[BNSoul]

Thank you! It won’t show up if you have anything other than balanced power plan selected in the old Power options plan menu. I selected balanced there and then high performance in the drop down. Who knows if it’ll make an difference. It I can’t get more than 4350, sometimes 4400 all core in c23. This is on a good custom loop, -30 all core and 77c temps max

With that temp your CPU shouldn't be throttling, it must be something else, have you tried the task I posted above where 4 cores boost past 4450 at the same time? Can you run it along HWInfo64 running on the side and take a screenshot of the cores running?

 

[bmagnien]

With that temp your CPU shouldn't be throttling, it must be something else, have you tried the task I posted above where 4 cores boost past 4450 at the same time? Can you run it along HWInfo64 running on the side and take a screenshot of the cores running?

not sure if I did this right. I selected cores 0-3 but it looks like only 0 and 1 are boosting?

 

[bmagnien]

I got cores 4-7 boosting to 4525 (screenshot shows 4514 but just didnt capture the right moment

and here's cores 0-3 but they would only go to 4470

 

[BNSoul]

...

Thanks for the screenshots, voltages and temps look okay, are you running this full stock? have you tried PBO2 Tuner with generic values such as 120 80 120 and a curve of -15 for best cores and -25 for the remaining six? What motherboard, BIOS and AGESA is your X3D running on?
In your last screenshot it shows all cores did boost but #2 and #3 are having a hard time, can you run the app "BoostTesterMannix.exe" by overclocker @Mannix ? Do it with HWInfo on the side and we'll see if the light load triggers a full boost in all available cores.

 

[bmagnien]

In other news, I think I just got the highest score I've gotten using my high efficiency 'optimized' settings:

 

[bmagnien]

...

I'm running -30 all core (which passed all per-core stability tests), and those ycruncher screenshots were at stock ppt, edc, tdc, but I normally run the efficient setting i came up with posted above that many people are running with on this thread. 1.2.0.7 on asus x570i. BoostTesterManix screenshot below, pretty weak IMO, nothing over 4540

 

[BNSoul]

I'm running -30 all core (which passed all per-core stability tests), and those ycruncher screenshots were at stock ppt, edc, tdc, but I normally run the efficient setting i came up with posted above that many people are running with on this thread. 1.2.0.7 on asus x570i. BoostTesterManix screenshot below, pretty weak IMO, nothing over 4540

not bad, 4540 is the sign of a good CPU sample if you get all cores to that point, I get anywhere from 4542 to 4544, sometimes 4545 on my humble Aorus Elite X570 board. All cores above 4542 though, but I got a really good sample, not gonna say Platinum because nobody really knows but really good considering its consistent great results on an average board and RAM kit.

Screenshots:
So according to HWInfo author, in the "Ryzen Snapshot" monitoring mode all clocks on the top half of the HWInfo panel are indeed effective, real and actual clock speeds attained by the cores during live monitoring, always affected negatively by the "viewer" or "spectator effect" but pretty accurate.
For those who don't know this is "Ryzen snapshot mode"

And here my results in a quick, dirty run using my day to day PBO2 settings, not gonna bother with -30 all core sin I just use that for Cinebench

all cores attaining actual, effective speeds higher than 4542, best core gets 4544 real close to full effectiveness / efficiency, I've seen it get 4545.5 twice.

and for reference so you can compare this was my last R23 run:

all results at 100.0MHz stock BCLK + PBO2 Tuner limits (of course)

so you still got some work to do (fun with y-cruncher and core cycler) optimizing that curve!! cheers mate.

edit: bonus Y-cruncher boosting 4 cores simultaneously, measured right with HWiNFO "Ryzen snapshot mode" (showing actual registered core speeds)

 

[modee79s]

Hi guys,
My Cinebench R23 score is 13,100-200; how can I improve it to match the average?
I have the latest Bios version from Asus and only applied the DOCP profile; otherwise, everything else is at default.
This is my first AMD PC and I am not sure what Bios tweaks will be applied. Could you guide me in applying the standard Bios settings (if any)?
*I am a beginner and not confident in using the PBO2 Tuner so I would instead stick with the basic settings.

I appreciate any help you can provide.

**My system (Win 10):
5800X3D
Case: Lian-li O11
MB: ROG Strix B550-F Gaming
Cooling: 360mm AIO + 6 case fans
RAM: 32GB DDR4 3600 CL17

 

[BNSoul]

Hi guys,
My Cinebench R23 score is 13,100-200; how can I improve it to match the average?

First of all read the opening post where the optimized BIOS settings are stated. Same for the PBO2 Tuner app: instructions, purposes and guidelines, tips and tricks as well as download links (provided by the author himself) have also been extensively discussed throughout this thread.

Your score is at the rock bottom compared to what this CPU can actually produce, it's no problem but it means you would need to catch up with the very basics. Having someone sprouting over and over the same long explanatory posts every time a below average Cinebench result gets posted would be considered spam.

All the info, examples and troubleshooting advices are in this thread. Get the latest AGESA 1.2.0.7 compatible BIOS for your motherboard, AMD chipset drivers, PBO2 Tuner (generic settings to start with will be 120 /85 /120 with a curve of -15 for the best cores and -25 for all the rest, then you can fine tune later), HWInfo64, Mannix' boost tester, Core Cycler and Y-Cruncher. Then start reading and tweaking to your heart's content. Make sure the CPU is not overheating/throttling during Cinebench runs (certainly it sure looks like your CPU does), if there's a problem with the cooling solution get it fixed before anything else.

If you get stuck with anything of the above you can always ask but try to provide some context the likes of a HWInfo64 screenshot or similar. Also, do mind that this is a forum for people who have their setup fine tuned to the point that they are willing to go further and overclock the CPU to extract every last drop of performance. It also serves to get your CPU fully maximized at stock settings, I guess for the time being you'd be interested in the latter. Good luck 👍

 

[lestatdk]

Hi guys,
My Cinebench R23 score is 13,100-200; how can I improve it to match the average?
I have the latest Bios version from Asus and only applied the DOCP profile; otherwise, everything else is at default.
This is my first AMD PC and I am not sure what Bios tweaks will be applied. Could you guide me in applying the standard Bios settings (if any)?
*I am a beginner and not confident in using the PBO2 Tuner so I would instead stick with the basic settings.

I appreciate any help you can provide.

**My system (Win 10):
5800X3D
Case: Lian-li O11
MB: ROG Strix B550-F Gaming
Cooling: 360mm AIO + 6 case fans
RAM: 32GB DDR4 3600 CL17

That's a very low score. Have you monitored the temperatures ? To make sure it's not throttling ?

 

[AXi0M]

not bad, 4540 is the sign of a good CPU sample if you get all cores to that point, I get anywhere from 4542 to 4544, sometimes 4545 on my humble Aorus Elite X570 board. All cores above 4542 though, but I got a really good sample, not gonna say Platinum because nobody really knows but really good considering its consistent great results on an average board and RAM kit.

Screenshots:
So according to HWInfo author, in the "Ryzen Snapshot" monitoring mode all clocks on the top half of the HWInfo panel are indeed effective, real and actual clock speeds attained by the cores during live monitoring, always affected negatively by the "viewer" or "spectator effect" but pretty accurate.
For those who don't know this is "Ryzen snapshot mode"

View attachment 2568631


And here my results in a quick, dirty run using my day to day PBO2 settings, not gonna bother with -30 all core sin I just use that for Cinebench

all cores attaining actual, effective speeds higher than 4542, best core gets 4544 real close to full effectiveness / efficiency, I've seen it get 4545.5 twice.

View attachment 2568632


View attachment 2568633

and for reference so you can compare this was my last R23 run:

View attachment 2568634

all results at 100.0MHz stock BCLK + PBO2 Tuner limits (of course)

so you still got some work to do (fun with y-cruncher and core cycler) optimizing that curve!! cheers mate.

edit: bonus Y-cruncher boosting 4 cores simultaneously, measured right with HWiNFO "Ryzen snapshot mode" (showing actual registered core speeds)

View attachment 2568636

View attachment 2568637

i see you have a bios with agesia 1.2.0.7. With my ASUS x570-PRIME board running 1.2.0.7 my voltages underload go UP even with a -30 all core curve (1.212v-1.231v) . running the previous bios with 1.2.0.6b i get similar voltages as you in cinebench ~1.16v with an all core offset -30. i guess ASUS messed up the 1.2.0.7 bios?

 

[BNSoul]

i see you have a bios with agesia 1.2.0.7. With my ASUS x570-PRIME board running 1.2.0.7 my voltages underload go UP even with a -30 all core curve (1.212v-1.231v) . running the previous bios with 1.2.0.6b i get similar voltages as you in cinebench ~1.16v with an all core offset -30. i guess ASUS messed up the 1.2.0.7 bios?

Hi there mate, yeah the 1.2.0.7 update on my Aorus X570 Elite did improve voltages (meaning, they are now lower across the board), temperatures ¨(now I won't go over 71-72º under heavy stress testing in the middle of the hot Spanish summer weather, with CPU usually running games at mid 40's and below, idling at 26-28º). Performance also increased accordingly among other things like the fTPM issues which were also addressed. In particular, single core boosts are now like really good, it's a basic board but with good VRMs. Power deviation was also further calibrated and under heavy load is in the 101-101.6% range which is pretty good too. This is not all the work of Gigabyte since by default AGESA 1.2.0.7 microcode does indeed address all of the above.

What I mean with all that is the fact that you should be seeing improvements after updating to 1.2.0.7 instead of performance regressions, in this case I wouldn't hesitate to complain directly to your local ASUS rep. Have you checked with different owners of the same board? are the issues replicable? Because if it's just you or a few users you could try a full CMOS reset and re-flashing the latest BIOS following by a clean/fresh OS install, always make sure to not have PBO2 Tuner running / PBO2 Tuner system service installed before removing or installing AMD chipset drivers. What's your CB23 score and your boost speeds with Mannix' ? Try HWiNFO64 Ryzen snapshot mode for improved readings.

 

[AXi0M]

Hi there mate, yeah the 1.2.0.7 update on my Aorus X570 Elite did improve voltages (meaning, they are now lower across the board), temperatures ¨(now I won't go over 71-72º under heavy stress testing in the middle of the hot Spanish summer weather, with CPU usually running games at mid 40's and below, idling at 26-28º). Performance also increased accordingly among other things like the fTPM issues which were also addressed. In particular, single core boosts are now like really good, it's a basic board but with good VRMs. Power deviation was also further calibrated and under heavy load is in the 101-101.6% range which is pretty good too. This is not all the work of Gigabyte since by default AGESA 1.2.0.7 microcode does indeed address all of the above.

What I mean with all that is the fact that you should be seeing improvements after updating to 1.2.0.7 instead of performance regressions, in this case I wouldn't hesitate to complain directly to your local ASUS rep. Have you checked with different owners of the same board? are the issues replicable? Because if it's just you or a few users you could try a full CMOS reset and re-flashing the latest BIOS following by a clean/fresh OS install, always make sure to not have PBO2 Tuner running / PBO2 Tuner system service installed before removing or installing AMD chipset drivers. What's your CB23 score and your boost speeds with Mannix' ? Try HWiNFO64 Ryzen snapshot mode for improved readings.

Im pretty sure its the bios, i flashed back and fourth between 1.2.0.6b and 1.2.0.7 a few times now and the voltages always seem to be 0.05v higher on 1.2.0.7

 

[Cuboy]

How are you not getting 4450 all-core? Also, CPU-Z or Cinebench are not a metric for anything meaningful really, clock frequency is just another variable to be factored in, be it in single or multi core results. The X3D can be sitting at the bottom of the benchmarks and still win by a considerable margin in Far Cry, Assetto C, Factorio, UE4 games, Tarkov, Path of Exile, World of Warcraft, CoD Warzone, Watchdog Legions, Division 2, Star Citizen, BF, Riftbreaker, Valorant, GTA and many, many others that I haven't tested personally but I've seen in reputable benchmarks.

Most of the above titles cannot be matched in performance by other Zen 3 CPUs just by upping frequencies, that's why I still got my reservations with regard to the 7600-7700 vs 5800X3D debate, of course there's also the IPC and L2 increase along with DDR5 but I'm sure in some titles it won't be enough.

In this sense, unless the 7800X3D brings some killer feature (such as no more hotspots in the CPU due to unoptimized interposer) I don't think a 5800X3D user should upgrade to it. A new GPU should be much cheaper and equally as effective.

That is what convinced me when i got the 5800x3d.

No motherboard, ram upgrade.
No waiting.
Less pricy.
And more performance out of the game i want to play.

There is always going to be something better, a 7800x3d will be better than a 5800x3d of course, but it also will be more pricy and i'll need to upgrade everything. I prefer to extend the life of my pc, and maybe upgrade when zen 5 is out if the performance jump is significant.

For now, i just have to worry about getting a better gpu.

 

[modee79s]

First of all, read the opening post where the optimized BIOS settings are stated. Same for the PBO2 Tuner app: instructions, purposes and guidelines, tips and tricks, as well as download links (provided by the author himself) have also been extensively discussed throughout this thread.

Hey mate, thank you for the advice.
I was hesitant to use PBO2 (due to lack of personal skill) and thought I only needed some Bios tweaks, but I did it and applied a basic curve, and now my score is ∼14,300. I am pretty happy with that.
Sorry for not sharing screenshots. Before PBO2, the temps always looked okay, mostly 82-86, never 89 or above. The only time I hit 90-91 was when I did a 10 minutes stress test with Prime95, but that's alright, I guess.

Once again, thank you!

 

[lunatik]

i see you have a bios with agesia 1.2.0.7. With my ASUS x570-PRIME board running 1.2.0.7 my voltages underload go UP even with a -30 all core curve (1.212v-1.231v) . running the previous bios with 1.2.0.6b i get similar voltages as you in cinebench ~1.16v with an all core offset -30. i guess ASUS messed up the 1.2.0.7 bios?

I think it's something on your side since that doesn't happen to me.. or maybe it's with x570 mobo's?

Switched back to 1.2.0.7- all auto, except for ram and related voltages.

Altough i do cmos reset before installing other versions..

Edit: Pic is with PBO2,-30 with 120/80/120. (I wouldn't use it on gaming tho)

 

[BNSoul]

Edit: Pic is with PBO2,-30 with 120/80/120. (I wouldn't use it on gaming tho)

Why wouldn't you use that while gaming? (And what settings do you use instead?)

 

[robolee]

Anyone using this tool can try this "moonrise" build with some minor changes in CLI:
(just curious about the CPU behavior with the FIT scalar set to 0)

Debug-cli-beta.7z

drive.google.com

• only full set of CO count arguments is mandatory if provided
• limits are not applied if zero or omitted.
• @KedarWolf - zero scalar value applied if provided (not the default 1 value)
• exccessive arguments are ignored

Example: (8-core CPU)

What the arguments if i want to set a temp limit for auto launch? For now it's -30 -30 -30 -30 -30 -30 -30 -30 for the CO but i want to set a temp limit of 65 as my country ambient temp is 32ish++ 24/7 all year long

 

[lunatik]

Why wouldn't you use that while gaming? (And what settings do you use instead?)

1. I don't understand how this extra cache behaves..Wasn't that the whole reason this chip was locked (oc/uv)?

2. I have no idea how to test cache performance under different voltages/gaming.

3. Since i only play competitive games, the most important thing is minimum fps and having the lowest possible/steady frametimes.

4. I think sotr benchmark atleast uses something along the lines of ram/cache and with those pbo2 settings i will lose atleast 10-15% min/max fps..

 

[BNSoul]

I would be thankful for anyone who can run HWiNFO64 on "Zen CPU Snapshot mode" (a toggle in main settings) while Mannix' boost tester is running on the side and can share a screenshot of clock speeds at default 100.0MHz, I'm trying to make sense of the "clock speed" and "effective clock" readings since according to the author of HWINFO64 all frequencies listed under clock speed are actual readings (effective speeds that were captured directly from the CPU and not averaged from other variables), so shouldn't the effective clock speeds readings be unnecessary in this case?