• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

AMD Ryzen 5000 Builders Thread

Page 19 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Justinus

Platinum Member
Oct 10, 2005
2,753
866
136
Finally done with the clockspeed tweaking :)
PBO used for everything.

Cinebench r23:

  • Singlethread = 1693
  • Multithread = 31124

Cinebench r20:

  • Singlethread = 661
  • Multithread = 12103

Cinebench r15:

  • Singlethread = 281
  • Multithread = 5262

Geekbench 5 @ https://browser.geekbench.com/v5/cpu/5247965

  • Singlethread = 1809
  • Multithread = 19615

Also some CPUZ scores in screenshot
Benching done on ~19-20 degrees ambiant.

Now i just need to wait for the MSI GeForce RTX 3090 SUPRIM X on its way in the mail, before i'm ready for cyberpunk :)
What kind of cooling are you using? I see peak tdie temps of just over 80C in your HWInfo.

Also, I have been struggling to learn how to configure PBO to get higher singlecore boosts. My first attempts brought basically zero performance improvements. Any tips?
 

Det0x

Senior member
Sep 11, 2014
558
810
136
What kind of cooling are you using? I see peak tdie temps of just over 80C in your HWInfo.

Also, I have been struggling to learn how to configure PBO to get higher singlecore boosts. My first attempts brought basically zero performance improvements. Any tips?
For cooling iam using custom EK watercooling, also have a TechN waterblock on its way im looking forward to test, should be notable difference.

Curve optimizer is used to maximize performance on the new 5000 series cpu, but it can be tricky to get everything right..
You can read more here:

https://www.reddit.com/r/Amd/comments/jrsz7x
 

Justinus

Platinum Member
Oct 10, 2005
2,753
866
136
For cooling iam using custom EK watercooling, also have a TechN waterblock on its way im looking forward to test, should be notable difference.

Curve optimizer is used to maximize performance on the new 5000 series cpu, but it can be tricky to get everything right..
You can read more here:

https://www.reddit.com/r/Amd/comments/jrsz7x
I look forward to seeing how the TechN block does for you. I have an Optimus lodged somewhere in the holiday transit mess.
 
  • Like
Reactions: Det0x

Kenmitch

Diamond Member
Oct 10, 1999
8,504
2,245
136
Sadly yes, anything above 3600 MT/s requires GDM on my setup with 4 memory sticks.
What's a couple of seconds anyways!

I just ordered 2 sets of the G.SKILL Trident Z Neo F4-3600C16D-16GTZN for my rig. Maybe they'll be cooperative and OC to at least 3600 CL14 with tight timings. I'll shoot for 3800MHz and see what happens, but default clocks are good enough anyways.

I wanted 32GB's and I guess the RGB bling. I usually just set a default in a softer white with my RGB's and exit the required app. No auto starting of any RGB software for me. Works best in the end and is easy on the eyes at night....Bling is just a couple of clicks away if needed.
 

Kocicak

Senior member
Jan 17, 2019
409
321
106
I just installed a 5900X and I have a weird feeling of confusion and unreal.

I put a standard Freezer 34 cooler on it (costs cca 25 Euro), the CPU rips through Cinebench, cooler fan does not even spin up and is INAUDIBLE. The computer sounds like nothing is happening but the CPU runs the highest possible load. How is this possible???

I did nothing to the settings and it runs at 4150 MHz at 1.13V full load, temperature is 65 °C. I did not read too many reviews so I do not know if this is typical?
 
  • Like
Reactions: lightmanek

Justinus

Platinum Member
Oct 10, 2005
2,753
866
136
I just installed a 5900X and I have a weird feeling of confusion and unreal.

I put a standard Freezer 34 cooler on it (costs cca 25 Euro), the CPU rips through Cinebench, cooler fan does not even spin up and is INAUDIBLE. The computer sounds like nothing is happening but the CPU runs the highest possible load. How is this possible???

I did nothing to the settings and it runs at 4150 MHz at 1.13V full load, temperature is 65 °C. I did not read too many reviews so I do not know if this is typical?
On my U12A, my 5950x at stock runs around 3800-3900MHz all core in cinebench at 1.0v, it doesn't even reach 55C.
 
  • Like
Reactions: lightmanek

Det0x

Senior member
Sep 11, 2014
558
810
136
Stage 2 memory tuning is done and i have dialed in my new 24/7 settings.
Happy with the results since this is a dual CCX cpu with 4 memory sticks :)

Did a quick memtest 100% task scope run to make sure everything is stable.
memtest 100 24-7 settings.png
 
  • Like
  • Love
Reactions: lightmanek and IEC

Elfear

Diamond Member
May 30, 2004
7,040
558
126
Stage 2 memory tuning is done and i have dialed in my new 24/7 settings.
Happy with the results since this is a dual CCX cpu with 4 memory sticks :)

Did a quick memtest 100% task scope run to make sure everything is stable.
View attachment 35456
Looking at your previous settings, it looks like volts have gone up (~1.27V --> ~1.34V average) and max clocks have gone down (5.1Ghz on some cores --> 4.8-4.9Ghz). Were your previous settings not very stable for 24/7 use?

Thanks for posting all your results by the way. I'll start tweaking my 5950 soon and you've helped with some great starting tips.
 

Det0x

Senior member
Sep 11, 2014
558
810
136
Looking at your previous settings, it looks like volts have gone up (~1.27V --> ~1.34V average) and max clocks have gone down (5.1Ghz on some cores --> 4.8-4.9Ghz). Were your previous settings not very stable for 24/7 use?

Thanks for posting all your results by the way. I'll start tweaking my 5950 soon and you've helped with some great starting tips.
I started hwinfo 11-12mins into the memtester run.. hwinfo was recording the ~4650-4700mhz effective clock on all cores when i was running memtest. (memtester run took 47min and hwinfo have only been running for 38min when i took the screenshot.
Look at the high average boost numbers a few mins after the run was over..

Aida which i ran before memtester show 5.1ghz boost.

In regards to volts = memtester only use ~90w on my setup while cinebench use ~230w

Settings and boost are the same as before :)
 

phillyman36

Golden Member
Jun 28, 2004
1,731
148
106
My Evga Rtx 3080 FTW3 came in. I just need a motherboard and im set. I want the Dark Hero but may have to grab a Hero Wifi. And of course Microcenter St davids had 5 Dark hero mobo. They never updated the website as far as stock. @5:15pm they had 1 left. I wouldnt make it in time to grab it. F my life lol
 

Elfear

Diamond Member
May 30, 2004
7,040
558
126
I started hwinfo 11-12mins into the memtester run.. hwinfo was recording the ~4650-4700mhz effective clock on all cores when i was running memtest. (memtester run took 47min and hwinfo have only been running for 38min when i took the screenshot.
Look at the high average boost numbers a few mins after the run was over..

Aida which i ran before memtester show 5.1ghz boost.

In regards to volts = memtester only use ~90w on my setup while cinebench use ~230w

Settings and boost are the same as before :)
Gotcha. That makes more sense.

Would you mind posting the settings you have in the Ryzen Curve Optimizer?
 

Kocicak

Senior member
Jan 17, 2019
409
321
106
I tested the 5900x a little bit more with R23, here is what I get:

1T - 4850MHz at 1,46V
4T 4700 1,40
8T 4600 1,35
12T 4400 1,25
24T 4100 1,125

Temperatures max 70°C.

Is this normal or a little weaker sample? With the 5800X I had previously when running R23 1T I was getting 4850 as well, that was 150MHz over specs, this one just goes 50 MHz over. With lighter load it goes up to 4950 MHz on one chiplet, the other one is slower.

5900boost.png
 
Last edited:

reqq

Member
Feb 26, 2020
27
33
51
I tested the 5900x a little bit more with R23, here is what I get:

1T - 4850MHz at 1,46V
4T 4700 1,40
8T 4600 1,35
12T 4400 1,25
24T 4100 1,125

Temperatures max 70°C.

Is this normal or a little weaker sample? With the 5800X I had previously when running R23 1T I was getting 4850 as well, that was 150MHz over specs, this one just goes 50 MHz over. With lighter load it goes up to 4950 MHz on one chiplet, the other one is slower.
Mine does that aswell on stock.. like 4.1-4.15.. then enable stock pbo it goes to 4.4 or something on 24T.. but then uses 1.3..
 

Kocicak

Senior member
Jan 17, 2019
409
321
106
Mine does that aswell on stock.. like 4.1-4.15.. then enable stock pbo it goes to 4.4 or something on 24T.. but then uses 1.3..
I tried turning PBO on and it does exactly the same thing, but due to high temperature (85°C) it drops to 4350 after a while.
 

Noid

Platinum Member
Sep 20, 2000
2,298
107
106
I put a standard Freezer 34 cooler on it ... How is this possible???
... is typical?
Yes, it's typical.
Heat Pipe tech is very effective.
My Noctua fans rarely " ramp up "
Even on Prime95 for 12 hours.

and , it's silent. ... shhhhh

( but a LARGE computer case is needed - they are enormous coolers )
 

Justinus

Platinum Member
Oct 10, 2005
2,753
866
136
I've been having a lot of trouble with PBO, and I'm not sure why. When I get a chance over Christmas break to watch some videos and learn a bit more I am going to give it another go.

It seems like no matter how or what configuration I enable PBO, it causes system instability. Crashing during cinebench singlethread, cinebench erroring, windows blue screening, failure to post.

I tried simply enabling PBO on auto limits with +0 MHz and it errors cinebench. Trying setting limits and a frequency offset causes the blue screens or worse.

Running a 5950x on a Dark Hero. Maybe PBO is not working correctly due to the MB being brand new, needing to get some more mature UEFI. Or maybe there's some secret sauce I simply don't know.

On the bright side I have 3800CL16 12 pass memtest stable even though it's some old 3200cl14 tridentz that's not on the QVL. Small victories, I guess.
 

ASK THE COMMUNITY