7950X3D PBO

[Mando772004]

i tried setting PBO to Auto in the bios, mind you i have nothing else tweaked on my 7950x3d, just prefer cache set to enabled, and i compared scores in cinbench and cs2 and i didnt notice much difference, to be clear i had it set to enabled before switching to auto which to my knowledge disables it, can someone tell me if its worth it to keep it on or off based on their experience, maybe ill gain a benefit in other games/apps? please tell me your experience with this

[Agall]

38 minutes ago, Mando772004 said:

i tried setting PBO to Auto in the bios, mind you i have nothing else tweaked on my 7950x3d, just prefer cache set to enabled, and i compared scores in cinbench and cs2 and i didnt notice much difference, to be clear i had it set to enabled before switching to auto which to my knowledge disables it, can someone tell me if its worth it to keep it on or off based on their experience, maybe ill gain a benefit in other games/apps? please tell me your experience with this

Set PBO to +200MHz and -15mV all core. SoC voltage to 1.15V.

 

Those are general rules for Ryzen 7000, the like 8 AM5 systems I've built, including a 7950x3D, all didn't mind those settings at all from the rip. The SoC voltage reduces the IOD wattage by like 33%, which on 3D v-cache CPUs can matter a lot with their lower TDP limit.

[Mando772004]

19 minutes ago, Agall said:

Set PBO to +200MHz and -15mV all core. SoC voltage to 1.15V.

 

Those are general rules for Ryzen 7000, the like 8 AM5 systems I've built, including a 7950x3D, all didn't mind those settings at all from the rip. The SoC voltage reduces the IOD wattage by like 33%, which on 3D v-cache CPUs can matter a lot with their lower TDP limit.

Ok i can try that, but generally speaking assuming one wanted to run stock, normal pbo vs auto, it doesnt make a difference right?

[Agall]

Just now, Mando772004 said:

Ok i can try that, but generally speaking assuming one wanted to run stock, normal pbo vs auto, it doesnt make a difference right?

PBO in general can make a huge difference. 

 

If you're already using cinebench R23, I have some older but likely still relevant data I took with my 7950x3D, before and after direct-die.

 

 

In this testing is where I figured out the SoC undervolt, which I've done on every system since to great success. I was able to get ridiculously low though with direct-die, so I take 1.15V as the normally configured standard.

[Stahlmann]

32 minutes ago, Mando772004 said:

Ok i can try that, but generally speaking assuming one wanted to run stock, normal pbo vs auto, it doesnt make a difference right?

For heavily multithreaded work it can make a decent difference, but in more everyday tasks and gaming, it doesn't really do anything other than maybe dropping temperatures at the cost of some system stability. The stability impact is heavily dependant on silicon lottery. YMMV

 

Especially with undervolting some X3D chips can run at the max -20 offset, some struggle with stability even at -10. And when you approach -5 you shouldn't bother because it doesn't do much good at this point.

 

My 5800X3D runs very hot no matter which cooler I use and can only do -10 voltage offset before running into instability. It's an example of bad luck with silicon lottery. So I don't bother with undervolting or custom PBO because it doesn't make a difference in my use case, which is 90% gaming and media consumption, with the occational multithread workload. And saving a few seconds here and there every few weeks isn't a big deal.

[Mando772004]

3 hours ago, Agall said:

PBO in general can make a huge difference. 

 

If you're already using cinebench R23, I have some older but likely still relevant data I took with my 7950x3D, before and after direct-die.

 

 

In this testing is where I figured out the SoC undervolt, which I've done on every system since to great success. I was able to get ridiculously low though with direct-die, so I take 1.15V as the normally configured standard.

do you happen to have a cpu score on windows 11 3dmark timespy test? i know a weird test to see CPU performance but im getting around 16k with background apps off and on windows 10 i did get a fair bit higher however cinebench is pretty much the same score or almost the same on windows 11 and 10, i also disabled core isolation. if you have a score for that with only PBO enabled no other OC or CO or anything that would be lovely thank you! i do get very similar results on cinebench with around 35-36k score with only pbo without any SOC voltage modifiers or Curver optimizer settings so i think my cpu is fine but in games however i wish i could know for a fact which is why i did try 3dmark because im guessing it could be similar to games

[Agall]

8 minutes ago, Mando772004 said:

do you happen to have a cpu score on windows 11 3dmark timespy test? i know a weird test to see CPU performance but im getting around 16k with background apps off and on windows 10 i did get a fair bit higher however cinebench is pretty much the same score or almost the same on windows 11 and 10, i also disabled core isolation. if you have a score for that with only PBO enabled no other OC or CO or anything that would be lovely thank you! i do get very similar results on cinebench with around 35-36k score with only pbo without any SOC voltage modifiers or Curver optimizer settings so i think my cpu is fine but in games however i wish i could know for a fact which is why i did try 3dmark because im guessing it could be similar to games

I've had some weird licensing issue with 3Dmark so I stopped using it a year ago or so. I primarily use cinebench r23 for CPU multicore testing and the games I play for practical CPU/GPU testing.

 

I should probably benchmark my current setup, but I feel like I'll run into the same nonsense with 3Dmark.

[Mando772004]

13 hours ago, Agall said:

I've had some weird licensing issue with 3Dmark so I stopped using it a year ago or so. I primarily use cinebench r23 for CPU multicore testing and the games I play for practical CPU/GPU testing.

 

I should probably benchmark my current setup, but I feel like I'll run into the same nonsense with 3Dmark.

Understandable, i actually used the free demo. Im guessing if cinebench scores are similar then thats enough of an indication that its running well

[Agall]

5 hours ago, Mando772004 said:

Understandable, i actually used the free demo. Im guessing if cinebench scores are similar then thats enough of an indication that its running well

I've found cinebench to produce a score consistent with average boost clocks. In terms of gaming performance, it's a good indicator that you're not thermally limited at high usage, as long as you've designed things properly to not have the GPU affect CPU performance significantly.