Question Raptor Lake - Official Thread

Page 92 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Hulk

Diamond Member
Oct 9, 1999
4,525
2,519
136
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.
 
  • Like
Reactions: vstar

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,786
136
Raptorlake improvements:
-Improved Intel 7 process with greater than 50mV reduction at the same frequency, 200MHz increase at the same voltage, and 600MHz increase in clocks at peak
-Machine learning prefetcher for the L3 allows INI caching scheme. INI = Inclusive/Non Inclusive, and it can switch between the two on the fly depending on the needs in 200us times. Was in Alderlake but didn't get used. Sounds like what Raichu was talking about regarding "secret sauce"? Though that was about Golden Cove.
-L2 cache in the E cores also use machine learning to improve prefetching. Performance improves from 2% in some circumstances to 16% in others.
-Fabric frequency increased by 900MHz to 5GHz. Apparently in leaks it said 4.6GHz with E cores active?
-The improvement with caching and prefetchers along with caches allow E cores to perform just like the original Skylake.
 
Last edited:

pakotlar

Senior member
Aug 22, 2003
731
187
116
What is this "PL1 105w TDP" intel keep takling about ?

Did they manually set PPT limit to 105w for the 5950x ??
Same PPT limit for both 5800x3d and 16core 5950x ? against 13900k @ 253watt ?
View attachment 68302
That 105 W TDP means PPT = 142W, and represents the maximum sustained (indefinite) draw. AMDs TDP means nothing, the CPU will draw up to the socket limit indefinitely, which is 142W. Thats the stock behavior.

Stock for 13900K is 125W PL1, but like AMD will indefinitely draw their socket limit, which is PL2 = 253W. So for 13900K and 12900K, unless tau is set to something other than the default indefinite value, PL1=PL2=253W for 13900K or 241W for 12900K.

So they are comparing stock 5800X 3D to stock 13900K.
 

nicalandia

Diamond Member
Jan 10, 2019
3,331
5,282
136
I'm curious about that bit. Is it rebranded Alder Lake, or Raptor Lake without the increase in L2? Not that it makes much of a difference either way, but still.
Occam's razor.

What is the chance of Intel expending resources on upgrading Golden Cove to Raptor Cove and then taking Raptor Cove and removing 60% of the L2 to lower the power requirements but still keeping some of the memory subsystem updates?

Or just using Golden Cove on a Mature Node to use resources on MTL.
 

pakotlar

Senior member
Aug 22, 2003
731
187
116
Indeed. Though nobody knows for sure when Raphael-X hits the streets. November? January? March? Complete unknown. So much waiting ugh.

Anyway it's hard to believe that Raptor Lake has to hold Intel over til Arrow Lake. It's a nice little upgrade for people that already liked Alder Lake, at least in those instances when it's actual Raptor Lake (and not just an Alder Lake refresh), but otherwise . . .

Intel has stated there will be desktop Meteor Lake bins, whats uncertain is whether there will be high end Meteor Lake desktop parts. For Intel’s sake that would be good, though I’m not optimistic given the talk of clock speed regressions and that the largest leaked part is 6P+8E.
 

Exist50

Platinum Member
Aug 18, 2016
2,452
3,102
136
Occam's razor.

What is the chance of Intel expending resources on upgrading Golden Cove to Raptor Cove and then taking Raptor Cove and removing 60% of the L2 to lower the power requirements but still keeping some of the memory subsystem updates.

Or just using Golden Cove on a Mature Node to use resources on MTL.
Why do you think it would be so much effort? Raptor Lake is like 99% identical to Alder Lake anyway. And Intel certainly isn't shy about doing more steppings, lol.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,786
136
Why do you think it would be so much effort? Raptor Lake is like 99% identical to Alder Lake anyway. And Intel certainly isn't shy about doing more steppings, lol.

Actually modern L2 caches on P cores are closely integrated with the core so it'll need layout changes to increase capacity.

Though they don't necessarily need to go that far. Maybe they are Alderlake+ where it gets the prefetcher changes that weren't able to be used and now they can due to maturity and more steppings.
 
  • Like
Reactions: lightmanek

nicalandia

Diamond Member
Jan 10, 2019
3,331
5,282
136
Why do you think it would be so much effort? Raptor Lake is like 99% identical to Alder Lake anyway. And Intel certainly isn't shy about doing more steppings, lol.
Why would they remove 60% of the L2 when it's clear that the Raptor Cove uArch is bringing better efficiency than Golden Cove? It does not make much sense to do that(from a performance/efficiency stand point anyways)
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,786
136
@nicalandia Caches are very power efficient and usually result in LOWER power at the chip level not the other way around.

Intel has stated there will be desktop Meteor Lake bins, whats uncertain is whether there will be high end Meteor Lake desktop parts. For Intel’s sake that would be good, though I’m not optimistic given the talk of clock speed regressions and that the largest leaked part is 6P+8E.

The ideal is skipping Meteorlake on the high end and go straight to Arrowlake like some leakers have said to get 20A as quick as possible.
 

moinmoin

Diamond Member
Jun 1, 2017
5,064
8,032
136
Raptorlake improvements:
-Improved Intel 7 process with greater than 50mV reduction at the same frequency, 200MHz increase at the same voltage, and 600MHz increase in clocks at peak
-Machine learning prefetcher for the L3 allows INI caching scheme. INI = Inclusive/Non Inclusive, and it can switch between the two on the fly depending on the needs in 200us times. Was in Alderlake but didn't get used. Sounds like what Raichu was talking about regarding "secret sauce"? Though that was about Golden Cove.
-L2 cache in the E cores also use machine learning to improve prefetching. Performance improves from 2% in some circumstances to 16% in others.
-Fabric frequency increased by 900MHz to 5GHz. Apparently in leaks it said 4.6GHz with E cores active?
-The improvement with caching and prefetchers along with caches allow E cores to perform just like the original Skylake.
Sounds like all of that could be foundry and microcode improvements that could be applied to existing ADL die masks? Then all the rebranding would actually be more than that and the new RPL die simply adds more cache and other silicon level optimizations on top.
 

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
Raptorlake improvements:
-Improved Intel 7 process with greater than 50mV reduction at the same frequency, 200MHz increase at the same voltage, and 600MHz increase in clocks at peak
-Machine learning prefetcher for the L3 allows INI caching scheme. INI = Inclusive/Non Inclusive, and it can switch between the two on the fly depending on the needs in 200us times. Was in Alderlake but didn't get used. Sounds like what Raichu was talking about regarding "secret sauce"? Though that was about Golden Cove.
-L2 cache in the E cores also use machine learning to improve prefetching. Performance improves from 2% in some circumstances to 16% in others.
-Fabric frequency increased by 900MHz to 5GHz. Apparently in leaks it said 4.6GHz with E cores active?
-The improvement with caching and prefetchers along with caches allow E cores to perform just like the original Skylake.

Was this information in the presentation?
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,786
136
Was this information in the presentation?

Yes, not all are in the slides and you have to listen to them, or was something that was asked by the interviewer.

Little more:
-Thread Director is overhauled with Windows 11 2022H2 release. Better handling of foreground vs background tasks using machine learning
-New Dynamic Tuning Technology feature in mobile with improved core parking techniques

I should also clarify that the max fabric frequency is 5GHz. They just say all core turbo for fabric goes up by 900MHz.

For the E cores the all core turbo goes up by 600MHz. Meaning same Turbo peak for both ST and MT.

Ok, so 900MHz ACT increase for fabric making it 5GHz makes sense. Alderlake with all cores active, it's limited by the Turbo frequency of the E cores or 4.1GHz. That's a pretty decent improvement. Intel themselves said 5% improvement in gaming.

This addresses bulk of the Alderlake hybrid issues. Just like what Ticks do.

(4.6GHz was from the early leaked benchmark so I guess we can disregard that)
 
Last edited:

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,786
136
"Raptor Cove" is Intel 7 Ultra(internal naming) process plus additional caches and prefetcher improvements.

The only part you can change without using a new die is the prefetcher algorithm. So you won't get the peak clock increase, you won't get the cache increase and you won't get the perf/watt increase.
 
  • Like
Reactions: Carfax83

nicalandia

Diamond Member
Jan 10, 2019
3,331
5,282
136
It's been confirmed by Sisoftware Sandra that The i5 13400 is Also using a rebranded Golden Cove configuration.

13th Gen Intel Core i5-13400 6C + 4c 20T 2.5 4.6 GHz, 7x 1.25 MB L2, 20MB L3)

According to Geekbench the rebranded CPUs are

13900HK and 13700H are
GenuineIntel Family 6 Model 186 Stepping 2

The 13400 is
GenuineIntel Family 6 Model 183 Stepping 1

The Processor ID does say that it is a different Model entirely.
Perhaps it's a refresh of the core like Lucienne was to Renoir Both are Zen2 based but have different models and code-name.

1664321797291.png
 
Last edited:
  • Like
Reactions: lightmanek