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.





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]

25% better perf/watt indicates a 25% increase in performance iso wattage. What you're looking at is % lower wattage iso performance. Two different things.

25% better perf/watt at isoperf mean that it s 1.25x less power for the same perf, or if you prefer 0.8x the power at same perf.

25% better perf (at isopower) is another matter, to get the perf/Watt improvement at isoperf out of this number you have to know the form of the power/frequency curve, for Intel they shoot generaly for P = F^2.2 for their processes, wich mean that it would amount to 1.25^2.2 = 1.63, that is, 63% better perf/watt at isoperf.

 

[Abwx]

It depends on which part of the performance spectrum you cherry pick your power figures. If you look at a score of 16,000 then Intel's only behind by about 18%. Either way, I agree Intel needs a giant leap in efficiency to catch up never mind pull ahead.

I wont go into that debate further but i explain shortly the thing.

Say a CPU A that score 10 000 pts at 30W and a CPU B that score say 7000 pts at the same 30W, the former will have only 1.4x better perf/watt than the latter but that s not accounting for the score.

If we downclock CPU A such that it score also 7000 pts its power will tank from 30 to something like 12W, hence its perf/watt at equal throughput will be 2.5x better than CPU B.

You can see that by maxing CPU A perf you can create an artificial comparison that will make the far less efficient and much slower CPU B looking half good, that s why isoperf comparison is the only one relevant when talking efficiency.

 

[H433x0n]

I wont go into that debate further but i explain shortly the thing.

Say a CPU A that score 10 000 pts at 30W and a CPU B that score say 7000 pts at the same 30W, the former will have only 1.4x better perf/watt than the latter but that s not accounting for the score.

If we downclock CPU A such that it score also 7000 pts its power will tank from 30 to something like 12W, hence its perf/watt at equal throughput will be 2.5x better than CPU B.

You can see that by maxing CPU A perf you can create an artificial comparison that will make the far less efficient and much slower CPU B looking half good, that s why isoperf comparison is the only one relevant when talking efficiency.

Using your metric, every new product release will double perf/watt. With your math Zen 4 is infinitely more efficient than Zen 3, more than doubling perf/watt!

Not even AMD marketing use this crazy metric.

 

[Abwx]

Using your metric, every new product release will double perf/watt. With your math Zen 4 is 125% more efficient than Zen 3, more than doubling perf/watt!

Not even AMD marketing use this crazy metric.

View attachment 86186

125% more efficient, that s 2.25x more efficient.

In the pic below Intel state that the 13900K@65W is 4x more efficient than the 12900K@241W, that is, 300% more efficient at same throughput :

Core i9-13900K, i7-13700K & i5-13600K: Gaming-Könige im Test: Leistungsaufnahme und Effizienz

Intel Raptor Lake im Test: Leistungsaufnahme und Effizienz / Leistungsaufnahme in Anwendungen ab Werk / Der Verbrauch steigt

www.computerbase.de

So much for the crazy metric, it is here in front of your eyes by who you know...

 

[H433x0n]

125% more efficient, that s 2.25x more efficient.

In the pic below Intel state that the 13900K@65W is 4x more efficient than the 12900K@241W, that is, 300% more efficient at same throughput :

Core i9-13900K, i7-13700K & i5-13600K: Gaming-Könige im Test: Leistungsaufnahme und Effizienz

Intel Raptor Lake im Test: Leistungsaufnahme und Effizienz / Leistungsaufnahme in Anwendungen ab Werk / Der Verbrauch steigt

www.computerbase.de

So much for the crazy metric, it is here in front of your eyes by who you know...

uhh... No, it's not lol. I'm pretty sure you're misreading this chart.

At 125W isopower, the 13900K has a 20% performance lead in the ComputerBase link you shared. At 115W isopower, Intel has the 13900K with a 21% performance lead over 12900K. There's nothing on this chart that says 300% performance efficiency improvement. All of the percentages listed are for performance at isopower.

 

[Geddagod]

25% better perf/watt at isoperf

This right here, is not the default of what perf/watt means.
When Intel says that they gain 20% perf/watt from Intel 4, it means they gained 20% more performance iso power vs Intel 7. This is shown in this graph:

Notice how they equalized power, not performance for their perf/watt claim.
"perf/watt iso perf" would be watt/perf. Phoenix uses 45watts/16000 points, RPL uses 65 watts/16000 points. RPL uses 44% more watts per point at 16,000 points.
The reason why the denominator is usually equalized is because the fraction sign is understood as "per", for example 60 miles per hour.

25% better perf (at isopower) is another matter, to get the perf/Watt improvement at isoperf out of this number you have to know the form of the power/frequency curve

You would need to know the form of the power/frequency curve for literally everything. Even perf/watt iso perf. For example, at 16000 points, RPL uses 44% more power, but at 12,000 points, RPL uses ~53% more energy.


So yes, MTL gaining 25% better perf/watt over RPL should make it enough to roughly match Phoenix, using that notebook check graph provided, at least for the power levels described on the graph. And yes, we should be looking at improvements over a 6+8 RPL.

 

[DrMrLordX]

Intel creating a N5 gpu tile for MTL & another N3E gpu tile for ARL is very very expensive and doesn't really make sense if they're going to ditch the N5 tile so soon. Just not worth it.

Why? Arrow Lake doesn't even launch until Q4 of next year, and if it's anything like Meteor Lake it'll be very limited. Intel is also on the hook for a lot on N3-family wafers (maybe just N3B, but possibly also N3E). Expect them to use that capacity.

 

[igor_kavinski]

At 125W isopower, the 13900K has a 20% performance lead in the ComputerBase link you shared. At 115W isopower, Intel has the 13900K with a 21% performance lead over 12900K. There's nothing on this chart that says 300% performance efficiency improvement. All of the percentages listed are for performance at isopower.

13900K and 12900K are not really comparable. 13700K is the actual 12900K successor. 13900K is a mutant with 8 extra e-cores.

 

[H433x0n]

13900K and 12900K are not really comparable. 13700K is the actual 12900K successor. 13900K is a mutant with 8 extra e-cores.

I know, I'm just explaining the graph he provided. The real numbers are ~7% higher perf/watt at 125W when comparing 13700K and 12900K.

 

[Geddagod]

Or - it could have +20% better performance at iso power (which is exactly what Intel 4 is advertised with) then voila it has superior perf/watt across entire curve.

I'm pretty sure "node perf/watt improvements" never translate very well into product improvements over the vast majority of the perf/watt curve, but that could just be me misremembering...

 

[H433x0n]

I'm pretty sure "node perf/watt improvements" never translate very well into product improvements over the vast majority of the perf/watt curve, but that could just be me misremembering...

It technically means 20% perf/watt on an industry standard ARM core.

I personally think it’ll exceed that based on the base frequency listed for MTL compared to RPL.

 

[Abwx]

uhh... No, it's not lol. I'm pretty sure you're misreading this chart.

At 125W isopower, the 13900K has a 20% performance lead in the ComputerBase link you shared. At 115W isopower, Intel has the 13900K with a 21% performance lead over 12900K. There's nothing on this chart that says 300% performance efficiency improvement. All of the percentages listed are for performance at isopower.

They say that the 13900K use 25% of the power to match the 12900K, they compare at equal throughput, 25% of the power mean 4x more efficient and hence 300% better efficiency at same throughput, have you some trouble with the basic arithmetic operations..?...

That s clealy stated on the right of the pic, but looks like you re in a cognitive dissonance trap now, so any further debate is useless since you refusing to simply read what is in Intel s slide.

13900K and 12900K are not really comparable. 13700K is the actual 12900K successor. 13900K is a mutant with 8 extra e-cores.

Of course because the core count is not the same, but still they do such a comparison knowing that the perf/watt numbers will be inflated comparatively to a same core count comparison.

Actually they wanted to have numbers as good as AMD, you ll notice that they use the same 65W comparison vs a stock previous gen, just that it s equivalent to AMD using a 5900X as comparison to a 7950X...

 

[Cstops]

So MTL appears to be coming to desktop after all if this is any indication from PC World's interview with Michelle Holthaus (pulled link from TPU).

edit: Added TPU's link below as well for reference.

Intel Confirms Meteor Lake Desktop CPUs Coming in 2024

 

[SiliconFly]

20% better perf/isowatt allow for about 50% better perf/watt at isoperf, quite better but it wont beat Phoenix, notice that Intel stated that it will be "their most efficent mobile SKU ever", not "the most efficient mobile SKU ever", notice the nuance in their wording...

Basing performance on word play? Excellent. 👍

 

[SiliconFly]

...MTL-S to fill in some of the low end of the 14th gen stack but I just don't see how it makes sense.

Maybe Intel wants to lead in perf-per-watt in desktops too!

 

[SiliconFly]

I'm pretty sure "node perf/watt improvements" never translate very well into product improvements over the vast majority of the perf/watt curve, but that could just be me misremembering...

Exactly. Node advantage is just only one indicator. No one here seems to be talking about power-efficiency increase due to architectural improvements in MTL.

 

[mikk]

So far existing "28W" SKUs boost to 64W , that s the current Intel way, stated as 28W and benched at 64W...

They can boost to 64W, it's a maximum theoretical value. It's up to the OEM and other parameters if it really boosts to the maximum. The OEM could completely disable the PL2 if they wanted to.

 

[eek2121]

I'm very skeptical that Meteor will be released for the desktop in a meaningful way. If it is released it will be like Broadwell. Looking back I don't think anyone really considers Broadwell a desktop release. Haswell was simply clocking too high, think Devil's Canyon for the immature 14nm Broadwell node to compete.

We are in a very similar situation now except that Meteor Lake isn't competing with the 2nd gen Alder Lake (Raptor), it's competing with the 3rd gen (Raptor Refresh), which has pushed clocks even higher. Meteor, with minimal IPC gains, if any, on a new node will not be able to compete from a performance point of view. In addition, Arrow Lake is expected next year and would release concurrently or there about. It doesn't make sense.

I will say this right now, there will be no Meteor Lake part that will compete with the 14900K from a pure performance point of view on the desktop.

I'm not going to "un-write" anything I wrote above but if anything I guess it would be possible for MTL-S to fill in some of the low end of the 14th gen stack but I just don't see how it makes sense.

Finally, the remark that Intel VP said was very nebulous. When prodded to talk about keeping mobile and desktop releases on the same core she said something like "That's the goal." I read that meaning for 15th gen and/or beyond.

I don’t think anyone made such a claim. Meteor Lake coming to the desktop means literally just that.

I have a small desktop PC sitting on a shelf that runs an AMD Cezanne mobile chip. The system isn’t much bigger than my wallet in terms of size. MTL-P 6+8 in one of those small boxes would be amazing to have.

 

[Glo.]

MLIDs video on the topic of Meteor Lake desktops and is leaking Cinebench score of MTL 155H SKU.

P.S. That power draw is awful.

 

[DrMrLordX]

I have a small desktop PC sitting on a shelf that runs an AMD Cezanne mobile chip. The system isn’t much bigger than my wallet in terms of size. MTL-P 6+8 in one of those small boxes would be amazing to have.

There's a lot of chips that might be great in there. Phoenix would be nice. Meteor Lake might be even better (or not, see below). It really comes down to a question of how much money can be made selling a product like that. If Meteor Lake is launching in ~$1500 laptops this December, don't expect it to show up in $500 NUCs anytime soon. Or even $900 ones.

MLIDs video on the topic of Meteor Lake desktops and is leaking Cinebench score of MTL 155H SKU.

P.S. That power draw is awful.

Looks like he's tapping into the same source that others have cited when stating that Meteor Lake has not-great efficiency.

 

[eek2121]

There's a lot of chips that might be great in there. Phoenix would be nice. Meteor Lake might be even better (or not, see below). It really comes down to a question of how much money can be made selling a product like that. If Meteor Lake is launching in ~$1500 laptops this December, don't expect it to show up in $500 NUCs anytime soon. Or even $900 ones.



Looks like he's tapping into the same source that others have cited when stating that Meteor Lake has not-great efficiency.

Some of them do have Phoenix.

Yeah I am not trusting MLID for anything benchmark related lol.

 

[dullard]

Anyone remembers this MLID "leak"?

Intel 14th Gen Core 'Meteor Lake': 15-21% IPC boost over Raptor Lake

Intel's new 14th Gen Core 'Meteor Lake' CPUs rumored to have 15-21% more IPC performance over 13th Gen Core 'Raptor Lake' CPUs.

www.tweaktown.com

"
We are being teased with 15-21% increases in IPC performance with the new Redwood Cove architecture over the Raptor Cove architecture, but with MLID apologizing for the huge range (15-21%) but I think that's fine."

No biggie, MLID missed it will be 0-1% , just a 100% failed prediction. I posted this here just to show how this guy either :
1) has no real sources
2) is making it all up for clicks and
3) has multiple "sources" who are actually trolls and who feed him BS info all the time

I wish people would actually look more closely at MLID's key in his posts. He has two types of information on them. Some stuff from real sources (dark blue font) and some stuff he is totally making up (white font). If you look at it with that lens, you can interpret it much better. His dark blue font text on Meteor Lake is far more likely to be spot on, other than LGA 2551 (that said, his 38 mm x 46 mm size is almost correct he was just incorrect with the number of pins).

• Clock speed regressions seem true
• Crestmont is true
• VPU or Neural Engine is true
• At least 2+8 and 6+8 mobile is true
• Raptor Lake being most of the volume and Meteor Lake being premium is true.

Then just ignore the stuff in white font.

 

[Hulk]

I don’t think anyone made such a claim. Meteor Lake coming to the desktop means literally just that.

I have a small desktop PC sitting on a shelf that runs an AMD Cezanne mobile chip. The system isn’t much bigger than my wallet in terms of size. MTL-P 6+8 in one of those small boxes would be amazing to have.

If "desktop" means Meteor Lake mobile parts in small form computers then I don't count that as desktop.
I'm obviously missing something here.
My conundrum here is how do do Meteor Lake and Arrow Lake coexist on the desktop? Wouldn't they both be released at about the same time? Arrow would be built on a newer node and have a better P core. Why Meteor desktop at all?

 

[jpiniero]

If "desktop" means Meteor Lake mobile parts in small form computers then I don't count that as desktop.

Intel does though.

 

[Glo.]

If "desktop" means Meteor Lake mobile parts in small form computers then I don't count that as desktop.

It may mean that MTL will come soldered to desktop motherboards.

Its an SOC with everything on the package, already, not needing the chipset on the MoBo.