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.)

 

[FlameTail]

Let the hype begin.

Wake me up when Intel catches upto Apple in efficiency

 

[Goop_reformed]

Pat Gelsinger 18 Month ago:

https://twitter.com/x/status/1552777744718651392

Intel can't be trusted.

Remember this is the core they have in Granite, Intel is F-CKED in CAPITAL, server wise

 

[coercitiv]

It's not even hype to expect RWC to improve perf/watt over RPC, it's common sense. It's on a new node, it definitely should bring PPW improvements.

That's a given, but remember the expectations were set by the jump in base clocks, leading us to believe that most of the claims about Intel 4 would transfer nicely into the new product... and then some. That's how we got to the hype @Kepler_L2 just mentioned above.

Anyway, it's still unclear to me what is happening, the noise floor around this launch is too high for the little time I have to read on the subject, but things will clear up soon. Maybe Chips and Cheese will give us a nice Christmas present this year

 

[FlameTail]

Geekerwan should test this.

 

[TESKATLIPOKA]

There it is again, the usual suspect(s).

This is MTL right now:

Expensive packaging technology: Check
New cores: Check
New node: Check
New branding: Check
New graphics tiles: Check


The end result? Can't even beat the competitor released 3Qs before, can't even beat their own previous cores. Also paper launch.

Any posters defending this type of behavior are objectively lying to themselves. Quite a sad launch, and does not bode well for arrow lake, which also is projected to have minor performance uplift over a 2(!!) year olds part.

If you can buy a MTL product then It's not a paper launch.
And you conveniently forgot to mention that Phoenix launch 3Q ago was a spectacular paper launch, you needed to wait a very long time to finally have some with a dGPU and without It, you had to wait even longer.

 

[TESKATLIPOKA]

Wake me up when Intel catches upto Apple in efficiency

Which gen of Apple product you mean?
Do you really want to sleep that long?

 

[SmokSmog]

Computerbase review has more technical content, FI a single core at 4.6GHz use 25W and 30W at 4.8GHz

Manually tuned ADL 4.6ghz 1.05Vcore
20W~ single core during R23 ST test.

 

[SpudLobby]

Uh, I think people don't like linking twitter stuff here, so here:
View attachment 89074
View attachment 89075
View attachment 89076
View attachment 89077View attachment 89078

I’m late on this, but matching the M1 GPU’s Perf/Watt at 12W — yet not mentioning the M2 at that power and only exceeding the M2 GPU at LNL’s peak is interesting. Basically suggests they’re gunning the GPU a bit and max power draw from the GPU is probably going to end up a bit higher than Apple (obviously those 17-30W TDP figures don’t represent what the GPU and CPU could run combined at peak).

And the M3 GPU can match an M1’s ~ 12W performance at half the power. No one games on a Mac really but this is still telling because the CPU efficiency gaps might still be there too from a workload MT perspective (they will albeit depends on the point in the power curve, realistically Intel is hoping for improvement + low idle).

Lunar Lake’s 10-17 +- CPU performance (measured properly from the wall) will be a real improvement over any existing AMD or Intel system of the time I think, somewhat competitive with an M1 or M2 and idle unmatched elsewhere for Intel/AMD, but I think expectations are kind of laughably high.

 

[SiliconFly]

Expensive packaging technology: Check
New cores: Check (correction: Same Old Cores)
New node: Check (correction: Not a very good node)
New branding: Check
New graphics tiles: Check

And ARL is going head to head with Apple in efficiency. Performance wise, no one knows about LNC (except for an untrustworthy leak).

 

[SiliconFly]

Pat Gelsinger 18 Month ago:

https://twitter.com/x/status/1552777744718651392

Intel can't be trusted.

Intel has met all goals that the tweet says. Whats there not to trust?

 

[Tarkin77]

Intel has met all goals that the tweet says. Whats there not to trust?

Unless "as good as 3nm from others" was meant as a joke, than yeah, you are right.

 

[Geddagod]

Did anandtech forums migrate to reddit and no one noticed?

Lol, thread.

There it is again, the usual suspect(s).

idk what you want people to say, I haven't seen a comment recently that was overwhelmingly positive abt Intel, except maybe 1? 2 if I stretch it?
Some people just love over reacting to products in general. If something is good, they will think it's amazing. If something is bad, like MTL, they will think it's terrible or a bulldozer.

Expensive packaging technology: Check

Which hurts power? Like any packaging tech does at low power to varying degrees?

New cores: Check

Literally just RPC with minor tweaks

New node: Check

Ye they f'ed up this

New branding: Check

What does this have to do with perf/power lol

New graphics tiles: Check

don't know why you are whining abt this, the iGPU was decent...

 

[SiliconFly]

Wake me up when Intel catches upto Apple in efficiency

9 more months

 

[Geddagod]

Unless "as good as 3nm from others" was ment as a joke, than yeah, you are right.

It's hard to argue with the math, HP libs on Intel 4 have the same density as TSMC 3. SRAM density is poop and perf/watt is almost certainly worse, but it's marketing for the vast majority of it, not an industry paper Intel is releasing at IEEE, so of course Pat is going to fixate on the metric he knows he can market.

But also, duh, never trust marketing. This was in a Q/A in a financial earnings call, and Pat is deff gonna "stretch" the truth to make Intel as good as it can seem. But on the flip side, I doubt you are legally allowed to just straight up lie to investors on hard numbers (such as Intel 7 shipping 35M units), while stuff like "MTL is doing good" is much more subjective.

 

[SiliconFly]

Unless "as good as 3nm from others" was meant as a joke, than yeah, you are right.

About Intel 4, a couple of members already mentioned that it's the cores and not the node!

In reality, Intel 4 is closer to TSMC N3 than N4.

And Intel 3 is expected to be slightly better than TSMC N3.

 

[Geddagod]

And Intel 3 is expected to be slightly better than TSMC N3.

Intel themselves don't think that.

 

[TESKATLIPOKA]

9 more months

And he can continue sleeping.
Seriously, why do you keep continuing this hype?
You were hyping MTL, didn't meet expectations and now you do the same with ARL.

 

[IEC]

There's some weird cherry-picking going on. The Raptor Lake part is +9% vs the MTL part even though it's only 4% faster in clocks. Doesn't that suggest an IPC regression versus Raptor Lake?

 

[FlameTail]

Can someone explain what this means? Only the 32 GB LNL SKU has greater performance than M2?

 

[jpiniero]

There's some weird cherry-picking going on. The Raptor Lake part is +9% vs the MTL part even though it's only 4% faster in clocks. Doesn't that suggest an IPC regression versus Raptor Lake?

I would say it's more likely that the Meteor Lake is not holding the 1T clock because of the loose binning. Although that is the 165H and not the 155H that people will actually buy.

 

[SiliconFly]

And he can continue sleeping.
Seriously, why do you keep continuing this hype?
You were hyping MTL, didn't meet expectations and now you do the same with ARL.

Me hyping? I think you are very confused. MTL has met all my expectations. I very clearly said MTL's performance will be similar to that of RPL (even a bit less in certain workloads) & it's efficiency will be a lot higher compared to RPL in very light loads due to LP E cores. Everything what I said has come true. Kindly read my words clearly before accusing me of hyping.

 

[SiliconFly]

And about the "9 more months" statement... I stand by my words.

20A is expected to bring in 35% to 40% PPW gains over the now dated Intel 4. And LNC is touted to be a power efficient design compared to RWC, which should bring in a minimum of 10% to 20% efficiency gains against RWC even in worst case scenarios. Best case, it might be a lot higher. Otherwise, intel wouldn't even consider ditching RWC and switching to LNC in the first place (LNC is more about efficiency than performance).

However we look at it, we can expect "a minimum of 50% PPW gain for ARL's CPU tile over MTL's CPU tile". This is my statement. ARL's CPU tile is gonna be hyper-efficient compared to MTL. No doubts there. Kindly do not misinterpret it or misunderstand it.

 

[tamz_msc]

So is your argument that Intel isn't being deceptive, just that they are purposefully crippling the competition to show that they have higher performance?

It's not Intel's problem that Phoenix equipped with 6400 MT/s memory.

 

[tamz_msc]

Oh, and before someone claims deceptive marketing because of the use of Intel's compiler, look at what the actual differences are between compilers on AMD CPUs:

https://icis.pcz.pl/~lszustak/sonata13/JCR_2022_9_ADES_Compilers.pdf

Emphasis mine:

The compiler’s influence on the execution time of simulations is
significantly more essential for the application versions with the dy-
namic intensity. The clear winner is now the Intel compiler, which on
AMD Rome permits improving the performance up to about 1.3 times
compared to the AMD AOCC compiler – the native software for EPYC
Rome and Milan processors. The PGI compiler is on a similar perfor-
mance level as AOCC, while GCC produces codes that are up to 0.86
times slower than AOCC. It should be added that the simulation results
achieved for GCC, Clang, PGI, and Intel compilers are in a good match
with those of the AOCC compiler. For AMD Milan, the performance
advantages of the Intel Compiler over AOCC is even more essential.

 

[maddie]

Me hyping? I think you are very confused. MTL has met all my expectations. I very clearly said MTL's performance will be similar to that of RPL (even a bit less in certain workloads) & it's efficiency will be a lot higher compared to RPL in very light loads due to LP E cores. Everything what I said has come true. Kindly read my words clearly before accusing me of hyping.

In your own words: With your AMD bias, I can tell you whats going to happen to your favourite brand starting December 14th.

Intel, for the last 5 years, have been releasing outdated products built on outdated nodes based on outdated and power-hungry core designs. Something that even I find a bit despicable. But in spite of all these severe shortcomings, they managed 88% desktop & laptop CPU market share consistently leaving AMD in dirt picking up scraps. And the credit goes to Intel marketing. They're one of the best in the industry.

Now Intel has given these sharks with an amazing product they've been waiting for all these years. A hyper power-efficient CPU thats far more advanced and a true engineering marvel compared to AMD's old and outdated chiplet design. Not only they're just going to simply outsell AMD like always, but this time it's going to a massacre and it's going to be very brutal.

AMD's time has come. Live with it. And I think it's time to start praying so that Intel has some decent competition in the future for our own good. 😧

Did you write this a few days ago?

Did you write this a few days ago?