Discussion Intel current and future Lakes & Rapids thread

[moinmoin]

I think that's the exception to the rule rather than the norm. Skylake refreshes were only truly super profitable because AMD couldn't compete regardless, right? Now that competition from AMD has returned, moving to new nodes and new architectures is a necessity to compete.

Yes and no. Any company should plan staying competitive against valiant competition as a rule indeed. But finding markets where there's less competition and more leverage to make money is also always a valid goal.

My worry is that after setting plenty promising expectations last year (IFS plus ecosystem, steady node progress, increasing flexibility in packaging) Intel's execution may stutter again, only this time due to financial necessity.

 

[A///]

Also, depending on IPC increases, a 6+16 part could be really competitive.

I said this a while back. AMD increasing core counts was great, but you can't increase them ad infinitum. There is an limit to how much you can without expanding pass possible thermal envelopes and package size. 6+16 seems week to the uninformed, but if those 16 e-cores are 11th or 12th gen performance two to one, then it may bring a lot of hurt to AMD. 6 next gen performance cores with 16 efficiency cores which can be seen as 8 11th or 12th gen cores sans ht will be incredible.

It'd be nice to see intel figure out how to include avx512 and ht on the e cores in the future. Intel realistically cannot go back to 4 or 6 core designs due to people getting a taste of high core designs and some software capable of capitalizing on such performance figures. They can make more efficient and incredibly fast cores. Power, efficiency and price can be on Intel's side.

Zen 5 next year is the big question. Is it everything amd says it'll be or will it be another Zen 4, plagued by superficial issues out of AMD's hands and only to be matched by Intel a few months down the line if not before with MTL.

It would be bizarre if MTLS's proposed 6+16 or 8P got close or matched to an 8950X/9950X. It would be a welcome challenge to AMD. Time will tell.

 

[Hulk]

I said this a while back. AMD increasing core counts was great, but you can't increase them ad infinitum. There is an limit to how much you can without expanding pass possible thermal envelopes and package size. 6+16 seems week to the uninformed, but if those 16 e-cores are 11th or 12th gen performance two to one, then it may bring a lot of hurt to AMD. 6 next gen performance cores with 16 efficiency cores which can be seen as 8 11th or 12th gen cores sans ht will be incredible.

It'd be nice to see intel figure out how to include avx512 and ht on the e cores in the future. Intel realistically cannot go back to 4 or 6 core designs due to people getting a taste of high core designs and some software capable of capitalizing on such performance figures. They can make more efficient and incredibly fast cores. Power, efficiency and price can be on Intel's side.

Zen 5 next year is the big question. Is it everything amd says it'll be or will it be another Zen 4, plagued by superficial issues out of AMD's hands and only to be matched by Intel a few months down the line if not before with MTL.

It would be bizarre if MTLS's proposed 6+16 or 8P got close or matched to an 8950X/9950X. It would be a welcome challenge to AMD. Time will tell.

I can't thank AMD enough for breaking the Intel created glass ceiling on core count and smashing right through to 16 cores for the consumer desktop. The 3xxx, 5xxx, and now 7xxx series are brilliant in performance, a streamlined stack, efficiency, pricing pretty much everything. 8, 12, and 16 cores.. boom, just like that. After struggling to get affordable 4 or 6 core parts for years.

And now Intel has been forced to produce a $500 24 core part thanks to AMD.

With all of my core fandom I should also say that *most* software is still ruled by 6 to 8 really strong P cores. While I think a 6+16 Meteor Lake part sound reasonable keep in mind that the 13900K can do 5.5GHz stock on all cores with adequate cooling, which admittedly is really a custom loop.

Assuming a 15% IPC increase for Redwood Cove would mean it would need 4.8GHz to compete core-for-core with Raptor Cove, but it would be down 2 cores. A counter argument would be that a "coolish" running Meteor Lake 6+8 running at 4.8GHz would actually be able to maintain those clocks, while it's hard for me to hold 5.0GHz all-core on my 13900K with a 280mm AIO. Other around here are doing much better so perhaps I'm an outlier data point.

While there are solid arguments on both sides I have a feeling Intel 4 just ain't there yet for the desktop with Raptor clocks hitting 6GHz. But sustained 4GHz at low power for mobile would indeed be a step forward, which is of course why I believe the Meteor mobile only rumors may come to pass.

 

[mderbarimdiger]

Then the whole process starts again with Intel 20A but I am assuming Intel is getting ahead of this better each time. We shall see.

with 20A - 18A Intel will not be in same situation.

Intel 4 competes with intel 10 which is now its 6th year in refinement.

lets assume 18A will be in mass production by 2026, by than intel3 will be in its 3th year of refinement, 18A as a smaller node should better compete.

 

[Hulk]

Intel 4 competes with intel 10 which is now its 6th year in refinement.

I didn't know that Intel 4 is now competing with Intel 10, which I assume you to mean Intel 7. That's great news for Intel depending on what "competing" means. Competing on a clock speed basis or competing like Raptor Cove on Intel 7 and Meteor on Intel 4 are achieving similar performance?

Where did you read this?

 

[mderbarimdiger]

MTL is using intel 4 (7nm) right?
Vs RPL using intel 10 nm

 

[poke01]

What Intel has to fix is efficiency. The new X3D chips are efficient while being powerful.

 

[uzzi38]

Which specific milestones are you referring to? Because they should be sampling to customers starting with ES1.

As far as I'm aware only ES1 is with OEMs at all (and even that was quite recent), so I have no clue where this talk of ES2 is coming from.

 

[Exist50]

As far as I'm aware only ES1 is with OEMs at all (and even that was quite recent), so I have no clue where this talk of ES2 is coming from.

Well that discussion was an aside about the old Arrow Lake leak, but on the topic of Meteor Lake, if they're shipping by the end of this year, they will need to start sending out ES2 samples relatively soon. So perhaps that's where this latest info is coming from?

 

[Exist50]

Was Rocket Lake more energy efficient than Comet Lake though? Maybe when equaling core count but I doubt the 11900k is more efficient than the 10900k. I think consumers are willing to accept a side grade if MTL ends up having a good gain in efficiency, which I think it will have. As long as the ST is within 95ish percent of the 14900k, I think Intel has a good argument for launching it. Consumers with a 12900k-14900k prob won't have a reason to upgrade if ST doesn't increase, but they prob won't upgrade anyway since MTL-S is on a new mobo regardless.

No, I don't think Rocket Lake was more efficient, but I think they're comparable in terms of "new gen that does important things worse than the old gen". If it was just part of a full lineup with e.g. Arrow Lake, that would be one thing, but Meteor Lake S vs Raptor Lake would probably be significantly more expensive when you factor in the new silicon + platform. And no one's going to trust Intel to keep the socket alive for long enough for that to be a selling point.

 

[uzzi38]

Well that discussion was an aside about the old Arrow Lake leak, but on the topic of Meteor Lake, if they're shipping by the end of this year, they will need to start sending out ES2 samples relatively soon. So perhaps that's where this latest info is coming from?

We're still a good bit away from it.

I don't think we'll see MTL ship in any significant volume to consumers this year (If at all - hell I genuinely wouldn't be surprised if we only see it ship in solid volumes in the latter half of Q1)

I'd point towards Dylan's commentary on that side of things though.

 

[Exist50]

We're still a good bit away from it.

I don't think we'll see MTL ship in any significant volume to consumers this year (If at all - hell I genuinely wouldn't be surprised if we only see it ship in solid volumes in the latter half of Q1)

I'd point towards Dylan's commentary on that side of things though.

They'll probably miss the holiday season for any significant volume to consumers, but if they want to even ship to OEMs, they're going to need to get ES2 out soon.

As for Dylan, he's been far off base on MTL since day one (e.g. claiming it used N3 as fact, speculating about ODI, etc). He just dresses up his flavor of BS as "analysis" instead.

 

[Hitman928]

If this tweet is correct, MTL-S may have 6+8, 6+16, and 8 core variants:

https://twitter.com/x/status/1629570261480792065

Interesting, if true. We will see.

Also, depending on IPC increases, a 6+16 part could be really competitive.

Tweet seems to be deleted now.

 

[Exist50]

Tweet seems to be deleted now.

A leaker just behind the curve. Figures. Not sure why people are assuming there will be a 6+16 mobile die. We've seen one, and it's 6+8. Even then, they can't just put MTL-P in a desktop socket. They'd need a new SoC.

 

[Hulk]

What Intel has to fix is efficiency. The new X3D chips are efficient while being powerful.

That's a fact.

 

[Hulk]

No, I don't think Rocket Lake was more efficient, but I think they're comparable in terms of "new gen that does important things worse than the old gen". If it was just part of a full lineup with e.g. Arrow Lake, that would be one thing, but Meteor Lake S vs Raptor Lake would probably be significantly more expensive when you factor in the new silicon + platform. And no one's going to trust Intel to keep the socket alive for long enough for that to be a selling point.

Hard to compare Rocket Lake and Comet Lake efficiency. Same node about the same die area. 10 Skylake cores vs. 8 Cypress Cove cores. In really optimized MT things like CB Comet was faster, at the same power I bet it was faster as well. But on ST things Rocket was faster or more efficient at iso performance.

The bottom line here is being constrained to a given process, that is 14nm it's the timeline dictating 10 Skylake cores vs. 8 Rocket Lake cores. If economics were not a concern they could have just made a huge die with 20 Cypress Cove cores on it, had the fastest chip, at lower clocks for MT work this unicorn would have been efficient as well. And Intel would have lost money on selling such a huge piece sand. So for Alder Lake they went hybrid so as to be able to compete while being behind with process technology.

AMD using TSMC's process along with the evolution of Zen got them to the performant and efficient place they are now.

At a given process (or if you're stuck on a process I think it would be fair to say you can pick two of the three below and it describes the reason we have hybrid parts from Intel.

Great MT performance (say 40 or so small area efficient cores)
Great ST performance (say 12 or so high IPC cores)
Reasonable die size

Intel needed to get as close to all three as possible and split the difference with the hybrid approach.

 

[Geddagod]

They'll probably miss the holiday season for any significant volume to consumers, but if they want to even ship to OEMs, they're going to need to get ES2 out soon.

As for Dylan, he's been far off base on MTL since day one (e.g. claiming it used N3 as fact, speculating about ODI, etc). He just dresses up his flavor of BS as "analysis" instead.

Can confirm. I'm not so proud as to not admit it, I thought Dylan was a legit analysist and MTL using N3 was a fact. I believe I even had this exact conversation with @Exist50 like a year or so ago on reddit lol, and ye I'm eating my words now haha
The problem is that he doesn't claim it's rumor or not confirmed by Intel themselves explicitly, so for new readers or people who don't follow him, they take his word as an analysist as info coming from Intel themselves, since he also groups that info with legit info as well.

 

[Geddagod]

As far as I'm aware only ES1 is with OEMs at all (and even that was quite recent), so I have no clue where this talk of ES2 is coming from.

ES2 should be out by now, using RPL development cycle as a guide line. Looks like ES-2 should start being sent out ~8 months before launch. ES-1s on the other hand should have been sent out a year before launch, so ye only like 3-5 months ago I'm guessing is when ES-1 samples were sent out.
RPL was stated to have a timeline of 6 months shorter than Alder Lake, since RPL is more of a refresh, but I'm guessing most of the time saving procedures came during the design phase since rather than stuff like validation.
We know MTL powered on in end of april 2022, and power on to launch is ~15 months gap. If this held true, MTL should be launching in July, or slightly more than half way into 2023. However that is pretty doubtful to happen, so I'm assuming there were a bunch of issues during the power on phase, which is why ES-1 samples only started coming out some what recently like ~3 months ago, since if they were on track they should be releasing July of 2022 rather than 2 months later in ~November of last year.

 

[Exist50]

Assuming a 15% IPC increase for Redwood Cove would mean it would need 4.8GHz to compete core-for-core with Raptor Cove, but it would be down 2 cores. A counter argument would be that a "coolish" running Meteor Lake 6+8 running at 4.8GHz would actually be able to maintain those clocks <snip>

While there are solid arguments on both sides I have a feeling Intel 4 just ain't there yet for the desktop with Raptor clocks hitting 6GHz. But sustained 4GHz at low power for mobile would indeed be a step forward, which is of course why I believe the Meteor mobile only rumors may come to pass.

See, that's the problem. Meteor Lake appears to be almost the opposite. A serviceable enough process, but a core so lackluster it's unable to compensate for any minor weaknesses. If RWC had +15% IPC, then they'd have a desirable desktop product. But instead they're forced to fall back to mobile where they can leverage the (hopeful) efficiency gains.

 

[jpiniero]

I assume the issue with Meteor-S is more manufacturing problems/yields/capacity. Thus Raptor Refresh.

It's much tougher to hide the actual volume in desktop versus mobile.

 

[Geddagod]

Doesn't Intel ship way more to the laptop market than the desktop market? Idk

 

[jpiniero]

Doesn't Intel ship way more to the laptop market than the desktop market? Idk

They do... but it's much easier to hide the volume since it's all soldered. Dell could for instance ship a couple laptop models with Meteor Lake but mostly sell the models with Raptor Lake.

 

[Exist50]

They do... but it's much easier to hide the volume since it's all soldered. Dell could for instance ship a couple laptop models with Meteor Lake but mostly sell the models with Raptor Lake.

Why doesn't the same logic apply to desktops? DIY is pretty much negligible in the big picture.

 

[A///]

I can't thank AMD enough for breaking the Intel created glass ceiling on core count and smashing right through to 16 cores for the consumer desktop. The 3xxx, 5xxx, and now 7xxx series are brilliant in performance, a streamlined stack, efficiency, pricing pretty much everything. 8, 12, and 16 cores.. boom, just like that. After struggling to get affordable 4 or 6 core parts for years.

And now Intel has been forced to produce a $500 24 core part thanks to AMD.

With all of my core fandom I should also say that *most* software is still ruled by 6 to 8 really strong P cores. While I think a 6+16 Meteor Lake part sound reasonable keep in mind that the 13900K can do 5.5GHz stock on all cores with adequate cooling, which admittedly is really a custom loop.

Assuming a 15% IPC increase for Redwood Cove would mean it would need 4.8GHz to compete core-for-core with Raptor Cove, but it would be down 2 cores. A counter argument would be that a "coolish" running Meteor Lake 6+8 running at 4.8GHz would actually be able to maintain those clocks, while it's hard for me to hold 5.0GHz all-core on my 13900K with a 280mm AIO. Other around here are doing much better so perhaps I'm an outlier data point.

While there are solid arguments on both sides I have a feeling Intel 4 just ain't there yet for the desktop with Raptor clocks hitting 6GHz. But sustained 4GHz at low power for mobile would indeed be a step forward, which is of course why I believe the Meteor mobile only rumors may come to pass.

Agreed and yes, I see your train of thought here. Arrowlake is a big interest to me because i am very curious just how good intel's interconnect will be to reduce or keep delay at bay. hah there I've rhymed, think I'l write a song and become the next garth brooks.

 

[inf64]

See, that's the problem. Meteor Lake appears to be almost the opposite. A serviceable enough process, but a core so lackluster it's unable to compensate for any minor weaknesses. If RWC had +15% IPC, then they'd have a desirable desktop product. But instead they're forced to fall back to mobile where they can leverage the (hopeful) efficiency gains.

Do we have any real reputable source that leaked the +15% IPC uplift figure for Redwood Cove?