. The "discrete" chipset is probably on it's way out even for higher performance products from Intel.

The big picture reason for Intel taking forever to integrate the PCH is because they used it to fill their older fabs. It is the same reason why they took forever to integrate the memory controller too. They had the CPU, MCH and the IOH, and competitive pressure made them integrate most of the MCH into CPU and rest into PCH.

Now with Foveros and Tiles they have zero reason not to do so. Of course the cost isn't free, but it's much better in terms of communication speeds and power. For high performance such as CPUs and GPUs it might be difficult to do it fully and always a compromise but PCH should be easier and have integration benefits.

 

[KompuKare]

The big picture reason for Intel taking forever to integrate the PCH is because they used it to fill their older fabs. It is the same reason why they took forever to integrate the memory controller too. They had the CPU, MCH and the IOH, and competitive pressure made them integrate most of the MCH into CPU and rest into PCH.

I always suspected that filling the old fabs was the reason that Atom was neglected for so long.
Well until years later they panicked and threw dollar bills at anyone willing to make tablets with Intel Atom. Think most of those £50 tablets have ended up in landfill by now.

 

[dullard]

It's pretty weird to see 20-30% IPC claims. Intel has given us some first-party numbers:

View attachment 102235

Take a look at the graph on the left. It shows an average (maybe median) of +14% IPC. But, that includes two benchmark results with negative IPC changes and two benchmark results with about double the average IPC change (roughly 25% to 30%). So, I don't think it is weird at all to see 20% to 30% IPC claims as long as people clearly state that that is just one benchmark--and it is on the rest of us to acknowledge that one benchmark is not representative of a processor as a whole. No conclusion should ever be made from just one benchmark unless that one benchmark is all you plan on using with that processor.

 

[Wolverine2349]

Intel Arrow Lake-S Engineering Sample Shows 25% Single-Thread Performance Improvement Over i9-13900K

An engineering sample of the Intel Core Arrow Lake-S desktop processor has been tested, revealing its performance metrics in the CPU-Z benchmarks. This sample demonstrated a score of 1143.

www.guru3d.com

26.71% single thread improvement over 13900K per CPU-Z i above source.

Though 22% regression in multi threaded performance.

Either HT makes that big a difference, the SKymont cores do not have near the IPC gain as thought over Skymont, they are using the 6+8 or all of the above. Even if 6+8 if Skymont had the IPC increase to put it at Raptor Cove, I would think that should more than make up for lakc of HT and onky 8 E cores if it indeed has 2% better IPC than Gracemont??

But Lion Cove oh my so good and the real real and IPC increase maybe 25% afterall if true and the latency can be good as well. Please Intel 12 P core Arrow Lake CPU.

 

[Hitman928]

Intel Arrow Lake-S Engineering Sample Shows 25% Single-Thread Performance Improvement Over i9-13900K

An engineering sample of the Intel Core Arrow Lake-S desktop processor has been tested, revealing its performance metrics in the CPU-Z benchmarks. This sample demonstrated a score of 1143.

www.guru3d.com

26.71% single thread improvement over 13900K per CPU-Z i above source.

Though 22% regression in multi threaded performance.

Either HT makes that big a difference, the SKymont cores do not have near the IPC gain as thought over Skymont, they are using the 6+8 or all of the above. Even if 6+8 if Skymont had the IPC increase to put it at Raptor Cove, I would think that should more than make up for lakc of HT and onky 8 E cores if it indeed has 2% better IPC than Gracemont??

But Lion Cove oh my so good and the real real and IPC increase maybe 25% afterall if true and the latency can be good as well. Please Intel 12 P core Arrow Lake CPU.

Tell me you didn’t read the last couple pages of the thread without telling me you didn’t read the last couple of pages. . .

 

[eek2121]

ICYMI: The CPU-Z benchmark is fake.

Is the 15W 8840U score with a 15W PL1/2, or just a 15W PL1? Because the guy who gave the 17W LNL score confirmed it's 17W PL1/PL2.

AMD chips use a singular power limit (PPT), not PL1/PL2.

It's pretty weird to see 20-30% IPC claims. Intel has given us some first-party numbers:

View attachment 102235

It makes sense to expect similar numbers for P-cores in Arrow Lake vs. Raptor Lake.

Also, if we take a closer look at Lion Cove's performance/power curve, we will see that it's steeper at the beginning and "saturates" faster. There are fewer reasons to push the power limits higher. Probably, the PL2 for Arrow Lake will be decreased to a more reasonable 180-200W. PL1 is expected to be 125W for ARL-S.

You are misrepresenting what Intel stated. They stated pretty clearly that the core in other products will differ from the one in Lunar Lake. Arrow Lake desktop, for example, will have faster cache, and more of it. It is debatable how much additional IPC uplift the cache changes will provide, but it will be > 0%. That excludes any other changes that Intel may include.

Absolute performance will be even greater, and I have seen NO indications there will be a clock regression vs the 13900k. 1T clocks should most certainly either be unchanged or possibly higher. Base clocks also appear to be up.

 

[Hitman928]

Absolute performance will be even greater, and I have seen NO indications there will be a clock regression vs the 13900k. 1T clocks should most certainly either be unchanged or possibly higher. Base clocks also appear to be up.

Are we memory holing the 14th gen then?

 

[Wolverine2349]

Tell me you didn’t read the last couple pages of the thread without telling me you didn’t read the last couple of pages. . .

Oh yeah thats right. People saying CPU-Z is flawed benchmark. Well for AMD CPUs it seems to be but it seems more accurate for Intel CPUs? Though did not fully read them just scanned briefly so may have missed?

So am I not right or is that source not right?

 

[Hitman928]

Oh yeah thats right. People saying CPU-Z is flawed benchmark. Well for AMD CPUs it seems to be but it seems more accurate for Intel CPUs? Though did not fully read them just scanned briefly so may have missed?

So am I not right or is that source not right?

Outside of the CPUz being worthless discussion, the original source was a post in the Baidu forum, but another user there apparently called it out as being an outright fake.

 

[SiliconFly]

Outside of the CPUz being worthless discussion, the original source was a post in the Baidu forum, but another user there apparently called it out as being an outright fake.

Thats sad. But I'm still hoping that it's somehow true!

I tried to analyze the image trying to find any ps edits, but the result was inconclusive. The (false) color bleed looks more like jpg compression artifact rather than alteration. Either the image is real or they've done an amazing pixel perfect job (which isn't difficult actually) or they might have faked it some other way. Hard to say.



But what if it isn't fake? 😶

 

[Kepler_L2]

Oh yeah thats right. People saying CPU-Z is flawed benchmark. Well for AMD CPUs it seems to be but it seems more accurate for Intel CPUs? Though did not fully read them just scanned briefly so may have missed?

So am I not right or is that source not right?

CPU-Z just benchmarks how good your CPU is at branchless SSE code, which is not a real workload.

CPU-Z’s Inadequate Benchmark

CPU-Z is a hardware information tool from a company called CPUID, not to be confused with the CPUID instruction. Besides showing basic CPU, motherboard, and memory information, CPU-Z features a bui…

chipsandcheese.com

 

[whoshere]

Thats sad. But I'm still hoping that it's somehow true!

I tried to analyze the image trying to find any ps edits, but the result was inconclusive. The (false) color bleed looks more like jpg compression artifact rather than alteration. Either the image is real or they've done an amazing pixel perfect job (which isn't difficult actually) or they might have faked it some other way. Hard to say.

But what if it isn't fake? 😶

I can forge any CPU-Z score you need with a 100% accuracy, meaning you'll never be able to tell it's fake.

There's no need to forensically examine this screenshot, as it's completely useless.