Question Raptor Lake - Official Thread

Page 56 - 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
4,191
1,975
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

Makaveli

Diamond Member
Feb 8, 2002
4,715
1,049
136
I wouldn't read too much into these results. IMO these subtests really depend on the exact rig and even benchmark run. At least in earlier versions I managed to influence some of these by more than 20% just by playing around with memory timings on my 3700X.

I particularly wouldn't equal these very specific micro-benches to overall web use experience. Web is complex, there are no true passable benchmarks for it. Speedometer 2.0 is probably the best available, as it at least runs an actual (rather simple but still) SPA on different frontend frameworks, doing multiple runs, in an actual web browser. It still isn't nowhere near good enough to make any overarching claims, but it certainly paints a different picture and is somehat relevant (unlike octane, kraken, which measures nothing similar to actual web code):

128194.png


And this is the area I do feel comfortable talking about, being a full stack developer for 15 years, out of which a good chunk I have optimised front-end code in different real-world web apps, to run well.

I really hope anandtech updates the browser they are using on this test.

on the most recent version of Edge I score 266 points with a 5800X
 
  • Like
Reactions: Lodix

John Carmack

Member
Sep 10, 2016
153
246
116
But please people, understand what 'leaks' and 'engineering samples' are for. Entertainment value. That's it.

I couldn't let this comment pass by. I hope most people can recognize that most 'leaks' are nothing more than carefully orchestrated PR by the marketing departments of the company in question to get out and shape narratives. Rarely is it ever a real leak the likes of someone sneaking a grainy low light phone camera grab of a prototype from Foxconn's factory floor.

It's no coincidence that after Intel's worst quarter in decades and all the bad news product announcements (more SPR delays, Optane abandonment, GPU shipments missing targets) we got a rapid fire barrage of 'leaks' the week after showing big clocks and big benchmarks.

Detailed RPL gaming benchmarks. Overclocked chips at 6GHz munching on 1.536V, 1.546V, with cores and other features disabled to flex on the single thread benches. Chips juiced up to 350W power to flex on some multi core benches.

Then it got really quiet for over a week until the day AMD had some boilerplate Ryzen 7000 PR to throw out there, and that's when we got a 'leak' with 7-Zip benchmarks.

They used the same playbook when Ian Cutress published his scathing Rocket Lake early review (March 5, 2021). Literally within a few days there was a deluge of 'leaks' detailing the next generation Alder Lake (March 6), Tiger Lake (March 7), and even a bit on Raptor Lake (March 11).

I wouldn't be surprised at all if a 'leaker' like OneRaichu gets all of this information directly from Intel.
 

shady28

Platinum Member
Apr 11, 2004
2,520
397
126
I couldn't let this comment pass by. I hope most people can recognize that most 'leaks' are nothing more than carefully orchestrated PR by the marketing departments of the company in question to get out and shape narratives. Rarely is it ever a real leak the likes of someone sneaking a grainy low light phone camera grab of a prototype from Foxconn's factory floor.

It's no coincidence that after Intel's worst quarter in decades and all the bad news product announcements (more SPR delays, Optane abandonment, GPU shipments missing targets) we got a rapid fire barrage of 'leaks' the week after showing big clocks and big benchmarks.

Detailed RPL gaming benchmarks. Overclocked chips at 6GHz munching on 1.536V, 1.546V, with cores and other features disabled to flex on the single thread benches. Chips juiced up to 350W power to flex on some multi core benches.

Then it got really quiet for over a week until the day AMD had some boilerplate Ryzen 7000 PR to throw out there, and that's when we got a 'leak' with 7-Zip benchmarks.

They used the same playbook when Ian Cutress published his scathing Rocket Lake early review (March 5, 2021). Literally within a few days there was a deluge of 'leaks' detailing the next generation Alder Lake (March 6), Tiger Lake (March 7), and even a bit on Raptor Lake (March 11).

I wouldn't be surprised at all if a 'leaker' like OneRaichu gets all of this information directly from Intel.


So yeah it's all a plot by Intel. Is it Intel's plot, or is the plot that it's Intel's plot?

Leaks are almost universally coming out of motherboard manufacturers in Taiwan and China. That's because they need ES samples before launch. Pretty obvious when you think about, I'd say about 99% chance that Raichu among others on twitter with early leaks works for Foxconn, Gigabyte, Asus, etc.

I'd say its also about 99% chance that they do *not* work for Intel.
 

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
Detailed RPL gaming benchmarks. Overclocked chips at 6GHz munching on 1.536V, 1.546V, with cores and other features disabled to flex on the single thread benches. Chips juiced up to 350W power to flex on some multi core benches

There's been benchmarks at default power and the 13900K still slaps like a mutha. Same or slightly higher power characteristics as the 12900K but with twice the efficiency cores, more L2 cache, enhanced L3 cache performance, higher clock speeds across all cores including efficiency cores

I don't know how Intel did it, but they managed to significantly refine the performance per watt of Raptor Lake without changing nodes.
 
Jul 27, 2020
15,738
9,800
106

Looks like Intel may still get to have the "most GHz" advantage.

1660880773531.png

The processor managed a Cinebench multi-core score of 42,790 with all performance cores clocked at 5.8 GHz, blowing the i9-12900KF's 26,939 out of the water in multi-core benchmarks. The water-cooled i9-13900KF also scored over 5,000 more points than its air-cooled scores. The Raptor Lake processor showed a 72% increase compared to the i9-12900KF in CPU-Z scores.
 

ZGR

Platinum Member
Oct 26, 2012
2,052
656
136
I really hope anandtech updates the browser they are using on this test.

on the most recent version of Edge I score 266 points with a 5800X

They will remain on these old versions forever, since retesting all the CPU's is sadly not possible. I hope I'm wrong.
 

John Carmack

Member
Sep 10, 2016
153
246
116
So yeah it's all a plot by Intel. Is it Intel's plot, or is the plot that it's Intel's plot?

Leaks are almost universally coming out of motherboard manufacturers in Taiwan and China. That's because they need ES samples before launch. Pretty obvious when you think about, I'd say about 99% chance that Raichu among others on twitter with early leaks works for Foxconn, Gigabyte, Asus, etc.

You've been around for 100 years but apparently you've never heard of companies facing negative press trying to change the script by getting people to talk about something else? That's not exactly a new idea. Would you be shocked if I told you I'm sure there are hired marketing firms being paid for their presence on these forums and on other social media?

I was actually on another site a few days ago reading a thread where people were reminiscing over a game from a couple of years ago. One person brought up an early 'leaked' screenshot that when the game actually came out much later, the same 'leaked' screenshot appeared in official press materials. Imagine that.

I'd say its also about 99% chance that they do *not* work for Intel.

I did not say I believed they worked for Intel so why are you telling me with emphasis that you don't think they're working for Intel? Do you believe some of these regular leakers are performing the benching and overclocking personally or do you think someone else is passing along screenshots to them?

There's been benchmarks at default power and the 13900K still slaps like a mutha. Same or slightly higher power characteristics as the 12900K but with twice the efficiency cores, more L2 cache, enhanced L3 cache performance, higher clock speeds across all cores including efficiency cores

I don't know how Intel did it, but they managed to significantly refine the performance per watt of Raptor Lake without changing nodes.

You had the answer written down. It's a combination of marginal process refinements moving from 10nm+++ to 10nm++++ and spending more die space to add 8 cores.

In R23, a 5800X scores over 15K (~108pts/W) and a 5950X scores over 25K (~178pts/W). Same architeture, same process, same power consumption, but over 60% higher score. The 5995WX scores over 66K (~237pts/W). Same architecture, same process, 100% more power consumption but 160% higher score. It should be fairly obvious why that is and why Raptor Lake has better performance per watt than Alder Lake.
 

Det0x

Golden Member
Sep 11, 2014
1,027
2,953
136

Looks like Intel may still get to have the "most GHz" advantage.

View attachment 66127
Intel this is ok haha.png

Dont even think it was done with a "normal chiller" using water.. If you check screenshots you'll see hwinfo reading "0 degrees" on sensors, making me belive chiller are running a other liquid (alcohol/glycol?) below the freezing point.
(read temperature in hwinfo should always be ~5 degrees higher than outlet temp from the chiller)
1660911067655.png 1660911174346.png

Lol reminds me of the Xeon W water chiller demo.
Yes indeed, brings back memories to "Intel: We 'Forgot' to Mention" :D
 
Last edited:

Asterox

Golden Member
May 15, 2012
1,026
1,775
136

Looks like Intel may still get to have the "most GHz" advantage.

View attachment 66127

Well, for Intel(CPU power consumption only for 18+)there is no other options if AMD increases TDP and CPU speed 5ghz+. :grinning:

We all now, how this looks with "old Alder Lake".

2022-08-18_213201.jpg
 
  • Like
Reactions: lightmanek

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136

Possible leak on the Raptor Lake processor list. Also, despite the headline, there's also 'confirmation' that the locked i5's and below are indeed Alder Lake. At least the i5's are getting small cores... even the 13400.
 

Gideon

Golden Member
Nov 27, 2007
1,608
3,569
136

Possible leak on the Raptor Lake processor list. Also, despite the headline, there's also 'confirmation' that the locked i5's and below are indeed Alder Lake. At least the i5's are getting small cores... even the 13400.
Only 13900 series seems to be new silicon, according to that. The rest is Alder-Lake refresh.

If that's true it getsquite a bit easier for AMD (no IPC gains and lower clocks)

But the lower budget CPUs should still be Intel's turf. Even the 13400 gets 4 efficency cores, 13500 gets 8. This + the possibility of DDR4 looks unbeatable for the value buolds
 
  • Like
Reactions: igor_kavinski

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,674
136
Also, despite the headline, there's also 'confirmation' that the locked i5's and below are indeed Alder Lake. At least the i5's are getting small cores... even the 13400.
That would good news overall, not perfect but so much better than 6+0 configs in the value segment. The 13400 vs. 13500 value comparison will be very interesting.

Also RIP i3.
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,674
136
If that's true it getsquite a bit easier for AMD (no IPC gains and lower clocks)
The extra cores would also come with higher clocks and extra L3, the resulting SKU would compete with 12600K in terms of performance. A 6+8 config would have higher L3 than 12600K, very juicy stuff for gaming.

Hope it's true.
 

uzzi38

Platinum Member
Oct 16, 2019
2,565
5,570
146

Looks like Intel may still get to have the "most GHz" advantage.

View attachment 66127

Those aren't stable for any more than CPU-z/R23 though. Meh.

Cool gimmicks but I wouldn't run or advise anyone else to run those in a million years.
 

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
You had the answer written down. It's a combination of marginal process refinements moving from 10nm+++ to 10nm++++ and spending more die space to add 8 cores.

In R23, a 5800X scores over 15K (~108pts/W) and a 5950X scores over 25K (~178pts/W). Same architeture, same process, same power consumption, but over 60% higher score. The 5995WX scores over 66K (~237pts/W). Same architecture, same process, 100% more power consumption but 160% higher score. It should be fairly obvious why that is and why Raptor Lake has better performance per watt than Alder Lake.

The 5950x has similar power consumption to the 5800x in Cinebench because it's all core boost is substantially lower; 3.9ghz vs 4.5ghz for the 5800x at stock. Zen 3's operating sweet spot in terms of performance per watt is the mid to upper 3ghz range.

The 13900K on the other hand doubled the efficiency cores, significantly increased the core frequencies, expanded the L2 and L3 cache for both performance and efficiency cores without a substantial increase in the power draw, which to me is impressive.