Question Raptor Lake - Official Thread

Page 169 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Hitman928

Diamond Member
Apr 15, 2012
4,375
5,766
136
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

Alt account confirmed?
 

Hitman928

Diamond Member
Apr 15, 2012
4,375
5,766
136
I dont have a 2nd account if that's what you are asking. But im not sure what you are asking

You find something offensive with my suggestion, to actually test those cpus ourselves? 😊
I just find it strange that you just joined today but immediately jumped in to talk all about Intel CPUs and started using the exact same arguments and referring to the same videos we've seen from others recently. That and you just replied to my post as if I had quoted you when I had quoted a different user but maybe you were just interjecting yourself into a different conversation and changing the topic for some reason.

To answer your question, I find it extremely challenging to accurately compare performance in games without taking many measures to eliminate as many variables as possible and doing verification checks. All of which is extremely difficult to do with a random someone miles away across the internet. Even if we just accept that the 13900K curb stomps Zen4 in a couple of games, it's not really the discussion at hand as that is the very definition of cherry picking and it can easily go both ways. I can show a 7950x beating a 13900k by 25% in a heavy RT game. That doesn't mean the 7950x is the destroyer of gaming worlds and dominates all others in RT calculations.
 

Herald

Member
Jan 23, 2023
124
57
56
I just find it strange that you just joined today but immediately jumped in to talk all about Intel CPUs and started using the exact same arguments and referring to the same videos we've seen from others recently. That and you just replied to my post as if I had quoted you when I had quoted a different user.

To answer your question, I find it extremely challenging to accurately compare performance in games without taking many measures to eliminate as many variables as possible and doing verification checks. All of which is extremely difficult to do with a random someone miles away across the internet. Even if we just accept that the 13900K curb stomps Zen4 in a couple of games, it's not really the discussion at hand as that is the very definition of cherry picking and it can easily go both ways. I can show a 7950x beating a 13900k by 25% in a heavy RT game. That doesn't mean the 7950x is the destroyer of gaming worlds and dominates all others in RT calculations.
Although i agree with what you said, if we test a couple of games and validate pcgh numbers, then that means their numbers are actually accurate and not made up.

From the video i posted, a 13900k is 25-30% ahead. Which is basically inline with pcghs review.

Which rt game is the 7950x 25% ahead? Really interested.
 

Hitman928

Diamond Member
Apr 15, 2012
4,375
5,766
136
Although i agree with what you said, if we test a couple of games and validate pcgh numbers, then that means their numbers are actually accurate and not made up.

From the video i posted, a 13900k is 25-30% ahead. Which is basically inline with pcghs review.

Which rt game is the 7950x 25% ahead? Really interested.
According to Eurogamer, Metro Exodus Enhanced Edition which is RT only, no pre-baked lighting in the game at all.

1674507193985.png

 
  • Like
Reactions: Herald

Herald

Member
Jan 23, 2023
124
57
56
According to Eurogamer, Metro Exodus Enhanced Edition which is RT only, no pre-baked lighting in the game at all.

View attachment 75188

That would be impressive... If zen 3 wasnt also winning against RPL 😁

I assume the game doesnt like ecores? even the i5 is 6% ahead, lol
 

Hulk

Diamond Member
Oct 9, 1999
3,680
1,275
136
This is a u12a at 330w in ycruncher

View attachment 75173

and in CBR23

View attachment 75172
Those are some mighty fine numbers especially considering you're on air.
What is your Vcore as indicated in HWinfo when running CB?

Couple things to keep in mind though. I also have that air cooler and my 280mm does better. Since I have both I have compared them. Your better result than mine can be explained by a combination of the following.

1. You have a KF CPU, which generates a little less heat due to not having the GPU on board.
2. I would bet your CPU has a higher SP rating than mine.
3. CB and other benchmarks tend to heat the CPU more evenly than many applications. I bet Vegas Pro Render+ would smoke your CPU like it does mine.
4. I would bet you are better at underclocking/fine tuning your rig than I am.
 

Herald

Member
Jan 23, 2023
124
57
56
Those are some mighty fine numbers especially considering you're on air.
What is your Vcore as indicated in HWinfo when running CB?

Couple things to keep in mind though. I also have that air cooler and my 280mm does better. Since I have both I have compared them. Your better result than mine can be explained by a combination of the following.

1. You have a KF CPU, which generates a little less heat due to not having the GPU on board.
2. I would bet your CPU has a higher SP rating than mine.
3. CB and other benchmarks tend to heat the CPU more evenly than many applications. I bet Vegas Pro Render+ would smoke your CPU like it does mine.
4. I would bet you are better at underclocking/fine tuning your rig than I am.
These are stock numbers with no undervolting, i just lowered light load from default 9 to 6 (which i guess counts as undervolting). Dont have a particularly good sample, p cores are fine (can run 5.8 all core) but ecores are disastrously bad 😁

Was thinking about buying a new one but this one has a decent IMC so I decided to stick with it

EG1. YCRUNCHER is wrong picture, just noticed i posted my 12900k numbers.
 

Herald

Member
Jan 23, 2023
124
57
56
Well i dont know what to tell you. A maxed out 12900k at 5.4 ghz and 6000c30 tuned ram loses to a stock 13900k with 7600c34 ram by about 12 to 15%. Said 12900k is faster than any zen 4, so imagine what the actual difference is between zen 4 and 13900K

Now regarding ald > zen 4, i have a 12900k, we can test this if you'd like
 

Carfax83

Diamond Member
Nov 1, 2010
6,781
1,480
136
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?
No, I'm saying that the 12900K will likely pull ahead when the situation is CPU bound, ie when maxed crowd settings are used and when RT is active.
 

Carfax83

Diamond Member
Nov 1, 2010
6,781
1,480
136
These are stock numbers with no undervolting, i just lowered light load from default 9 to 6 (which i guess counts as undervolting). Dont have a particularly good sample, p cores are fine (can run 5.8 all core) but ecores are disastrously bad 😁

Was thinking about buying a new one but this one has a decent IMC so I decided to stick with it

EG1. YCRUNCHER is wrong picture, just noticed i posted my 12900k numbers.
Nice system dude. Have you ever done an Aida64 cache and memory bandwidth test? I'm currently tuning my memory, so I'm eager to compare my scores to someone with a similar system. I have my RAM at DDR5 7800 right now, but I'm still testing stability. I'll post a screenshot when I get home as I'm at work right now.
 

Herald

Member
Jan 23, 2023
124
57
56
Nice system dude. Have you ever done an Aida64 cache and memory bandwidth test? I'm currently tuning my memory, so I'm eager to compare my scores to someone with a similar system. I have my RAM at DDR5 7800 right now, but I'm still testing stability. I'll post a screenshot when I get home as I'm at work right now.
Off the top of my head, 121gbs read and 52ns latency. Clocking cache to 5ghz should give me another 1.5ns drop in latency roundabout, but i prefer running 13900k at stock.
 

Carfax83

Diamond Member
Nov 1, 2010
6,781
1,480
136
Well that's certainly nonsense. RPL and Zen 4 are generally competitive, with the 13900k vs 7950x appearing to be a slight victory for Intel, but ADL >>> Zen 4? Lol, no.
As I recall, Zen 4 never decisively beat ADL in gaming. The 12900KS was still edging it out, especially if it was paired with DDR5 6400.
 
  • Like
Reactions: Henry swagger

Carfax83

Diamond Member
Nov 1, 2010
6,781
1,480
136
Off the top of my head, 121gbs read and 52ns latency. Clocking cache to 5ghz should give me another 1.5ns drop in latency roundabout, but i prefer running 13900k at stock.
How the hell is your latency that low? I can't get mine to drop below 60ns. Have you tuned the secondary and tertiary timings as well? I think the reason my latency is still 60ns is because I've only messed with the primaries. I've also never touched the uncore.
 

Herald

Member
Jan 23, 2023
124
57
56
How the hell is your latency that low? I can't get mine to drop below 60ns. Have you tuned the secondary and tertiary timings as well? I think the reason my latency is still 60ns is because I've only messed with the primaries. I've also never touched the uncore.
Havent tuned a lot yet, just primaries trefi and trfc. Could drop it to around 47ns with more aggressive tuning but since i dont have any cooling on my ram i dont want to sacrifice stability.

Trefi and trfc should get you at 55ns or below. You can run trefi maxed out but id rather play it on the safe side and use 65535, especially considering i dont use any cooling for the ram.

What frequencies and primaries are you running? Im at 7600 34 44 44 34
 
  • Love
Reactions: Carfax83

Carfax83

Diamond Member
Nov 1, 2010
6,781
1,480
136
Trefi and trfc should get you at 55ns or below. You can run trefi maxed out but id rather play it on the safe side and use 65535, especially considering i dont use any cooling for the ram.
I'm going to look for those settings in the UEFI when I go home. So do you suggest 65535 for the Trefi for an air cooled system? What about the trfc?

What frequencies and primaries are you running? Im at 7600 34 44 44 34
Currently at 7800 36-44-44-44-92.
 

Herald

Member
Jan 23, 2023
124
57
56
I'm going to look for those settings in the UEFI when I go home. So do you suggest 65535 for the Trefi for an air cooled system? What about the trfc?



Currently at 7800 36-44-44-44-92.
Ive no idea with 7800, you can try higher trefi and test it. Trefi increases dimm temperatures a LOT, after 60c usually ram starts losing stability, so you need to keep them below that.

Then you need to play with your rtls.

But is it worth it? I could spend a couple of days and tune tertiaries to drop to 50ns but whatever, my 4090 is still a massive bottleneck in games 😂
 

Carfax83

Diamond Member
Nov 1, 2010
6,781
1,480
136
Ive no idea with 7800, you can try higher trefi and test it. Trefi increases dimm temperatures a LOT, after 60c usually ram starts losing stability, so you need to keep them below that.
I just want to drop into the 50s, ideally 55ns to 58ns. When I first bought my Hynix A die memory kit, I thought it would be easy to get into the 50ns range due to the high frequencies, but that seems to not be possible without messing with the other sub timings.

Even at 8ghz, still at 60ns! :rolleyes:

But is it worth it? I could spend a couple of days and tune tertiaries to drop to 50ns but whatever, my 4090 is still a massive bottleneck in games 😂
Yeah I agree it's not worth it. RPL definitely prefers bandwidth over latency (unlike Zen 4) as the massive 512 ROB can hide the memory latency.
 

Timmah!

Golden Member
Jul 24, 2010
1,172
364
136
It crossed my mind today that Ashes of Singularity is no more part of latest gaming benchmarking suites, at least i dont recall seeing any of the new CPUs and GPUs being tested in it. What´s up, is it too dated at this point to be even useful as a performance benchmark, even if nobody really plays it anymore? Cause it was never ultra-popular game, got nowhere near Supreme Commander, but it was highly threaded and whatnot, yet CPUs/GPUs used to be tested in it a lot.
 

ASK THE COMMUNITY