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

[Tigerick]

Wildcat Lake (WCL) Preliminary Specs

Intel Wildcat Lake (WCL) is upcoming mobile SoC replacing ADL-N. WCL consists of 2 tiles: compute tile and PCD tile. It is true single die consists of CPU, GPU and NPU that is fabbed by 18-A process. Last time I checked, PCD tile is fabbed by TSMC N6 process. They are connected through UCIe, not D2D; a first from Intel. Expecting launching in Q2/Computex 2026. In case people don't remember AlderLake-N, I have created a table below to compare the detail specs of ADL-N and WCL. Just for fun, I am throwing LNL and upcoming Mediatek D9500 SoC.

		Intel Alder Lake - N	Intel Wildcat Lake	Intel Lunar Lake		Mediatek D9500
Launch Date		Q1-2023	Q2-2026 ?	Q3-2024		Q3-2025
Model		Intel N300	?	Core Ultra 7 268V		Dimensity 9500 5G
Dies		2	2	2		1
Node		Intel 7 + ?	Intel 18-A + TSMC N6	TSMC N3B + N6		TSMC N3P
CPU		8 E-cores	2 P-core + 4 LP E-cores	4 P-core + 4 LP E-cores		C1 1+3+4
Threads		8	6	8		8
Max Clock		3.8 GHz	?	5 GHz
L3 Cache		6 MB	?	12 MB
TDP		7 W	Fanless ?	17 W		Fanless
Memory		64-bit LPDDR5-4800	64-bit LPDDR5-6800 ?	128-bit LPDDR5X-8533		64-bit LPDDR5X-10667
Size		16 GB	?	32 GB		24 GB ?
Bandwidth			~ 55 GB/s	136 GB/s		85.6 GB/s
GPU		UHD Graphics		Arc 140V		G1 Ultra
EU / Xe		32 EU	2 Xe	8 Xe		12
Max Clock		1.25 GHz		2 GHz
NPU		NA	18 TOPS	48 TOPS		100 TOPS ?

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.

 

[511]

M3 has the ST and MT win. GPU idk. We will have to see productivity benchmarks but for gaming Lunar will easily win.
288V(pre-release) https://browser.geekbench.com/v6/cpu/7102066

M3

Mac15,13 - Geekbench

Benchmark results for a Mac15,13 with an Apple M3 processor.

browser.geekbench.com

This is windows vs macOS but scores for the 288V will likely be higher in Linux.

In linux with few% advantage it can be close in MT

 

[Doug S]

it’s going to another year of flops. We’ll see how the M4 series turns out.

But I have a feeling that Lunar lake will be the best x86 client product this year.

What defines a "flop"?

Failure to meet wildly unrealistic expectations?

 

[gdansk]

What defines a "flop"?

Failure to meet wildly unrealistic expectations?

Hmm, I don't think it is wildly unrealistic to expect more than 5-10% 1T gain after 2 years and a huge process shrink.

 

[poke01]

Failure to meet wildly unrealistic expectations?

Not delivering meaningful advances.

But I can’t cross off Zen 5 yet because there is Strix Halo

 

[Philste]

That is not true. 9950X is 8% faster than 7950X

Well Computerbase tests with spec RAM, which means 5200 for ZEN4 and 5600 for ZEN5. That's already 3-4% of the 7-8% they got.

 

[DavidC1]

Unfortunately it’ll probably end up being +5% 1T and +15% 1T and be basically tied against Zen 5 and launch to a flurry of negative reviews and x86 doomer talk.

14900K is 6GHz ST so, assuming 90% clock scaling 14% better uarch results in 5.7GHz Arrowlake being 9% faster.

15% MT is realistic for average since Lion Cove loses HT and if you assume 2 E cores = 1 P core in Raptorlake generation. Maybe 20% is possible since Skymont clocks slightly higher than the "Raptormont" core. This assumes 50% average gain for Skymont with a combined Int/FP application and being 30% faster in Int and 65% faster in FP.

Worst case scenario, in MT Int it might be 6-8% faster.

What defines a "flop"?

Failure to meet wildly unrealistic expectations?

Zen 5 offering 32% may have been too much but Snapdragon Elite at least looked promising based on initial specs. It didn't live up to the projections in their own slide, so they lied or missed quite a bit.

While Apple's gains are lot less than before, they are coming from an absolute leadership over x86. So the fact that both Zen 5/Lion Cove is under historical 15-20% gain is a disappointment.

There's no particular reason to expect it should be so far behind the ARM cores either such as the embarassing ST perf/watt difference.

 

[jdubs03]

I think we should expect to see around a 12-15% increase in multi thread for ARL. It’d be nice if there was a 9% improvement in single thread, but I feel like it’s gonna be more like 2-5%.

It’s a shame because I feel like they had an opportunity to come back pretty strong against Zen 5. But if they’re gonna be able to limit the power significantly (rumor has it consuming 100W less at peak performance level), and we see that in the benchmarks, then it won’t be so disappointing.

Since it’s directly relevant to this conversation, I’ll just leave these two here as a reminder of what the expectations are for ARL. I know we’ve all seen it, but doesn’t hurt to repeat it.

 

[coercitiv]

14900K is 6GHz ST so, assuming 90% clock scaling 14% better uarch results in 5.7GHz Arrowlake being 9% faster.

I think @H433x0n 's take is still within the realm of possibility, even if leaning on the pessimistic side. The 14% figure was given, as you already know, for uplift versus Redwood Cove in Meteor Lake. AFAIK RWC showed some PPC regression even against mobile Raptor Lake.

Here's some SPECint 2017 result for example, since integer is Zen's weak point moving forward. Even if half the regression is due to different memory subsystem used on RPL system, we'd still be looking at a potential 4% penalty. Combine that with your estimate of 9% perf uplift and we get closer to the 5% mentioned by H433x0n.

 

[H433x0n]

I think @H433x0n 's take is still within the realm of possibility, even if leaning on the pessimistic side. The 14% figure was given, as you already know, for uplift versus Redwood Cove in Meteor Lake. AFAIK RWC showed some PPC regression even against mobile Raptor Lake.

Here's some SPECint 2017 result for example, since integer is Zen's weak point moving forward. Even if half the regression is due to different memory subsystem used on RPL system, we'd still be looking at a potential 4% penalty. Combine that with your estimate of 9% perf uplift and we get closer to the 5% mentioned by H433x0n.

That data seems out of date or weird. In the past month there's been multiple reviewers that had MTL ahead of Zen 4 in Specint when at a fixed 4ghz. I don’t remember seeing a similar comparison for RPC v Zen 4 but I doubt it did much better than RWC. Below is geekerwan's data, there was another reviewer that got the same results too that I'm trying to find.



David Huang got similar results when he compared RWC, Zen 4 and Zen 5 at a fixed 4.2ghz.



Point being, I'm not sure if RWC is a pure regression in IPC.

 

[coercitiv]

That data seems out of date or weird. In the past month there's been multiple reviewers that had MTL ahead of Zen 4 in Specint when at a fixed 4ghz. I don’t remember seeing a similar comparison for RPC v Zen 4 but I doubt it did much better than RWC. Below is geekerwan's data, there was another reviewer that got the same results too that I'm trying to find.

Be that as is may, I only took less than half of the regression in that table into consideration. David Huang's initial data showed just under 9% regression, I took 4%. That would put RWC 2.4% ahead of mobile Zen 4, which aligns with geekerwan's data and exceeds the advantage shown in David Huang's later Spec test with LPDDR memory for all CPUs.

Let me ask you this in return though, how did you arrive at the 5% ST advantage scenario (even as a worse case versus the possible 10% option)? Was it just "Intel will mess something up" and drop 5% in the process?

 

[TwistedAndy]

There are some new Geekbench v5 and v6 results for Arrow Lake:

COLORFUL iGame Z890 ULTRA - Geekbench

Benchmark results for a COLORFUL iGame Z890 ULTRA with an Intel Core Ultra 5 245K processor.

browser.geekbench.com

Gigabyte Technology Co., Ltd. Z890 AERO G - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z890 AERO G with an Intel Core Ultra 7 265KF processor.

browser.geekbench.com

It looks pretty nice.

 

[H433x0n]

There are some new Geekbench v5 and v6 results for Arrow Lake:

COLORFUL iGame Z890 ULTRA - Geekbench

Benchmark results for a COLORFUL iGame Z890 ULTRA with an Intel Core Ultra 5 245K processor.

browser.geekbench.com

Gigabyte Technology Co., Ltd. Z890 AERO G - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z890 AERO G with an Intel Core Ultra 7 265KF processor.

browser.geekbench.com

It looks pretty nice.

The GB5 score is good and would put it at 2500-2600 range. The GB6 score is mid but would still put it on par with Zen 5.

Have to see if there’s any improvements with more mature BIOS and microcode.

 

[511]

If user can have 10-15% Better MT 5-9.69%ST vs 14600K while consuming only 125W vs 181W it would be a nice upgrade also GB is nice but doesn't tell whole picture even GB6 ST is going to be a meme

 

[vanplayer]

Since it’s directly relevant to this conversation, I’ll just leave these two here as a reminder of what the expectations are for ARL. I know we’ve all seen it, but doesn’t hurt to repeat it.

You would like to repeat it in the future. These 1 year old slides exposed what Intel could do in the next 3-4 years, count from leaking date July 2023. Guess if Intel could survive until then.

Intel’s internal performance projection for Raptor Lake S Refresh and Arrow Lake S - How fast the CPU and iGP are expected to be | Exclusive | igor´sLAB

The Raptor Lake refresh is expected to arrive this year as the 14th CPU generation for the desktop. But the real successor, Arrow Lake-S for desktop, is also increasingly materializing along with the…

www.igorslab.de

 

[eek2121]

The GB5 score is good and would put it at 2500-2600 range. The GB6 score is mid but would still put it on par with Zen 5.

Have to see if there’s any improvements with more mature BIOS and microcode.

That is going to depend on power consumption. Leaks that came out over a year ago indicate it should be competitive with Zen 5 in terms of performance/watt, but we will see. Things change. If they actually outperform AMD in 1T/nT workloads and also manage to keep power consumption in check…shoot I might be going Intel.

EDIT: I am surprised that after remarks made by Intel, they didn’t enable SMT. Now they have a thread handicap, but as you can see, that doesn’t matter! 🤣

 

[511]

That is going to depend on power consumption. Leaks that came out over a year ago indicate it should be competitive with Zen 5 in terms of performance/watt, but we will see. Things change. If they actually outperform AMD in 1T/nT workloads and also manage to keep power consumption in check…shoot I might be going Intel.

EDIT: I am surprised that after remarks made by Intel, they didn’t enable SMT. Now they have a thread handicap, but as you can see, that doesn’t matter! 🤣

Geekebench doesn't scale with threads also we didn't knew about ChadMont

 

[Wolverine2349]

How good will latency be on Arrow Lake.

I know now on a tile instead of monolithic die, but isn't it monolithic on a tile instead of 10nm? Like

So wouldn't it still be much better than AMD chiplets because its a ring bus on the tile and all cores on ring bus and Intel designing it like that? So could latency be as good as the 10nm Alder and Raptor Lake?

 

[dullard]

There are some new Geekbench v5 and v6 results for Arrow Lake:

COLORFUL iGame Z890 ULTRA - Geekbench

Benchmark results for a COLORFUL iGame Z890 ULTRA with an Intel Core Ultra 5 245K processor.

browser.geekbench.com

Gigabyte Technology Co., Ltd. Z890 AERO G - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z890 AERO G with an Intel Core Ultra 7 265KF processor.

browser.geekbench.com

Thanks. I always need to compare them to other CPUs since I don't memorize scores on various benchmarks. Here is Videocardz' take on it (with the obvious grain of salt of not knowing necessary details of the platform like power consumption):


Going by that chart: this mid-range chip is 9% faster ST and 1% faster MT than the comparable 14700K. The single thread performance is on par with what the Intel leak claimed (posted by jdubs03 a few posts above). But the expected 6% to 19% MT performance gain isn't there yet.

 

[9949asd]

Thanks. I always need to compare them to other CPUs since I don't memorize scores on various benchmarks. Here is Videocardz' take on it (with the obvious grain of salt of not knowing necessary details of the platform like power consumption):
View attachment 105395

Going by that chart: this mid-range chip is 9% faster ST and 1% faster MT than the comparable 14700K. The single thread performance is on par with what the Intel leak claimed (posted by jdubs03 a few posts above). But the expected 6% to 19% MT performance gain isn't there yet.

GB6 only can look at st. The MT for x86 is always not right.

 

[DavidC1]

GB6 only can look at st. The MT for x86 is always not right.

Geekbench scales fine as long as you look at the Skylake era CPUs.

 

[Nothingness]

GB6 only can look at st. The MT for x86 is always not right.

I won't even waste my time debunking that. Clueless people have won.

 

[MarkPost]

Geekbench v6 results for Arrow Lake:

Gigabyte Technology Co., Ltd. Z890 AERO G - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z890 AERO G with an Intel Core Ultra 7 265KF processor.

browser.geekbench.com

Not good to say the least.

That 265KF compared with my 13900K (Intel extreme profile and DDR5 @5600):

ST: 265KF 4.3% faster than 13900K
MT: 265KF 2.5% slower than 13900K

Gigabyte Technology Co., Ltd. Z890 AERO G vs Gigabyte Technology Co., Ltd. Z690 AORUS ELITE AX - Geekbench

 

[Hitman928]

Not good to say the least.

That 265KF compared with my 13900K (Intel extreme profile and DDR5 @5600):

ST: 265KF 4.3% faster than 13900K
MT: 265KF 2.5% slower than 13900K

Gigabyte Technology Co., Ltd. Z890 AERO G vs Gigabyte Technology Co., Ltd. Z690 AORUS ELITE AX - Geekbench

ST should be 8% faster with fastest ARL (5.7 GHz). Of course, the difference will be smaller if compared to a 14900k (<= 5%).

 

[Wolverine2349]

ST should be 8% faster with fastest ARL (5.7 GHz). Of course, the difference will be smaller if compared to a 14900k (<= 5%).

Yeah looks like another underwhelming uplift.

Though its welcome because stability is needed and the 8 + 16 RPL die has stability and degradation issues that no microcode update can fix IMHO and gut feeling!!

 

[Wolverine2349]

Not good to say the least.

That 265KF compared with my 13900K (Intel extreme profile and DDR5 @5600):

ST: 265KF 4.3% faster than 13900K
MT: 265KF 2.5% slower than 13900K

Gigabyte Technology Co., Ltd. Z890 AERO G vs Gigabyte Technology Co., Ltd. Z690 AORUS ELITE AX - Geekbench

Not good. Though if latency is still good, its a welcome uplift as Raptor Lake was good if only it were stable and did not degrade so easily.

SO Arrow Lake fixes that.

Zen 4 is stable so Zen 5 underwhelming or no performance improvement in general consumer workloads and gaming is bad because Zen 4 is stable and does not have degradation issues.

Though hard to believe MT would regress. I know Arrow Lake has no HT but isn't the e-cores Skymont supposed to be so much better than Gracemont that it should easily more than make up for it given that HT is not a real core. Like isn't Skymont supposed to have IPC and latency of Raptor Cove or maybe 2% better or is that just wishful thinking?