Question Raptor Lake - Official Thread

Page 168 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Abwx

Diamond Member
Apr 2, 2011
9,868
2,313
136
So you're saying he should change his name to CarpFrameX given how fishy his numbers are?

:p
Well, PCGH is a quite commercial site whose part of the content is behind a paywall, they get quite some money from announcers among others Intel, as a hint they got a 13900KS to play with while Computerbase apparently got nothing despite good relationships with Intel as well (since they got a dual Xeon 8xxx server for free)...
 

Mopetar

Diamond Member
Jan 31, 2011
7,096
4,540
136
Almost no one got a KS because it's really not worth buying over a 13900K unless you absolutely want that last ~2% performance and don't care about the cost. Why widely sample something for review if you know that the result will be a collective "meh"?
 

jpiniero

Lifer
Oct 1, 2010
12,811
4,098
136
Almost no one got a KS because it's really not worth buying over a 13900K unless you absolutely want that last ~2% performance and don't care about the cost. Why widely sample something for review if you know that the result will be a collective "meh"?
Page hits?
 

Hulk

Diamond Member
Oct 9, 1999
3,680
1,275
136
3 bins on fully operational 13900's when you consider the KS, the K, and the non K.
They should just make like 10 bins and price accordingly. 13900A (choice, the best), 13900B (next best), etc... It would stop a lot of the returns to get "a good one."
 

biostud

Lifer
Feb 27, 2003
17,103
2,980
126
3 bins on fully operational 13900's when you consider the KS, the K, and the non K.
They should just make like 10 bins and price accordingly. 13900A (choice, the best), 13900B (next best), etc... It would stop a lot of the returns to get "a good one."
There is also the KF :p
 

Herald

Member
Jan 23, 2023
123
57
56
Hmm, I thought that running only one company's product with overclocked memory and overclocked CPU while leaving the other at stock CPU and dog slow JEDEC memory was supposed to be evidence of clear bias and that their results couldn't be trusted. Seems no one has an issue with it in this case though. Strange o_O.
But he is testing the 13900k with just 5600, there is still a massive gap. You can even drop the 13900k, the 12900k with 4400 is basically matching the best zen 4 has to offer.

It is known that RPL > ALD >>> Zen 4 for gaming. Dunno why it's still contestable
 
Last edited:

Hulk

Diamond Member
Oct 9, 1999
3,680
1,275
136
13900K and hot spots. As most of you know I have a 13900K with a 280mm AIO. In some applications I can run nearly 250W and not throttle, like Cinebench or Prime 95, which seem to load/heat the CPU evenly. But in others, namely, Vegas Pro 19 while using Render+ to render I will throttle at 200W. Render+ frameserves the Vegas timeline to an external x265 encoder. It's a quite compute heavy process as Vegas is assembling the timeline, which is then served frame-by-frame in uncompressed video form to the external x265 encoder.

But apparently it loads some cores more than others and thus causes hotspots, which causes throttling.

I've tried a number of things to get to a "set it and forget it" approach so that I'm good for any application and I think I've found a good setting that I wanted to share here.

I have set by core usage to 5.8GHz for 2 cores, 5.5GHz for 4 cores, 5.2GHz for 6 cores, and 5.0GHz for 8 cores. I've also capped the E's at 4.0GHz. The reason being if an application is using more than the 8 P threads then it's probably using a lot of the 16 E's as well so I might as well keep them a little cooler for a small reduction in compute.

So it's working well. I can still put 37K in Cinebench, get to 5.5GHz when processing fx in Photoshop, and the CPU stays 86C max while rendering in Vegas Pro.

I currently have PL1 set at 225W but in with my current settings it only pulls about 180W in Vegas Pro, which if the cores were left "uncapped" they would throttle. But holding them to 5GHz stops that hotspot issue.
 

Hitman928

Diamond Member
Apr 15, 2012
4,375
5,763
136
But he is testing the 13900k with just 5600, there is still a massive gap. You can even drop the 13900k, the 12900k with 4400 is basically matching the best zen 4 has to offer.

It is known that RPL > ALD >>> Zen 4 for gaming. Dunno why it's still contestable
I think you missed the point of my post.

And no, your performance statement is not accurate in the slightest. Just saying it is, "known", doesn't make it true. Every outlet that I've seen that uses faster memory for both platforms shows that Zen4 is competitive at worst with RPL in gaming and several show Zen4 being pretty darn close (6-8% behind). Zen4 actually takes the lead when talking lower end or lower midrange SKUs (i.e. <$300). Many reviews have already been posted that support this in this thread, so I won't bother repeating them all again.
 
  • Like
Reactions: Schmide and Elfear

Herald

Member
Jan 23, 2023
123
57
56
I think you missed the point of my post.

And no, your performance statement is not accurate in the slightest. Just saying it is, "known", doesn't make it true. Every outlet that I've seen that uses faster memory for both platforms shows that Zen4 is competitive at worst with RPL in gaming and several show Zen4 being pretty darn close (6-8% behind). Zen4 actually takes the lead when talking lower end or lower midrange SKUs (i.e. <$300). Many reviews have already been posted that support this in this thread, so I won't bother repeating them all again.
In any non GPU bottlenecked review the difference is 25 to 30% between Zen 4 and RPL. I can post you a bunch of them, from purepc to club386. And I know cause ive tested myself, a tuned 7950x with 6200 tuned ram ties a stock 12900k with 6000 ram in averages in cp2077 but loses by 35% in minimums. And that's just in the built in benchmark that's relatively light. In game the difference is actually bigger.

But if you don't trust the reviews, I guess we can test it ourselves. I have both a 12900k and a 13900k paired with 7600c34 ram, anyone wants to post a cp2077 or a spiderman run with their zen 4, ill be glad to help.
 

Herald

Member
Jan 23, 2023
123
57
56
13900K and hot spots. As most of you know I have a 13900K with a 280mm AIO. In some applications I can run nearly 250W and not throttle, like Cinebench or Prime 95, which seem to load/heat the CPU evenly. But in others, namely, Vegas Pro 19 while using Render+ to render I will throttle at 200W. Render+ frameserves the Vegas timeline to an external x265 encoder. It's a quite compute heavy process as Vegas is assembling the timeline, which is then served frame-by-frame in uncompressed video form to the external x265 encoder.

But apparently it loads some cores more than others and thus causes hotspots, which causes throttling.

I've tried a number of things to get to a "set it and forget it" approach so that I'm good for any application and I think I've found a good setting that I wanted to share here.

I have set by core usage to 5.8GHz for 2 cores, 5.5GHz for 4 cores, 5.2GHz for 6 cores, and 5.0GHz for 8 cores. I've also capped the E's at 4.0GHz. The reason being if an application is using more than the 8 P threads then it's probably using a lot of the 16 E's as well so I might as well keep them a little cooler for a small reduction in compute.

So it's working well. I can still put 37K in Cinebench, get to 5.5GHz when processing fx in Photoshop, and the CPU stays 86C max while rendering in Vegas Pro.

I currently have PL1 set at 225W but in with my current settings it only pulls about 180W in Vegas Pro, which if the cores were left "uncapped" they would throttle. But holding them to 5GHz stops that hotspot issue.
This is a u12a at 330w in ycruncher

yccruncher.JPG

and in CBR23

41161.JPG.d242353c36f1f535810d3d35af6c246f.jpg
 

igor_kavinski

Diamond Member
Jul 27, 2020
7,598
4,394
106
And I know cause ive tested myself, a tuned 7950x with 6200 tuned ram ties a stock 12900k with 6000 ram in averages in cp2077 but loses by 35% in minimums.
I would believe that more if it was the 13900K due to its increased cache but 12900K beating a 7950X in minimums with that wide a margin? Hmm...

It could also be just specific to CP2077 and probably this behavior is not representative of other popular games.
 

Herald

Member
Jan 23, 2023
123
57
56
I would believe that more if it was the 13900K due to its increased cache but 12900K beating a 7950X in minimums with that wide a margin? Hmm...

It could also be just specific to CP2077 and probably this behavior is not representative of other popular games.
You can ask anyone with the 7950x, is struggles to break 70 fps in the built in benchmark, in order to hit anything close to 80 you need to tune the crap out of the ram. The 12900k easily hits 95+, and the 13900k isn't that far better with around 100-105 in minimums.
 

Hitman928

Diamond Member
Apr 15, 2012
4,375
5,763
136
In any non GPU bottlenecked review the difference is 25 to 30% between Zen 4 and RPL. I can post you a bunch of them, from purepc to club386. And I know cause ive tested myself, a tuned 7950x with 6200 tuned ram ties a stock 12900k with 6000 ram in averages in cp2077 but loses by 35% in minimums. And that's just in the built in benchmark that's relatively light. In game the difference is actually bigger.

But if you don't trust the reviews, I guess we can test it ourselves. I have both a 12900k and a 13900k paired with 7600c34 ram, anyone wants to post a cp2077 or a spiderman run with their zen 4, ill be glad to help.
Smells like an alt account but whatever.

You're talking about 1 - 2 games. I'm talking about on average across a large range of games. If we were to completely cherry pick, I could show the 7950x dominating a 13900k by over 20% in gaming.
 

Herald

Member
Jan 23, 2023
123
57
56
Smells like an alt account but whatever.

You're talking about 1 - 2 games. I'm talking about on average across a large range of games. If we were to completely cherry pick, I could show the 7950x dominating a 13900k by over 20% in gaming.
Im validating PCGH's numbers with those 1-2 games. If they are right about those 2 games, then they are probably right about the rest.
 

Herald

Member
Jan 23, 2023
123
57
56

Elfear

Diamond Member
May 30, 2004
7,064
583
126
In any non GPU bottlenecked review the difference is 25 to 30% between Zen 4 and RPL. I can post you a bunch of them, from purepc to club386. And I know cause ive tested myself, a tuned 7950x with 6200 tuned ram ties a stock 12900k with 6000 ram in averages in cp2077 but loses by 35% in minimums. And that's just in the built in benchmark that's relatively light. In game the difference is actually bigger.

But if you don't trust the reviews, I guess we can test it ourselves. I have both a 12900k and a 13900k paired with 7600c34 ram, anyone wants to post a cp2077 or a spiderman run with their zen 4, ill be glad to help.
Do you have a bunch of game review averages where RPL and ADL dominate Zen4?

https://forums.anandtech.com/threads/raptor-lake-official-thread.2599551/post-40930085 (could be somewhat GPU limited)

All of those posts in this thread show average performance across a mix of games. Taking the results from the fastest RPL and Zen4 chip in each review, the differences are below. Across all those reviews, the 13900K(S) was 5.5% faster than Zen4. If you take out the two biggest outliers (PCGH and Eurogamer), the average is 4.6% faster in favor of RPL.

1674497591544.png

Posting one or two games where RPL truly dominates Zen4 doesn't paint an accurate picture. Just like if people post up the few games where Zen4 dominates RPL. What reviews of a broad swath of games show is that gamers won't be able to tell a difference between RPL and Zen4. Both provide a great gaming experience. I'm even doubtful gamers will be able to tell a noticeable difference with Zen43D even though it will likely bring big performance gains.
 

In2Photos

Senior member
Mar 21, 2007
885
799
136
In any non GPU bottlenecked review the difference is 25 to 30% between Zen 4 and RPL. I can post you a bunch of them, from purepc to club386. And I know cause ive tested myself, a tuned 7950x with 6200 tuned ram ties a stock 12900k with 6000 ram in averages in cp2077 but loses by 35% in minimums. And that's just in the built in benchmark that's relatively light. In game the difference is actually bigger.

But if you don't trust the reviews, I guess we can test it ourselves. I have both a 12900k and a 13900k paired with 7600c34 ram, anyone wants to post a cp2077 or a spiderman run with their zen 4, ill be glad to help.
Oh yay, this argument again! This should be what, the 4th time this calendar year.
 

Herald

Member
Jan 23, 2023
123
57
56
Do you have a bunch of game review averages where RPL and ADL dominate Zen4?

https://forums.anandtech.com/threads/raptor-lake-official-thread.2599551/post-40930085 (could be somewhat GPU limited)

All of those posts in this thread show average performance across a mix of games. Taking the results from the fastest RPL and Zen4 chip in each review, the differences are below. Across all those reviews, the 13900K(S) was 5.5% faster than Zen4. If you take out the two biggest outliers (PCGH and Eurogamer), the average is 4.6% faster in favor of RPL.

View attachment 75182

Posting one or two games where RPL truly dominates Zen4 doesn't paint an accurate picture. Just like if people post up the few games where Zen4 dominates RPL. What reviews of a broad swath of games show is that gamers won't be able to tell a difference between RPL and Zen4. Both provide a great gaming experience. I'm even doubtful gamers will be able to tell a noticeable difference with Zen43D even though it will likely bring big performance gains.
Purepc uses heavy settings in heavy game scenes. And yes, ald and rpl dominate in those. Honestly im too bored to have a discussion about something that is just a fact, so if someone has a zen 4 he can step forward and we can test 5-10 games and see whats up.

I have both a 12900k and a 13900k with 7600c34 ram, so i can test both against a zen 4.
 

Carfax83

Diamond Member
Nov 1, 2010
6,781
1,480
136
I would believe that more if it was the 13900K due to its increased cache but 12900K beating a 7950X in minimums with that wide a margin? Hmm...

It could also be just specific to CP2077 and probably this behavior is not representative of other popular games.
I don't know why people are surprised that a 12900K would beat a 7950x in a CPU bound situation. The 12900K has a 6 wide core, 7950x has a 4 wide core. 12900K also has a ROB that is 60% larger than Zen 4's (512 vs 320).

ADL's main problem was that it lacked the memory and cache bandwidth to feed the cores, but this was fixed with RPL.... though not entirely. RPL still has an insatiable need for increased bandwidth to perform at full capacity.
 
  • Like
Reactions: Henry swagger

Carfax83

Diamond Member
Nov 1, 2010
6,781
1,480
136

View attachment 75176

Using DDR5-6000 CL30 for both systems, 7950X has no issue beating 12900K in BOTH minimum and average FPS.
The problem with that review is that it was done with a 6700xt. Much of the confusion about the extent to how much RPL or Zen 4 is faster in games is due to reviewers being GPU bottlenecked..... typically out of ignorance or incompetence.
 
  • Like
Reactions: Henry swagger

Hitman928

Diamond Member
Apr 15, 2012
4,375
5,763
136
The problem with that review is that it was done with a 6700xt. Much of the confusion about the extent to how much RPL or Zen 4 is faster in games is due to reviewers being GPU bottlenecked..... typically out of ignorance or incompetence.
So you're saying that the 7950x would be more than 10% faster than the 12900k in their test case if they used a faster GPU?
 

Herald

Member
Jan 23, 2023
123
57
56
So you're saying that the 7950x would be more than 10% faster than the 12900k in their test case if they used a faster GPU?
Im saying... Instead of arguing about reviewers, let's test them ourselves? Isnt there anyone with a zen 4 and a high end gpu? 😁

For example this guy is testing in one of the heaviest parts of the game, at 1440p + rt on + dlss performance, i can upload the same area with my stock 13900k and compare if you like me to

 

Harry_Wild

Senior member
Dec 14, 2012
805
143
106
Can you believe it, most of retailers are either sold out or marking up the MSRP on the non K 13th gen. CPU!😬☹

They claim excessive demand right now!🫣
 

ASK THE COMMUNITY