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.

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	12 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)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop Only	Desktop & Mobile H&HX	Mobile U Only	Mobile H
Process Node	Intel 4	Intel 20A	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Q1 2025 ?	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	6P + 8E ?	8P + 16E	4P + 4E	4P + 8E
LLC	24 MB	24 MB ?	36 MB ?	12 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.)

SiliconFly · Sep 8, 2024

jdubs03 said:
On single thread, the 285K will be ahead. Looking at computerbase, the 14900K was 9% ahead of the 7950X. I don’t think it’ll be ahead by nearly that much. I said 0-3% before, I think it’ll be around that 3%. In multithread the 7950X was level with the 14900K. I think the 285K will at least be equal again.

3%?

Compared to AMD HX 370, Lunar Lake 288V is already ahead by around ~15% in ST perf I think.

Same should translate well for 9950X vs 285K in ST perf. The difference should be in excess of >10% to 15% easily. My guess is ~15%.

It's a common trend I'm seeing with many people here. They're still not able to accept the fact the Zen 5 doesn't have a good IPC uplift this gen and has fallen below expectations. It has underperformed.

Whereas, Intel this gen has shown good performance uplift thereby widening the gap more than it should.

Meaning, the perf difference between previous gen and this gen is gonna be more. A lot more!

Abwx · Sep 8, 2024

cebri1 said:
Talk about cherry picking

Using Cinebench as only metric is the same as using V-Ray as only metric.

I would think that 4 renderers are a better test than using one of the 2 outliers in this set.

jdubs03 · Sep 8, 2024

igor_kavinski said:
Intel's benefit this time is that they can count on real physical cores and not be hampered by HT related issues dragging down performance. I think you should brace yourself for 9950X losing in some benchmarks because SMT cannot match a real core. I just have to call it like it is. Intel has a slight advantage this time around. Well, until AMD boosts 9950X performance with an AGESA update or something. I would've preferred 9950X having CCD2 comprised of 16 Zen5c cores.

I don’t know why my post was ignored (not by you obv). It was very even handed. I’m not playing favorites.

Abwx said:
Even in single core, we ll see in a month when they release ARL but given the gap between the 9950X and the 14900K i doubt that it can be closed given the improvement in CB, as with other renderers it will start from a lower step.

FTR the former is 21% faster in Blender, 30% faster in Corona, 51% faster in V-Ray and only 5% faster in CB 2024, wich by those number is the convenient exception used as a rule, the geomean average of those 4 numbers, 25.6%, is way more representative than an outlier.

As you can see it require a much better ST perf to get those numbers given that SMT bring a comparable gain for all tests.

In single thread you’re basically claiming the 9950X is going to be even to ahead of the 285K (pulling back that 9%) when all of the data out there so far says this isn’t going to be the case.

SiliconFly said:
3%?

Compared to AMD HX 370, Lunar Lake 288V is already ahead by around ~15% in ST perf I think.

Same should translate well for 9950X vs 285K in ST perf. The difference should be in excess of >10% to 15% easily.

Yeah I’m trying to be conservative. In CB my guess is it will be 8–9% higher, but I think the other benchmarks will be lower single digits.

The 288V won’t be much higher (maybe a few %) in GB I don’t think just going based on their releases so far. But for CB it’s somewhere around 12%. Time will tell to fully confirm those and other benchmarks.

(I think AMDs desktop core is more competitive than their mobile in ST)

511 · Sep 8, 2024

Abwx said:
Even in single core, we ll see in a month when they release ARL but given the gap between the 9950X and the 14900K i doubt that it can be closed given the improvement in CB, as with other renderers it will start from a lower step.

Single core avx-512 will be the only win for 9950X vs 285K

Abwx said:
FTR the former is 21% faster in Blender, 30% faster in Corona, 51% faster in V-Ray and only 5% faster in CB 2024, wich by those number is the convenient exception used as a rule, the geomean average of those 4 numbers, 25.6%, is way more representative than an outlier.

Corona/vray uses intel embree which has avx-512 support so there is that as well blender doesn't support it and we are expecting 15-20% mt improvement so blender would be within 1-2%

AVX-512

forum.corona-renderer.com

Abwx · Sep 8, 2024

jdubs03 said:
In single thread you’re basically claiming the 9950X is going to be even to ahead of the 285K (pulling back that 9%) when all of the data out there so far says this isn’t going to be the case.

What are all thoses tests, Cinebench.?.

In the GB ST submissions ARL is not ahead even with 7200 RAM, from the few leaks it s neither ahead in WebXPRT4 wich is tailored by Intel and is single threaded, nor in Speedometer 2, only in CPU Z, lol.

511 · Sep 8, 2024

As a fact i would like to say one thing if yhe particular test uses AVX-512 Zen 5 wins else intel will win in ST

Abwx · Sep 8, 2024

511 said:
Single core avx-512 will be the only win for 9950X vs 285K

Corona/vray uses intel embree which has avx-512 support so there is that as well blender doesn't support it and we are expecting 15-20% mt improvement so blender would be within 1-2%

AVX-512

AVX-512

forum.corona-renderer.com

Cinebench R15/R20/R23 all use Embree as well.

jdubs03 · Sep 8, 2024

Abwx said:
What are all thoses tests, Cinebench.?.

In the GB ST submissions ARL is not ahead even with 7200 RAM, from the few leaks it s neither ahead in WebXPRT4 wich is tailored by Intel and is single threaded, nor in Speedometer 2, only in CPU Z, lol.

No it’s this:

You got one leak for at best a QS 2-3 months away from a final product. Both Zen 5 and Lion Cove are around 15pct higher in IPC and both have a clock speed reduction of 5%. I don’t see how Zen 5 claws back a 9% gap with those numbers.

511 · Sep 8, 2024

https://github.com/RenderKit/embree

Looks like it but maybe it depends on the kernel used anyway these are MT benchmark where is the ST one maybe grace mont is dragging it down we have like a month to go before launch

Josh128 · Sep 8, 2024

9949asd said:
it’s 220w when running r23, but run fpu will be 250w. Ecore is 4.6.

Source?

Abwx · Sep 8, 2024

jdubs03 said:
No it’s this:
View attachment 107061
You got one leak for at best a QS 2-3 months away from a final product. Both Zen 5 and Lion Cove are around 15pct higher in IPC and both have a clock speed reduction of 5%. I don’t see how Zen 5 claws back a 9% gap with those numbers.

The 9950X has on average 11.57% better IPC than the 14900K in this graph, wich is duplicated by GB 6 wich is exactly on the average, and so far there s GB submissions where ARL is no better than the 9950X in ST.

Other leaks for WebXPRT, Speedometer or even Cinebench just go in the same direction given that the tests frequencies or power were stated, it s not like it was some hasardous ES clocking randomly.

igor_kavinski · Sep 8, 2024

Abwx said:
so far there s GB submissions where ARL is no better than the 9950X in ST.

What's the frequency for ARL in these submissions?

jdubs03 · Sep 8, 2024

Abwx said:
The 9950X has on average 11.57% better IPC than the 14900K in this graph, wich is duplicated by GB 6 wich is exactly on the average, and so far there s GB submissions where ARL is no better than the 9950X in ST.

Other leaks for WebXPRT, Speedometer or even Cinebench just go in the same direction given that the tests frequencies or power were stated, it s not like it was some hasardous ES clocking randomly.

I’m just saying; going by the information provided by each company. The numbers align with Intel being slightly ahead on single thread. For the prior generation Zen 4 was quite a bit behind, I’m acknowledging that Zen 5 has bridged some of the gap, but I don’t think they’ll bridge all of it. I think that is too much of a stretch.

Abwx · Sep 8, 2024

igor_kavinski said:
What's the frequency for ARL in these submissions?

At GB these are the regular frequencies, other than this there were leaks of scores at stock and at a given power, FI 43118 CB R23 pts at stock and 32 200 at 120W, those two numbers say that at stock it s at least 250W.

cebri1 · Sep 8, 2024

According to Guru3D, Z5 has about an 10-11% IPC advantage over RPC.
LNC mobile has 14% IPC advantage over RWC, which has a similar IPC to RPC.

It's really a matter of clocks. In ST if LNC desktop is anywhere near 16% IPC over RPC, and can indeed clock to 5.7 Ghz it should take the ST crown. MT is a different story because of the lack of HT, so chadmont will need to show its value.

At the end, both processors will have its weaknesses and advantages and the consumer will be able to choose the one that fits more its needs.

igor_kavinski · Sep 8, 2024

cebri1 said:
MT is a different story because of the lack of HT, so chadmont will need to show its value.

Skymont has a nickname? Or is the desktop mont different in some way to LNL's mont?

Josh128 · Sep 8, 2024

cebri1 said:
According to Guru3D, Z5 has about better 10-11% IPC over RPC.
LNC mobile has 14% IPC advantage over RWC, which has a similar IPC to RPC.

It's really a matter of clocks. In ST if LNC desktop is anywhere near 16% IPC over RPC, and can indeed clock to 5.7 Ghz it should take the ST crown. MT is a different story because of the lack of HT, so chadmont will need to show its value.

At the end, both processors will have its weaknesses and advantages and the consumer will be able to choose the one that fits more its needs.

People have to remember that the removal of HT only applies to the P cores, the E cores never had it to begin with-- its not that much of a negative. So, assuming +25% average MT on the Raptor Cove P cores in MT w/ HT, Lion Cove 's 8 cores, with their +14% IPC uplift, will see a net 11% reduction in MT for P core cluster. Meanwhile, the 16 Skymont cores, with ~+38% IPC over Crestmont.

P-cores only get about 22.5K in R23 at stock (correct me if Im wrong). Take 11% from that and you get ~20K. E-cores add about 18K for a total of ~40K. Take your new 20K for non-HT Lion Cove cores and add to it 24.85K (1.38 x 18K) and you get 44.85K. The biggest assumption here is that the +14% IPC for Lion Cove and +38% for Skymont figures given apply directly to CB R23. If they do, you are going to end up around 44.9K in R23 MT. Simple as that. Thats +7.4% stock vs stock 9950X.

I think the above is actually a bit conservative. My prediction was ~+13% MT average over 9950X at stock, and I still stand by that. So the R23 score could easily end up around 47K stock for 285K.

DokiDoki · Sep 8, 2024

igor_kavinski said:
Skymont has a nickname? Or is the desktop mont different in some way to LNL's mont?

According to rumors, the core layout in ARL die is different from the RPL die. It's no longer one long row of P-cores followed by the E-cores.

That, and Skymont in ARL has access to 36 MB LLC instead of only 8 MB SLC on Lunar Lake.

fastandfurious6 · Sep 8, 2024

FlameTail said:
Is it?
View attachment 107060

I wonder if CPU architects have tracing/monitoring granular enough to see what's going on step-by-step in that depth

FlameTail · Sep 8, 2024

DokiDoki said:
That, and Skymont in ARL has access to 36 MB LLC

Wow. This is for ARL-H I believe?

DokiDoki · Sep 8, 2024

FlameTail said:
Wow. This is for ARL-H I believe?

No, ARL-S.

Ghostsonplanets · Sep 8, 2024

FlameTail said:
What laptop price segment would Wildcat Lake belong to?

ADL-N spiritual successor. So <$500 - 400 laptops.

2P + 0E + 4LPE Cougar Cove + Darkmont
2 Xe³ Cores (256 ALUs)
20 TOPs NPU (?)
64-bit memory bus (?)

naukkis · Sep 8, 2024

Performance is ipc*clocks. Arl will also clock it's e-cores higher than rpl. +100mhz from 14900ks and 200 from k. P-cores clock will also vary by amount of power budget left - at sane power limits arl should clock its p-cores higher too. So there might be good amount of mt power coming - if they can sustain those clocks.

FlameTail · Sep 8, 2024

Ghostsonplanets said:
ADL-N spiritual successor. So <$500 - 400 laptops.

2P + 0E + 4LPE Cougar Cove + Darkmont
2 Xe³ Cores (256 ALUs)
20 TOPs NPU (?)
64-bit memory bus (?)

If 20 TOPS NPU is true, that means it won't even support Copilot+ 1st gen (minimum 40 TOPS).

And by that time in 2026, Microsoft would have unveiled Copilot+ 2nd gen (minimum 80 TOPS?).

Does Microsoft not want to bring Copilot+ to the sub-$500 laptop segment?

cebri1 · Sep 8, 2024

FlameTail said:
If 20 TOPS NPU is true, that means it won't even support Copilot+ 1st gen (minimum 40 TOPS).

And by that time in 2026, Microsoft would have unveiled Copilot+ 2nd gen (minimum 80 TOPS?).

Does Microsoft not want to bring Copilot+ to the sub-$500 laptop segment?

It's a chrome book, you also need you differentiate your devices. If the consumer wants CoPilot+ they'll need to pay more.

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

Senior member

Attachments

Golden Member

Lifer

Senior member

Senior member

Lifer

Senior member

Lifer

Senior member

Senior member

Senior member

Lifer

Lifer

Senior member

Lifer

Senior member

Lifer

Senior member

Member

Member

Diamond Member

Member

Senior member

Senior member

Diamond Member

Senior member