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.

 

[nicalandia]

If the less than 10% is in MT, I don't know what the point of RPL is.

"Enhanced Overclocking features"

This is the most laughable marketing of modern CPUs. Practically zero OC but you got more knobs to adjust them!

10% will be due to speed and IPC combined. MT workload will be a 30% boost at least on 13900K

 

[IntelUser2000]

10% will be due to speed and IPC combined. MT workload will be a 30% boost at least on 13900K

I thought so. More honest wording is a very good thing and overdelivering compared to marketing statements a year before release.

 

[Exist50]

10% will be due to speed and IPC combined. MT workload will be a 30% boost at least on 13900K

Where did you get that 30% number?

 

[IntelUser2000]

Where did you get that 30% number?

+8 E cores?

 

[Exist50]

+8 E cores?

Iso power?

Edit: I meant "Is there a rumor/leak of that number?"

 

[nicalandia]

Iso power?

Edit: I meant "Is there a rumor/leak of that number?"

That's been know for quite a while now and confirmed today



I even drew a Mock Up die shot with info based on Locuza die annotations

 

[Exist50]

That's been know for quite a while now and confirmed today

View attachment 57577

Are you just applying a naive scalar to Alder Lake then? No adjusting for power or anything else?

 

[nicalandia]

Are you just applying a naive scalar to Alder Lake then? No adjusting for power or anything else?

You can bet your house that the 13900K will reach those 30% performance boost in MT at any cost, Efficiency be damned.

 

[IntelUser2000]

Are you just applying a naive scalar to Alder Lake then? No adjusting for power or anything else?

Actually we know that the E cores add about 25% gain for 12900K. And according to the SPEC test on Anandtech review, they said 1P core is equal to 2E core but it ends up being less, because the 8P cores are enough to saturate the TDP.

Refreshes themselves can add couple hundred mhz. So take that, lower it a bit further especially for the Ps, and we might get another 25-30% again.

 

[Exist50]

Actually we know that the E cores add about 25% gain for 12900K. And according to the SPEC test on Anandtech review, they said 1P core is equal to 2E core but it ends up being less, because the 8P cores are enough to saturate the TDP.

Refreshes themselves can add couple hundred mhz. So take that, lower it a bit further especially for the Ps, and we might get another 25-30% again.

Call me a skeptic, at least...

 

[eek2121]

You can bet your house that the 13900K will reach those 30% performance boost in MT at any cost, Efficiency be damned.

Power limits are already set in stone for RPL-S. They have been for a long time. Short of the chip moving to a new socket. Power consumption will be about the same as it was for Alder Lake.

Actually we know that the E cores add about 25% gain for 12900K. And according to the SPEC test on Anandtech review, they said 1P core is equal to 2E core but it ends up being less, because the 8P cores are enough to saturate the TDP.

Refreshes themselves can add couple hundred mhz. So take that, lower it a bit further especially for the Ps, and we might get another 25-30% again.

Right, you have 50% more cores using the same amount of power and are expecting the gains to be just as big. Intel has made some changes to improve efficiency, but it won't be enough. "Diminishing returns" is a thing you know. Zen 4 is going to make this chip irrelevant.

 

[IntelUser2000]

Right, you have 50% more cores using the same amount of power and are expecting the gains to be just as big.

Not a fan of people quoting misleadingly to make a point. Alderlake doesn't have 16P cores, therefore isn't a straightforward 50% increase in cores. Same as those saying Intel is beating AMD in mobile having 14 cores. No, 8 of them are tiny E cores.

And I said slight reduction in P cores open up quite a bit of headroom for the E's.

We complained from the beginning that the Alderlake chip was meant for mobile and it simply didn't have enough threads. 8+16 solves that. Plus the P cores themselves have an improved uarch which will contribute.

 

[HurleyBird]

"Up to double digit performance boost"

Up to. Intel doesn't exactly have a history of understatement. Unless they've recently decided to do a 180 and start sandbagging, Zen 4 is going to trounce RL.

 

[desrever]

"Up to double digit performance boost" is the worse language used in a market slide I have seen in a while. Makes it seem like it will have a hard time getting 10% improvement. In pure compute perspective (power draw be damned), 8p+16e is going to be significant over 8p+8e they have now. If we use the rough estimate of each e core is about 1/2 a p core, that would put it roughly 33% more compute power. Quoting up to double digit is seriously underselling it unless they are lower clocks significantly for the e cores.

 

[DrMrLordX]

Quoting up to double digit is seriously underselling it unless they are lower clocks significantly for the e cores.

It's a bad look but nobody will really know what it means until Raptor Lake actually gets benchmarked in commercial silicon. Remember Intel's "up to 19%" rhetoric on Rocket Lake? Sometimes it was true, sometimes it wasn't.

Regardless, people should be less worried about how much 8 more Gracemont cores will help it in rendering benchmarks and more concerned about all those workloads where the extra Gracemont will add exactly 0% to total performance.

 

[deasd]

IIRC last time 'up to double digit' being used is in the Rocketlake statement, but it was refered to IPC uplift from CML to RKL. This time no statement about the IPC or ST or MT, such statement sounds utterly meaningless....

 

[coercitiv]

Regardless, people should be less worried about how much 8 more Gracemont cores will help it in rendering benchmarks and more concerned about all those workloads where the extra Gracemont will add exactly 0% to total performance.

It's too late for that, people are fixated on E cores and that's how it's gonna' stay until the wake-up call from Zen4. Only then you'll see a sudden interest in Raptor Cove optimizations.

Quoting up to double digit is seriously underselling it unless they are lower clocks significantly for the e cores.

It's either saying "up to double digit" or splitting the info into ST and MT. Remember this?

 

[dark zero]

That's been know for quite a while now and confirmed today

View attachment 57577

I even drew a Mock Up die shot with info based on Locuza die annotations

View attachment 57579

So... they will retain Big Octa and going full 16 efficient cores.

My question is... the config of those efficient cores will be on Quad Cores or will move to Octa Cores?

If the latter happens (being in the configuration of Octa Cores), a potential Raptor Lake -N might be at minimun octa? And considering the improvement it might get, it will definately put any Big Dual Core solution (aka current Celeron and Pentium) out of the bussiness unless it includes efficient cores.

Also, seeing a potential 16 core Raptor Lake -N would be brutal to say the least since each core would be on the Kabylake tier. That jump would be on par of the Conroe jump Intel pulled a lot of time ago.

And of course, seeing a 16 or even a 12 small core laptop would be crazy to watch.

It's a bad look but nobody will really know what it means until Raptor Lake actually gets benchmarked in commercial silicon. Remember Intel's "up to 19%" rhetoric on Rocket Lake? Sometimes it was true, sometimes it wasn't.

Regardless, people should be less worried about how much 8 more Gracemont cores will help it in rendering benchmarks and more concerned about all those workloads where the extra Gracemont will add exactly 0% to total performance.

Agreed with you, however... what happens if those efficient cores goes solo in a potential Raptor Lake -N?
Just 16 Small core processor is enough to cause problems to the competition. Even at 12 core is enough to be noticeable.

And yeah, AMD might counter, but how about ARM with Qualcomm and Mediatek?

 

[Abwx]

If the less than 10% is in MT, I don't know what the point of RPL is.

"Enhanced Overclocking features"

This is the most laughable marketing of modern CPUs. Practically zero OC but you got more knobs to adjust them!

The point is to improve perf/watt since ADL has not enough throughput at reasonable TDP to compete with an AMD 16C.

With 8 extra E cores they can dial down frequencies in MT and have higher thtoughput at lower TDPs, you have noticed that they aknowledge that the competition has the perf/watt crown and that they hope to get it back in 2024.

 

[IntelUser2000]

With 8 extra E cores they can dial down frequencies in MT and have higher thtoughput at lower TDPs, you have noticed that they aknowledge that the competition has the perf/watt crown and that they hope to get it back in 2024.

Ok, but they'll need the full TDP again to compete with Zen 4. I don't really believe they'll cut the top TDP figure, because it can be used for more performance. The trend is increasing TDPs.

That doesn't mean chips are less efficient. They are just increasing the dynamic range. The low power chips are awesome nowadays.

 

[Abwx]

Ok, but they'll need the full TDP again to compete with Zen 4. I don't really believe they'll cut the top TDP figure, because it can be used for more performance. The trend is increasing TDPs.

That doesn't mean chips are less efficient. They are just increasing the dynamic range. The low power chips are awesome nowadays.

They can either increase TDP by 50W and get 25% better perfs, keep the same TDP and still increase perfs by something like 15% or dial down the TDP such that perfs are slightly better with a more favourable perf/Watt ratio.

Their slide wich state up to double digit perfs improvement point to the third option even if they release some halo variant with increased TDP for the enthusiasts, the most meaningfull will be the mainstream SKUs.

As for competing against Zen 4 this cant be adressed with their current node, they ll have to wait for their equivalent of TSMC s 5nm, in the meantime they ll surely rely on high TDPs.

 

[IntelUser2000]

As for competing against Zen 4 this cant be adressed with their current node, they ll have to wait for their equivalent of TSMC s 5nm, in the meantime they ll surely rely on high TDPs.

We KNOW from leaks that Raptorlake has the same 250W PL as Alderlake. I mean details change a bit but overall it's the same.

With the next gen GPUs going in the 450W range, and the multi-tile GPUs probably in the 600W range, the trend is increasing power for more performance.

 

[coercitiv]

We KNOW from leaks that Raptorlake has the same 250W PL as Alderlake. I mean details change a bit but overall it's the same.

Well... about that: the PL2 limit for 12900K is 241W. If the leak about 12900KS is true then PL2 for this new SKU goes up to 260W.

We also know about two "new" TDP points on AM5, one for 120W and one for 170W, so we'll see SKUs with higher than 142W Package Power Tracking, possibly 200W+. In other words, between the performance of ADL-S and the upcoming performance of Zen4 flagship, Raptor Lake's power limit is going anywhere but down. My money is on ~260W PL2 for 13900K.

Intel is caught in a vicious circle and it's going to take an outstanding CPU generation to bring power usage under control (from the power limit PoV). Or new executive leadership, whichever comes first.

 

[mikk]

View attachment 57566

So, less than 10%?

Did they test with the final clock speeds of Raptor Lake? In their demo it was an Intel 0000, they are usually lower clocked in such early state.

 

[Exist50]

Did they test with the final clock speeds of Raptor Lake? In their demo it was an Intel 0000, they are usually lower clocked in such early state.

It's an incremental revision. There's no real reason the chip would clock low. In any case, the figure wasn't from the actual demo, and refers to the actual product expectations.