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.

 

[SpudLobby]

None of that is good lol.

SMT doesn't cost any die area (more or less).
It's just more validation time.
Intel only killed it because Apple has no SMT.
That is the only reason.

Pretty good to me, mobile matters much more than silly 6GHz DIY stuff.

It does in fact cost die area, it's just minimal.

Also, depending on how the core is designed, if they aren't especially front end limited with LNC they may not benefit from HT as much.

 

[SpudLobby]

Apple introduced excellent power gating in the A13 chip.

We will have see how Lion cove and skymont are like.

This is an interesting point about power gating and the microarchitecture. I know Intel and AMD's gating isn't as extensive as Apple or most Arm cores, but didn't realize they introduced (more) fine power gating with the A13 in particular. Any links?

 

[adroc_thurston]

Well there's still PHX2e/Bandit Crest, unless that has been cancelled recently.

Yea I know (and wow someone actually named the next AMD 9W part), but that's beyond niche and idk if they'll actually make it a product (skinny iGP so idk if they'll fit Deck2 with it).
Chromebooks are totaled and PHX2 is a part that's a wee bit too expensive for them anyway.

Pretty good to me, mobile matters much more than silly 6GHz DIY stuff.

Server, brother.
SMT is paramount in server (outside of scale-out spam favelas).

I know Intel and AMD's gating isn't as extensive as Apple or most Arm cores

Core-wise they aren't really ever different, just that Apple has a more efficient fabric (but it sucks at scaling up, RIP 4t Mac Pro dreams).

but didn't realize they introduced (more) fine power gating with the A13 in particular

That's because they didn't.
A13 was nothing special.

 

[SpudLobby]

There is link above the chart showing leaks from Twitter about that and I made some modifications..

Yeah there is 0 mention of a 256-bit bus from Qualcomm anywhere, that modification is basically entirely fan fiction or speculation man. There are multiple leakers on this topic and every single one states 128-bit bus for all the parts or does not specify that there is any difference. I doubt there will be a 256-bit part from QC with this iteration based on AMD/Intel currently, their target market and the fact that the GPU, while solid isn't really M-Pro/Max class, just higher clocked Adreno 740 (so probably better than an M2 iGPU by a bit).

That Adreno 740 on the 8cx Gen 4 would still benefit from a larger bus width even for some MT CPU workloads by a smidge (if 8+4) but ultimately it would come down to the iGPU and I doubt it's justifiable.

I would love to be wrong.

 

[SpudLobby]

Yea I know, but that's beyond niche and idk if they'll actually make it a product (skinny iGP so idk if they'll fit Deck2 with it).
Chromebooks are totaled and PHX2 is a part that's a wee bit too expensive for them anyway.

Server, brother.
SMT is paramount in server (outside of scale-out spam favelas).

Core-wise they aren't really ever different, just that Apple has a more efficient fabric (but it sucks at scaling up, RIP 4t Mac Pro dreams).

That's because they didn't.
A13 was nothing special.

I am assuming they're keeping SMT for server, which makes sense given that's where those workloads actually matter, yes outside of GMT/4C/Graviton-style stuff.

Don't disagree there.

RE: Core wise: GLC is closer to Apple's core than Zen 4, but that's ridiculous, Apple have roughly a 30% perf/GHz lead on Intel's Golden Cove even at moderate clocks e.g. comparing at 3.5GHz and I doubt GLC scales down particularly well like Apple can. The core is also bloated due to HP spam, not just Intel 7. Let me know when the fabric explains that, and I'm sure Meteor Lake will still gun it to 15-20W for what Apple or probably QC and Nuvia perform @ 4-5W.

RE: A13, it was a significant boost on the same process. Apple could've launched an M chip with 4 Lightning cores and 4 thunder cores and it'd still blow a 5600/5800U out of the water on energy efficiency with perfectly fine performance (and unplugged...).

Mac Pro sucks. I don't see why anyone really cares. They gave that up once and for all at this point with the transition to custom SoC's and dropping PCIE GPU's. Would much rather have an HEDT PC.

 

[adroc_thurston]

I am assuming they're keeping SMT for server

But the core has no SMT!
lol.

GLC is closer to Apple's core than Zen 4

Yes Intel is no good at making good CPU cores lol, that's not news.

The core is also bloated due to HP spam, not just Intel 7. Let me know when the fabric explains that.

Yes but that's active power (Apple isn't anything special as far as MT perf on N5 class node goes).
We're talking idle and media aka the real reason Apple parts are so efficient.

A13, it was a significant boost on the same process.

It was like 13% IPC and a PPW regression over Vortex, lol.
a 2.66GHz Lightning was already a 5W core in specINT give or take.

and it'd still blow Zen 3 out of the water on energy efficiency.

Ehh at idle yea, in MT loads nah.

Mac Pro sucks

That's because Apple failed.
It happens.
An actual 4T part would've been interesting but all they've managed it 2t so it's just a Mac Studio in a big useless chassis.

 

[poke01]

A13 was nothing special

Start at 3:19

 

[adroc_thurston]

Start at 3:19

That's nondescript Apple marketing.
I'm talking real, measured results done by Andrei here, years ago.
A13 was kinda bad (fat SLC aside).

While the performance figures are quite straightforward and not revealing anything surprising, the power and efficiency figures on the other hand are extremely unexpected. In virtually all of the SPECint2006 tests, Apple has gone and increased the peak power draw of the A13 SoC; and so in many cases we’re almost 1W above the A12. Here at peak performance it seems the power increase was greater than the performance increase, and that’s why in almost all workloads the A13 ends up as less efficient than the A12.

The total power use is quite alarming here, as we’re exceeding 5W for many workloads. In 470.lbm the chip went even higher, averaging 6.27W. If I had not been actively cooling the phone and purposefully attempting it not to throttle, it would be impossible for the chip to maintain this performance for prolonged periods.

 

[SpudLobby]

But the core has no SMT!
lol.

Yes Intel is no good at making good CPU cores lol, that's not news.

Yes but that's active power (Apple isn't anything special as far as MT perf on N5 class node goes).
We're talking idle and media aka the real reason Apple parts are so efficient.

It was like 13% IPC and a PPW regression over Vortex, lol.
a 2.66GHz Lightning was already a 5W core in specINT give or take.

Ehh at idle yea, in MT loads nah.

That's because Apple failed.
It happens.

I could see them keeping a separate one for servers. It's Intel. You really think they would do that, just increase costs or do something insane like that? Who knows though, only saw it mentioned in context of ARL and LNL. If they don't, you are correct it will hurt, but I think you probably have a financial conflict of interest so I don't expect a reasoned assessment there.

Yes but that's active power (Apple isn't anything special as far as MT perf on N5 class node goes).
We're talking idle and media aka the real reason Apple parts are so efficient.

Idle is huge yeah. Efficient but performant ST is pretty beneficial - you can get a good bit of extra perf over Intel/AMD restricted to similar power levels. There's also concurrent management of lower priority tasks siphoned to more efficient (and not in the Intel way) cores which is a huge benefit in mixed use. Those E Cores for background tasks take anything AMD and Intel have to the woodshed on energy efficiency. That adds up. Not everyone logs on for a 15-20W fully threaded Cinebench R23 run until the battery kills over which has become a common meme point of comparison to M chips. (Though in that case the 5800U would probably perform better at those power levels, and in a full MT performance test the 5800U would likely win vs this hypothetical A13-based chip).

It was like 13% IPC and a PPW regression over Vortex, lol.
a 2.66GHz Lightning was already a 5W core in specINT give or take.

13% IPC for an annual upgrade while having the industry's lead on perf/GHz is pretty decent, barely a process change. A 6.5% frequency improvement to boot.

On the *peak* PPW regression: Probably because they pushed the frequency a bit higher vs the A12, these slopes are steep, and the microarchitecture changed. The chip stayed within 5W or so, that's a great tradeoff for an annual upgrade whilst keeping the power around 5W. It's not like they deviated from their general power range trends or that the absolute power is nuts.

The most likely case is that the core is 30% (probably highballed but still) more efficient iso-performance relative to an A12, as Apple's material suggested and Andrei's review which you are referencing suggested this likely based on his experience with the A12's v/f + W curve.

From Andrei:

One possible explanation for the quite shocking power figures is that for the A13, Apple is riding the far end of the frequency/voltage curve at the peak frequencies of the new Lightning cores. In the above graph we have an estimated power curve for last year’s A12 – here we can see that Apple is very conservative with voltage up until to the last few hundred MHz. It’s possible that for the A13 Apple was even more aggressive in the later frequency states.

The good news about such a hypothesis is that the A13, on average and in daily workloads, should be operating at significantly more efficient operating points. Apple’s marketing materials describe the A13 as being 20% faster along with also stating that it uses 30% less power than the A12, which unfortunately is phrased in a deceiving (or at least unclear) manner. While we suspect that a lot of people will interpret it to mean that A13 is 20% faster while simultaneously using 30% less power, it’s actually either one or the other. In effect what this means is that at the performance point equivalent to the peak performance of the A12, the A13 would use 30% less power. Given the steepness of Apple’s power curves, I can easily imagine this to be accurate.

Obviously, that 20% and 30% are mutually exclusive contra Apple's marketing, but still.

Anyway that 5W performance was pretty solid. A 5800U's ST gets cut down significantly if we're taking it down to Apple's power or freq levels instead of manically boosting above to 10-15W for the APU. At 4.45GHz AMD were at what, 1400-1450 GB5? Lol. Okay, so you drop it down to a low TDP for ST but you're almost certainly below Apple's perf at that point.

RE: A13 Perf/W and power gating claim: I'd agree there probably wasn't any power gating magic introduced on the A13.

That's because Apple failed.
It happens.

Sure does. They're absolutely awful with dumb shit like the Mac Pro or their whole Fisher Price OS, lightning ports, blah blah. Just one of the worst firms out there frankly. Which is why it sucks they build better mobile hardware than others do, for now.

Interestingly, Andrei F. seems fairly disappointed in their execution with Zen 4 in Phoenix or on the desktop. ST power not looking great. Maybe AMD will deal a more competent blow with Zen 5's mobile lineup.

 

[SpudLobby]

Why would anyone leave premium cores out?
lol
Compacts are for area eff MT grunt and not much else.

Boosting to upper 4GHz to lower 5GHz+ is so monkeybrained as to be stupid on a laptop and those designs are leakier. Cut the tumor right out, give me something I'll actually use.

 

[adroc_thurston]

I could see them keeping a separate one for servers. It's Intel

They're struggling with validation as is.
Don't think having an even more separate server core variant is quite a good idea there.

13% IPC for an annual upgrade while having the industry's lead on perf/GHz is pretty decent, barely a process change.

Yea but that was the last one in years.
Firestorm was like high single digits and then poof, party's over.

A 5800U's ST gets cut down significantly if we're taking it down to Apple's power or freq levels instead of manically boosting above to 10-15W for the APU

Yea but Cezanne had poverty caches versus what Apple did on their parts.

Interestingly, Andrei F. seems fairly disappointed in their execution with Zen 4 in Phoenix or on the desktop.

Andrei is in no position to talk stuff about anyone given how anemic Nuvia Phoenix is turning out to be.

 

[adroc_thurston]

Maybe AMD will deal a more competent blow with Zen 5's mobile lineup

Yea Zen5 is the big tuna kahuna that will roadkill a whole lot of competing designs.
Mikey C. wasn't joking about sleeping and dreaming about the thing being a real product.

Hopefully Panther Lake core is a meaner peak 1t design than LNC, otherwise lol.

Boosting to upper 4GHz to lower 5GHz+ is so monkeybrained as to be stupid on a laptop and those designs are leakier

Peak 1t matters in race to idle scenarios, sorry.
The bulk (8 out of 12) cores on STX1 is compacts anyway.

plus the cool factor of making THAT thing run at IBM z16 speeds.

 

[SpudLobby]

They're struggling with validation as is.
Don't think having an even more separate server core variant is quite a good idea there.

Yea but that was the last one in years.
Firestorm was like high single digits and then poof, party's over.

Yea but Cezanne had poverty caches versus what Apple did on their parts.

Andrei is in no position to talk stuff about anyone given how anemic Nuvia Phoenix is turning out to be.

They're struggling with validation as is.
Don't think having an even more separate server core variant is quite a good idea there.

Haha. When has reason held any sway over Intel in the last 5 years? I'm all for the chaos. Lately we're getting a bit of it. So if your AMD shares and cherished rep among AMD groupies takes a bit of a knick because Intel is half alive as opposed to this vegetative flailing right now, that's great. If people like SteaknPork (or whoever, you know him), or the Singaporean with a US Zumwalt-class destroyer pfp strap in and watch Intel blow every last market and sulk in their overconfidence, that's great too.

At any rate I have no idea what they'll do for servers. I'd agree it's stupid - I didn't endorse it - to go for a separate designs given the nimrods in their design teams. And yes SMT is valuable there. Validation part is obvious. See SPR.

Yea but that was the last one in years.
Firestorm was like high single digits and then poof, party's over.

Obviously. But let's wait to see someone catch them - or just get within 10% on perf/GHz, and similar margins at 4-5W. Until then I'm not super interested in clowning them on this particular part until there's an alternative. Mac Pro is effing [redacted] yes, no one is buying an Apple desktop other than creatives or Tik Tokers with too much money.

Yea but Cezanne had poverty caches versus what Apple did on their parts.

Yes and that counts for something from a functional perspective. And it still sucks today AMD still won't throw 32MB of L3 on Zen 4 mobile given caching's influence on DRAM and system active power. "lol BOM" yep, sucks! Either way that's regressing on an intrinsic quality, business choice (muh merchant economics) or microarch-related or both, that's part of it. At least Zen 4's L2 is doubled to 1MB all around though, which helps. Obviously if AMD wanted to conserve extra power they could do it with extra cache AKA a bigger mm^2 budget even on existing parts. You know this. But keep in mind consumers don't care why. But they know what is. Deep thonks.

I'd hazard a guess Zen 5 will go relatively heavier (2-4MB/core) on L2 cache both as a uarch priority to feed the core with a side benefit towards keeping data closer, reducing power consumption. Maybe L3 will increase too.

 

[SpudLobby]

Yea Zen5 is the big tuna kahuna that will roadkill a whole lot of competing designs.
Mikey C. wasn't joking about sleeping and dreaming about the thing being a real product.

Hopefully Panther Lake core is a meaner peak 1t design than LNC, otherwise lol.

Peak 1t matters in race to idle scenarios, sorry.
The bulk (8 out of 12) cores on STX1 is compacts anyway.

plus the cool factor of making THAT thing run at IBM z16 speeds.

Peak 1t matters in race to idle scenarios, sorry.
The bulk (8 out of 12) cores on STX1 is compacts anyway.

Depends on the parameters. It's not going to generally outweigh the disproportionate power consumption that AMD and Intel gladly indulge for those big cores at their peaks due to obvious quadratic power demands with increased voltage. If it did, none of their balanced much less power efficiency modes would cut ST to the extent they do or e.g. when unplugging. Sure, going full [redacted] where static losses are huge like with 1.5GHz is just stupid, but that's not what I'm talking about. A Zen 4C that can handle 4GHz is fine, I'd prefer it. That said, since AMD cares about neckbeards that don't understand any of this but see matchbox benchmarks and froth at the mouth for 15-20W memes to own [other team] as they burn their testicles, they'll ship it - and I'll just change the power settings away from that one way or another depending on the vendor.

On desktops wrt Zen 5 ST, sure that's fine.

Yes 4+8 seems wise in light of Intel's corespam strategy versus the 8+4 alternative or just in general.

 

[SpudLobby]

Yea Zen5 is the big tuna kahuna that will roadkill a whole lot of competing designs.
Mikey C. wasn't joking about sleeping and dreaming about the thing being a real product.

Hopefully Panther Lake core is a meaner peak 1t design than LNC, otherwise lol.

It should be interesting based on AMD's claim about a re-pipelined frontend or Clark's insistence it's good. Suspect it's AMD's GLC basically, albeit possibly larger structurally and less bloated physically or in terms of power/performance. I expect a Zen 2 -> Zen 3 size perf uplift and probably better gating or power efficiency via more cache, wider, etc.

We'll see about LNC and PNC yeah. I've seen claims LNC is smol on N3 but that's no surprise with N3 logic + lower frequency in Lunar Lake.


RE: Andrei and Nuvia:

Let's wait and see. The apparently leaked benchmarks no longer look as great as in 2020 if it's 1700-1800 GB5 and 4-5W which is M1+ territory. I'd take it over Phoenix but suspect Zen 5 will be tougher competition. DVFS rumor (lack thereof) is pretty bad though. If true on the mobile chip and not just the canceled server part, that's a showstopper.

You oughtta be rooting for them if you want someone to hand it to Apple and actually nail the details like idle power. It's QC though so we'll see.

 

[deasd]

I start wondering who sent these confidential materials to MLID......

Intel Meteor Lake specs leak lists 14 CPU cores and 8 Xe-Cores based on Xe-LPG architecture - VideoCardz.com

First Intel Meteor Lake and Raptor Lake Refresh details What may be the very first Intel slides for the next-gen laptop series have just been shared by YouTuber Moore’s Law is Dead. The latest presentation slides finally reveal the core count for the upcoming mobile platform. While the company...

videocardz.com

 

[adroc_thurston]

So if your AMD shares and cherished rep among AMD groupies takes a bit of a knick because Intel is half alive as opposed to this vegetative flailing right now, that's great

Meh I mostly talk with Intel/Intel-adjacent folks anyway, ironically enough.

But let's wait to see someone catch them - or just get within 10% on perf/GHz, and similar margins at 4-5W

We're like mere quarters away from two vendors doing so (at once, basically).

Depends on the parameters.

It definitely matters.
That's the very reason things like Cortex-X exist.

Yes 4+8 seems wise in light of Intel's corespam strategy versus the 8+4 alternative or just in general.

It's really not an answer to Intel estrogen core spam, it only starts to make sense once you know what Zen5(c) actually is.

Suspect it's AMD's GLC basically

It's a lot bigger than GLC, lol.

I expect a Zen 2 -> Zen 3 size perf uplift

Considerably higher.
Kinda the gimmick there, really.
I've seen Turin numbers, they're insane.

 

[adroc_thurston]

I'd hazard a guess Zen 5 will go relatively heavier (2-4MB/core) on L2 cache both as a uarch priority to feed the core with a side benefit towards keeping data closer, reducing power consumption. Maybe L3 will increase too.

Don't think even Zen6 changes L2/L3 capacity per core.

 

[SpudLobby]

It definitely matters.
That's the very reason things like Cortex-X exist.

Man, the Cortex-X is in a power profile closer to a Zen 4C core (and actually certainly lower at peak relative to 4C in PHX2), which is just directionally in favor of what I've said. This is comical. I didn't say it doesn't matter as a principle, but that's not going to help you at the extremes 20% faster for 3x the power where the latter is 15W. 7840U will see longer battery life avoiding 4.8+GHz. Cortex X is more efficient than the A7x's for most of the curve anyways. it's okay to concede a reasonable point. That is a fine trade to have for neckbeards plugged in or whatever but not my thing, not a fan of blowing loads like that but I've got better things to do. I'd just buy a desktop.

It's really not an answer to Intel estrogen core spam, it only starts to make sense once you know what Zen5 actually is.

It's a lot bigger than GLC, lol.

Considerably higher.

Interesting.

 

[SpudLobby]

We're like mere quarters away from two vendors doing so (at once, basically).

Yes, *hopefully* QC and AMD with Oryon this fall and Zen 5 at CES.

 

[SpudLobby]

Don't think even Zen6 changes L2/L3 capacity per core.

Lmao this will get interesting if it's really that large. Maybe shared cache on L2 IDK. Looking forward to it either way.

RE: Zen 2 -> Zen 3: I don't see this as disappointing fwiw. If they hit a 20% uplift on IPC, that's tremendous. Would be huge for Zen 5c in particular. I'm not even sure what Intel will do at that point for their E cores in servers lmao. If they get more, well, popcorn time.

 

[adroc_thurston]

Man, the Cortex-X is in a power profile closer to a Zen 4C core

Laptop configs ARM envisions are pretty toasty, just that no one really ships em.

Cortex X is more efficient than the A7x's for most of the curve anyways

That's mostly because ARM has been very aggressive in cutting mainline Cortex-A area, unlike A76/77 days.
And A710 was a perf/W regression over A78. Sozz.

Yes, *hopefully* QC and AMD with Oryon this fall and Zen 5 at CES.

Unless both catastrophically explode (unlikely) we'll be there soon 'nuff.

Lmao this will get interesting if it's really that large. Maybe shared cache on L2 IDK. Looking forward to it either way.

Yea, then we wait for LNC and the real funny begins.

 

[adroc_thurston]

Zen 2 -> Zen 3: I don't see this as disappointing fwiw

Yea it was good but this was under area constraints which are now relaxed due to cost-sensitive markets (laptop/favelas/edge) being served by Compact piles.
Exactly the same strategy as Intel, just done in a technically different way.

I'm not even sure what Intel will do at that point for their E cores in servers lmao

Yea SRF and Ampere One are both getting roadkilled next year.
Hopefully CWF has a lot more cores than 144 cuz lol

 

[SpudLobby]

Laptop configs ARM envisions are pretty toasty, just that no one really ships em.

Not the point but yes I agree. They do suck but far and away from 15-20W ST stuff.

8Cx Gen 3 is a total POS and I'd rather have Rembrandt, Cezanne or Phoenix (obviously on the last). Qc laziness but also Samsung's 5NM clusterfuck (though I'm cautiously optimistic about them turning things around).

That's mostly because ARM has been very aggressive in cutting mainline Cortex-A area, unlike A76/77 days.
And A710 was a perf/W regression over A78. Sozz.

Either way it wasn't really a huge gap before, but yes. If they took it seriously we'd see vendors doing e.g. 1MB of L2 on the A7x but that's not even an option. Anyways A715 clawed back some efficiency, A720 should again (with the larger implementation). It's a fine core for mobile on perf/w or area and probably the only one worth anything, X cores are a bit bizarre now with the L1I and width decisions, they will be turkey basted by Nuvia if things even half pan out.

Unless both catastrophically explode (unlikely) we'll be there soon 'nuff.

It's going to be pretty great. Intel + Apple meltdowns are going to be be pretty A1.

Yea, then we wait for LNC and the real funny begins.

What's your forecast on LNC IPC gain?

 

[SpudLobby]

Yea it was good but this was under area constraints which are now relaxed due to cost-sensitive markets (laptop/favelas/edge) being served by Compact piles.
Exactly the same strategy as Intel, just done in a technically different way.

Yea SRF and Ampere One are both getting roadkilled next year.
Hopefully CWF has a lot more cores than 144 cuz lol

Right.

Yeah Ampere One looks like trash. If you're going to do a custom core, at least have the dignity to beat Arm's. 16kb L1 or whatever is a meme.

I think Intel's E Cores will still be alright if only because I don't see any other Arm competition (and Nvidia stuff is niche for HPC), so they'll probably grab some clients but if Zen 5 shapes up like it should then AMD will win the next generation too for cloud providers.