Question Raptor Lake - Official Thread

[Hulk]

Since we already have the first Raptor Lake leak I'm thinking it should have it's own thread.
What do we know so far?
From Anandtech's Intel Process Roadmap articles from July:

Built on Intel 7 with upgraded FinFET
10-15% PPW (performance-per-watt)
Last non-tiled consumer CPU as Meteor Lake will be tiled

I'm guessing this will be a minor update to ADL with just a few microarchitecture changes to the cores. The larger change will be the new process refinement allowing 8+16 at the top of the stack.

Will it work with current z690 motherboards? If yes then that could be a major selling point for people to move to ADL rather than wait.

 

[Asterox]

Well, it's very obvious why they did it. But why did they do that, when they had to specify the configuration details anyway.Data is not invisible, so complete absurdity or an attempt to show the i9 13900K in a better light.

i9 12900K, ddr5 4800mhz

i9 13900K, ddr5 5600mhz

 

[inf64]

Does anyone know when the review NDA is going to be lifted for 13th Gen?

 

[jpiniero]

Probably isn't actually Raptor Cove, unless it's a bizarre early look a refresh, but rather the much-awaited Sapphire Rapids monolithic die. And with a full 8 memory channels too.

A 34 core SPR would be way too big so it has to be Raptor Cove. It's still quite the big boi.

I checked Dell and Lenovo and they are both still not selling Icelake-W so you figure that is dead. But something like this 34 core Raptor Cove might have appeal to OEMs.

 

[Carfax83]

Well, it's very obvious why they did it. But why did they do that, when they had to specify the configuration details anyway.Data is not invisible, so complete absurdity or an attempt to show the i9 13900K in a better light.

i9 12900K, ddr5 4800mhz

i9 13900K, ddr5 5600mhz

I touched on this in the other Intel thread and I don't see a problem with it. Alder Lake is validated to run officially at DDR5 4800 speeds. Yes we all know it can typically use much higher memory frequencies, but it's not validated. There's presumably been a few people that have bought DDR5 6400 kits along with their 12900K CPUs and have had to run them at slower speeds and more relaxed timings because the memory controller is incapable of hitting that speed without being unstable.

Raptor Lake's memory controller is validated to run at at least 2800mhz so that it can fully take advantage of DDR5 5600. Intel could have cherry picked a 13900K with a beast memory controller that could comfortably run at DDR5 7000+ speeds if they wanted to, and skewed the benchmarks even more but they didn't because that memory speed is unvalidated and not guaranteed by the manufacturer to work with Raptor Lake.

 

[nicalandia]

I touched on this in the other Intel thread and I don't see a problem with it. Alder Lake is validated to run officially at DDR5 4800 speeds. Yes we all know it can typically use much higher memory frequencies, but it's not validated. There's presumably been a few people that have bought DDR5 6400 kits along with their 12900K CPUs and have had to run them at slower speeds and more relaxed timings because the memory controller is incapable of hitting that speed without being unstable.

Raptor Lake's memory controller is validated to run at at least 2800mhz so that it can fully take advantage of DDR5 5600. Intel could have cherry picked a 13900K with a beast memory controller that could comfortably run at DDR5 7000+ speeds if they wanted to, and skewed the benchmarks even more but they didn't because that memory speed is unvalidated and not guaranteed by the manufacturer to work with Raptor Lake.

The 7600X at DDR6000 also kicks the 12900k at DDR4800 In a simlar fashion as the 13900K.

Truth is there is not going to be much difference between both CPUs at DDR6400

 

[inf64]

I touched on this in the other Intel thread and I don't see a problem with it. Alder Lake is validated to run officially at DDR5 4800 speeds. Yes we all know it can typically use much higher memory frequencies, but it's not validated. There's presumably been a few people that have bought DDR5 6400 kits along with their 12900K CPUs and have had to run them at slower speeds and more relaxed timings because the memory controller is incapable of hitting that speed without being unstable.

Raptor Lake's memory controller is validated to run at at least 2800mhz so that it can fully take advantage of DDR5 5600. Intel could have cherry picked a 13900K with a beast memory controller that could comfortably run at DDR5 7000+ speeds if they wanted to, and skewed the benchmarks even more but they didn't because that memory speed is unvalidated and not guaranteed by the manufacturer to work with Raptor Lake.

Although it's true that official spec is 4800 for AL, they could have ran the tests with 5600 memory just to show apples to apples comparison. AL does gain noticeable fps when going to 5600+ so the real review numbers are really not going to be comparable to what intel showed. Most AL parts can run at least DDR5 5600 without any problems.

Also, suspect is the fact they used DDR4 3200 on AMD platform but that's besides the point. As nicalandia noted, there should not be much of a difference between all chips mentioned (Z3D, AL, RC and Z4), not until we get Nvidia 4090 to remove the GPU wall.

 

[Carfax83]

Although it's true that official spec is 4800 for AL, they could have ran the tests with 5600 memory just to show apples to apples comparison. AL does gain noticeable fps when going to 5600+ so the real review numbers are really not going to be comparable to what intel showed. Most AL parts can run at least DDR5 5600 without any problems.

I think the reason why they do this is to prevent lawsuits due to misrepresentation and false advertisement. Intel always does their product launches with the officially supported memory frequency. AMD also did the same thing until Zen 4. I was honestly shocked that AMD paired their Zen 4 review samples with DDR5 6000 memory kits, because as far as I know, DDR5 6000 isn't validated to run on Zen 4.

 

[inf64]

I think the reason why they do this is to prevent lawsuits due to misrepresentation and false advertisement. Intel always does their product launches with the officially supported memory frequency. AMD also did the same thing until Zen 4. I was honestly shocked that AMD paired their Zen 4 review samples with DDR5 6000 memory kits, because as far as I know, DDR5 6000 isn't validated to run on Zen 4.

It's for legal reasons I guess (in intel's case), but so far we have even reports that Zen 4 can run 4 dimms with DDR 6000 without any problems (even on 7600X which is the lowest quality silicon).
My point is that the difference in games with current GPUs will be much smaller than what intel showcased, but lets wait and see.

 

[Carfax83]

 

[LightningZ71]

Assuming that Intel can bin 32-34 core models of the RPLS34 die, they should comfortably sit between the 5975wx and 5995wx threadrippers in most tasks. I don't see it getting past the 64 core -95 in most any MT task that properly scales beyond 32 threads, but it should do quite well in anything in the 8-32 thread set.

I'm just happy that Intel hasn't fully abandoned that market as it at least keeps AMD's prices reasonable.

 

[nicalandia]

I'm just happy that Intel hasn't fully abandoned that market as it at least keeps AMD's prices reasonable.

Do you have an Idea how long it will take for the actual product to come to market? Having a Wafer on display is not the same as being production ready

 

[Exist50]

Others beat me to it but that would mean another set of masks for yet another variation of Raptor Lake. Raptalder Lake? Whatever you would call it. It would be a waste of resources.

And if it gets them some extra performance? Raptor Lake will need to hold them over for a year+. Even marginal gains may be worth the effort. Also, if DLVR actually shows up, that'll require new silicon too.

A 34 core SPR would be way too big so it has to be Raptor Cove. It's still quite the big boi.

Dude, come on. It's right there in the picture. They certainly wouldn't have created such a chip just for the workstation market.

Do you have an Idea how long it will take for the actual product to come to market? Having a Wafer on display is not the same as being production ready

Whatever they decide to call it, it's really Sapphire Rapids. So should hopefully be shipping in some form by year's end.

 

[uzzi38]

Probably isn't actually Raptor Cove, unless it's a bizarre early look a refresh, but rather the much-awaited Sapphire Rapids monolithic die. And with a full 8 memory channels too.

Yup it's SPR MCC.

And on that topic: Monolithic Sapphire Rapids - by SkyJuice - Angstronomics

 

[jpiniero]

Dude, come on. It's right there in the picture. They certainly wouldn't have created such a chip just for the workstation market.

I didn't say doing a 34 core Raptor Lake would be worth the engineering effort. Raptor Cove is probally like half the size of the Sapphire core so you can see why they did it.

 

[nicalandia]

Yup it's SPR MCC.

And on that topic: Monolithic Sapphire Rapids - by SkyJuice - Angstronomics

From the Article:

Core Count Regression
"Sapphire Rapids introduces a much larger, faster CPU core combined with more IO and acceleration features such as AMX, which all consumes area. This results in a regression in core count for a given die size vs the previous generation. A 10nm wafer can fit 84 dies of 40-core Ice Lake, but only 68 dies of 34-core Sapphire Rapids. So, despite having fewer cores, the new generation takes 24% more wafers to fit the same number of chips. This gap widens once factoring a lower yield for larger chips."

 

[nicalandia]

Raptor Cove is probally like half the size of the Sapphire core so you can see why they did it.

Where are you getting this info from? Raptor Cove if anything is a Larger(due to Larger L2, but not by much) than Golden Cove. They are both built on the same process node.

 

[jpiniero]

Where are you getting this info from? Raptor Cove if anything is a Larger(due to Larger L2, but not by much) than Golden Cove. They are both built on the same process node.

AMX and the extra AVX-512 unit take up a ton of space.

Yup it's SPR MCC.

And on that topic: Monolithic Sapphire Rapids - by SkyJuice - Angstronomics

If it was actually 770 mm2 (it doesn't look that big but the Toms angle isn't that great), it'd be unsellable. They can't even get OEMs to use Icelake-W.

 

[nicalandia]

AMX and the extra AVX-512 unit take up a ton of space.

It's not that much, 1/5th at most. But what make you believe Intel will put client based Raptor Cove cores on a Monolithic Die? On the Sapphire Rapids based Xeon W9 they are using the full fat AMX/AVX-512 Core. There is no reason to believe they will not be using the Emerald Rapids Raptor Cove on those as well.

 

[jpiniero]

It's not that much, 1/5th at most. But what make you believe Intel will put client based Raptor Cove cores on a Monolithic Die? On the Sapphire Rapids based Xeon W9 they are using the full fat AMX/AVX-512 Core. There is no reason to believe they will not be using the Emerald Rapids Raptor Cove on those as well.

They are calling it Raptor Lake instead of Sapphire Rapids-W.

 

[inf64]

They are calling it Raptor Lake instead of Sapphire Rapids-W.

Wouldn't AVX512 and AMX be very useful for workstation use? I find it odd that they would cut these features out for Xeon part.

 

[nicalandia]

They are calling it Raptor Lake instead of Sapphire Rapids-W.

Raptor Lake is just product name that can and will certainly encompass different core uArchs(For example Rebranded Golden Cove/Golden Cove+, Client Raptor Cove and Server based Raptor Cove on HEDT)

 

[Exist50]

I didn't say doing a 34 core Raptor Lake would be worth the engineering effort. Raptor Cove is probally like half the size of the Sapphire core so you can see why they did it.

By all indications, it's Golden Cove in the full Sapphire Rapids configuration, AVX512, AMX, and 2MB L2 included. Not that client Raptor Cove would be much smaller.

If it was actually 770 mm2 (it doesn't look that big but the Toms angle isn't that great), it'd be unsellable. They can't even get OEMs to use Icelake-W.

Unsellable? Looking at how many 32c SKUs they have for SPR, it'll probably sell more units than the bigger XCC die/package. Icelake-W sucks because the performance gains are so-so over Cascade Lake, and OEMs thought Sapphire Rapids was coming sooner than it will. SPR should be a much bigger deal for workstations.

 

[jpiniero]

Wouldn't AVX512 and AMX be very useful for workstation use? I find it odd that they would cut these features out for Xeon part.

IIRC AVX-512 is still there in the Raptor Cove core.

 

[inf64]

IIRC AVX-512 is still there in the Raptor Cove core.

Yes but Is it the same full AVX512 unit as in original Sapphire Rapids?

 

[jpiniero]

Yes but Is it the same full AVX512 unit as in original Sapphire Rapids?

No.