Question Raptor Lake - Official Thread

Page 101 - 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
4,227
2,015
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

cebri1

Member
Jun 13, 2019
126
133
116
So 13900K around 8-9% faster than the 7950X in CB23? Intel will put a lot of pressure on the AM5 platform if these results are translated to real world performance. AMD is just to expensive to justify.
 

Abwx

Lifer
Apr 2, 2011
10,953
3,472
136
So 13900K around 8-9% faster than the 7950X in CB23? Intel will put a lot of pressure on the AM5 platform if these results are translated to real world performance. AMD is just to expensive to justify.


At 335W and an AIO vs 220W air cooled 7950X, using the same air cooler as the latter it would run at 130°C...
 

nicalandia

Diamond Member
Jan 10, 2019
3,330
5,281
136
I do not play games, I do not have any discrete graphic card now. I want to get RX6600, but it is still not dropping in price for some reason.
So it's a HEDT/Workstation on a Budget? Yeah 13900K and budget would not seem to go hand to hand, but HEDT it's just too expensive at this time thanks to Intel Lagging behind.
 

cebri1

Member
Jun 13, 2019
126
133
116
At 335W and an AIO vs 220W air cooled 7950X, using the same air cooler as the latter it would run at 130°C...

Well it's running on an "older" node. Intel competing with a worse manufacturing process and even beating the competition (at the expense of efficiency) is good news for them. MTL-S should be a good test for Intel to see if they can regain a significant advantage over AMD again.
 

inf64

Diamond Member
Mar 11, 2011
3,702
4,027
136
Seems like with good cooling and DDR5 41K, maybe 42K, should be doable.
The 40K figure is achieved with auto-OC feature in the BIOS, the P cores are running at 5.5Ghz and E cores at 4.3Ghz. The P cores might have a bit more room left but I doubt it's worth it as the chip is already consuming 330W using these settings.
 
  • Like
Reactions: cebri1

inf64

Diamond Member
Mar 11, 2011
3,702
4,027
136
I was not hitting any thermal limit during the test - maximal temperature was 85°C. I also did not tune the CPU in any way. I believe I could get higher score with this CPU if I tried to overclock it, undervolt it, etc.
Could you set the PL2 to 250W and see what the stock performance is?

For reference, this is the stock Turbo spec according to intel : https://www.intel.com/content/www/u...-36m-cache-up-to-5-80-ghz/specifications.html

The unlimited mode is analog to PBO2 on 7950X (when using a premium cooling), but no review so far (to my knowledge) has tested with PBO2 enabled.
 

inf64

Diamond Member
Mar 11, 2011
3,702
4,027
136
If you tell me where it is, I looked in the BIOS and did not find it. It is 21:32 here already and I may do it tomorrow or over the weekend.

It should be called Power Limit or TDP limit, but I'm not sure to be honest. We had no BIOS leaks that show how it's called (probably depends on the motherboard maker).
 

deasd

Senior member
Dec 31, 2013
519
755
136
So 13900K around 8-9% faster than the 7950X in CB23? Intel will put a lot of pressure on the AM5 platform if these results are translated to real world performance. AMD is just to expensive to justify.
R15 is more like a median result than R20/R23 when summarized together with other realworld workload. Computerbase has a good 7950X review, R23 gives Intel more advantage than any other tests.

It's quite funny that so many people blindly trust MAXON software.
 

Wolverine2349

Member
Oct 9, 2022
157
67
61
So 13900K around 8-9% faster than the 7950X in CB23? Intel will put a lot of pressure on the AM5 platform if these results are translated to real world performance. AMD is just to expensive to justify.


Yeah that's true. Why do you think AMD has gotten so greedy when they had a small time with performance lead (Only Zen 3 whipping everything Intel for 1 year until Alder Lake dropped and retook the 8 or less P core performance crown and often trade blows with the hybrid arch and AMD's much better Zen 3 exra cores compared to Intel's Gracemont getting the assistance form the superior to Zen 3 Golden Cove cores) and they are smaller than Intel with less resources??

The fact Intel has better P cores at same clock speed even on Alder Lake (by a little bit anyways), yet AMD is charging more for a 7600X than the 12600K when the 7600X is 6 cores and 12600K has 6 P cores and 4 e-cores and e-cores are useful for some even though they are not to me. You can buy a 12600K, disable the e-cores and have a 6 core CPU just as good as the 7600X at least when it comes to IPC at same clock for clock or maybe a little better for INtel side. Likewise you can get the 12700K for cheaper ($350) and it has 8 P cores and 4 e-cores and even shutting off e-cores you have an 8 core CPU with slightly better or same IPC clock for clock than $400 7700X which only has 8 cores and no e-core option for those that find them useful.

What is AMD thinking now. They wanted to be the premium brand but there 8 P core parts are losing in single threaded stuff at same clock to Intel. Now at top end as they have 16 strong cores yeah they sometimes beat Intel and no hybrid compatibility. But at somewhat high and and mid range and 8 P cores or less counterparts they are so badly over charging.

Nevermind the motherboards are insanely priced too.

Perhaps is AMD charging a premium more for the platform costs than outright CPU performance cause outright CPU performance is close enough even if IPC is slightly worse it is negligible and pus the platform longevity of AM5 they are charging more for for giving an upgrade path to Zen 4 3D cache and then Zen 5 and maybe beyond?? Where as Intel Z790 and 13900KS is the best as it gets and no other CPU upgrades down the road.
 

nicalandia

Diamond Member
Jan 10, 2019
3,330
5,281
136
I found it. Here are the numbers at 160 and 250W power limits: 32 716 and 38 318. Temperatures hit 51 and 67°C. I did not observe frequencies.

So 38,000 at stock and 40,000 at unlimited power. Well that's about the same of what a 7950X gets...!

AMD has them beat at Performance/Efficiency. I think we can put that argument to rest for good.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,564
14,518
136
Well it's running on an "older" node. Intel competing with a worse manufacturing process and even beating the competition (at the expense of efficiency) is good news for them. MTL-S should be a good test for Intel to see if they can regain a significant advantage over AMD again.
I will wait for real benchmarks to see if it even beats Zen 4. There is no way it can maintain 337 watt usage for long. I did see he said he also got water cooling, but how big ? AIO ? Custom ? chilled ?

And this from the guy that started a thread saying the government should ban these "inefficient" high power draw CPU's.
 

Abwx

Lifer
Apr 2, 2011
10,953
3,472
136
I measured two more, here are the results:

View attachment 69082

View attachment 69083

View attachment 69084

It seems to be pretty efficient at lower power/frequency.

At 100W 13900K has the same score as 12900K had at 250W?

Intel said 13900K@65W is matching 12900K@250W, i once said that it was plain lie to mimick s AMD s stated perf/improvement and that it would require about 100W to match ADL...

As a comparison for your number the 5950X does 26150 pts@130W or so.
 

Hitman928

Diamond Member
Apr 15, 2012
5,313
7,957
136
I measured two more, here are the results:

View attachment 69082

View attachment 69083

View attachment 69084

It seems to be pretty efficient at lower power/frequency.

At 100W 13900K has the same score as 12900K had at 250W?

Very impressive increase for Intel in efficiency. Obviously adding 8 additional E-cores is probably the biggest reason, but it seems they were able to tweak the process pretty well also. It can basically match a 7950x on TSMC 5 at higher freqencies / power in this test. At lower frequencies/power the TSMC 5 node's advantage just becomes too much and the 7950x takes a sizeable efficiency/performance lead. For desktop, Raptor Lake is plenty for Intel, at least until whenever Zen 5 lands. Mobile and server is a different story, but for desktop purposes, both are looking like solid options.