PBO limits for 5900X

[heatonpkmassive]

Is -30 all core stable? That's like the max you can dial in.
I also raid that touching the PBO scalar is not recommended, since it can have a dangerous effect on voltages.

It's what I started with and haven't changed for a year, boosts to 4.5Ghz all core in Cinebench R23 at a max temp of around 75-76c. That's at PPT 175, TDC 110, and EDC 155. I guess I got lucky, but there are clearly others here who have had the same luck.

Edit: Not touched PBO scalar.

 

[Trick]

@Forceflow be aware that since AGESA 1.2.0.5 EDC is borked and limited to 140 on a 5900X. If you set EDC higher than 140 then boost voltage is throttled.

 

[Cyber-Mav]

@Forceflow be aware that since AGESA 1.2.0.5 EDC is borked and limited to 140 on a 5900X. If you set EDC higher than 140 then boost voltage is throttled.

Thats an advantage which i use. I set edc to 141 so max voltage wont go over 1.425v. otherwise it shoots to 1.5v. and my benches show no loss in single thread speed or clocks but if you want to overclock the best cores to do 5.1ghz then you will need that 1.5v.
Im still not convinced that 1.5v long term is safe for these chips and amd dropping it to 1.425v for higher edc makes sense since i believe 1.425v is the true upper limit.

 

[xPETEZx]

Thank you for all the replies!

Up until yesterday, I had appearantly been running it on PPT 500, TDC 210, EDC 280, which I now see are motherboard values that correspond to "let er' rip".
I never minded these values, I thought these were safe, and they probably are, but they are a wasteful way of (not even getting the best, mind you) better performance in all-core loads. I don't understand why just enabling PBO (what an end-user would do, just set it to "enabled" in BIOS) results in these wasteful values.

What I did *not* understand is how my CPU got hotter and got better all-core boosts when I lowered EDC. More limited, less power draw, right? But after watching and reading a lot I now know that there's a certain optimum to tuning these parameters, and how they work together. After continuous tweaking on just the PBO2 params I landed on PPT 180, TDC 120 and EDC 165. This results in an all-core load of 4,35 Ghz at 72 degrees in CB R23.

Since this is mostly a gaming system, I decided against dialing these in, and just went back to stock. The gains in-game are non-existent or very minimal, even in games that like a lot of threads (warzone, darktide). Lower thermals and lower power usage is beneficial for my chip and for the environment, at the moment.

I'll keep that PBO2 profile handy through for when I encounter a situation in which I need more oompf, and maybe have a look at Curve Optimizer later.

My stock Cinebench R23 all-core scores are in the low 22's, but my RAM probably isn't the best (3600 Mhz with 16-19-19-19 timings) and I've read up that recent AGESA versions introduced some odd clockspeed bugs.

Does this all sound reasonable?

This ended up being my conclusion.
I spent some time trying to dial in these values and curve optimiser. I started with -30 all core and although that seemed stable 95% of the time, I got 1 or 2 game crashes in the 3 months I had it... just turned out to be more work than it was worth, so I went back to stock no PBO.

I get far lower CB23 scores bone stock though...usually about 19.9k is my top. I have given up chasing down why though... its a rabbit hole I dont have time / interest for.

 

[Cyber-Mav]

I did curve optimiser 1 core at a time and started on the weakest core identified by windows, hwingo and ryzen master can also identify the core preferred order.
Used core cycler to do the testing. Pc been rock solid since it was done

 

[NoobCannon]

Sorry to bump an older post. But Adrenaline does the curve optimiser for you now including stress testing.

ran it yesterday on my 5900x and changed to the values it suggested and has been great.

https://www.amd.com/en/support/kb/release-notes/rn-ryzen-master-2-9-0-2093

 

[Wing-Man]

always set BPO to motherboard, as that will be the limit if there is any... but it wont be

do CO for all cores til you have a fail then you know what the weakest core can do.. saves doing on core at a time as its pointless
i could do -9 on my 5800x but -15 on core 2 and 6 but i got zero improvements over just running -9 all cores

 

[NutritioN]

Personally don't like the idea of using the MB PBO limits, as it effectively says 'use as much as you can' and didn't change my stock Cinebench score. I explored MB PBO limit while testing my 5900x just to see how far it went before something else held it back. Following numbers are what the motherboard set limits at, and then what Ryzen Master actually hit during Cinebench MC R23 testing:

Using 175W
PPT 540 -> 175
TDC 300 -> 115
EDC 215 -> 208

Personally, I like to control it a bit more and fancied a lower power usage, definitely not going for high scores. So for context my daily settings are as follows for a quieter fan profile + custom CO curve stressed with corecycler : 135,90,100 using 137W.

No change to Cinebench score.

 

[alesgs]

-30 on curve optimiser
TDC 170A ( 118A R23)
EDC 140 A (140A R23)
PPT 230 W (180W R23)
10X
200MHz

all core 4.674 in R23 72-75c

CPU-Z benchmark https://valid.x86.fr/ccf48u

Hello, my question is after more than a year with these values, have you had any problems with the CPU? Would you recommend the same values today (May 2024) and with the latest updates to agesa 1.2.0. California. ? Since for me, even if you put your values, my maximum values in the r23 test are PPT 178w TDC 123 A AND EDC 140 A with an average of 4500 MHz and a score of 21500 points. I suppose it will depend on the silicon lottery, one thing that seems curious to me is that when you put PBO activated on the board, that is, it is managed by the same motherboard, the EDC value rises to 190 A and the TDC to 135 A, that's why my question is whether you continue to recommend your values, or whether a higher EDC is currently supported and more stable. I hope you understand my question, thank you very much.

 

[Ross Thomson]

Hello, my question is after more than a year with these values, have you had any problems with the CPU? Would you recommend the same values today (May 2024) and with the latest updates to agesa 1.2.0. California. ? Since for me, even if you put your values, my maximum values in the r23 test are PPT 178w TDC 123 A AND EDC 140 A with an average of 4500 MHz and a score of 21500 points. I suppose it will depend on the silicon lottery, one thing that seems curious to me is that when you put PBO activated on the board, that is, it is managed by the same motherboard, the EDC value rises to 190 A and the TDC to 135 A, that's why my question is whether you continue to recommend your values, or whether a higher EDC is currently supported and more stable. I hope you understand my question, thank you very much.

nope no issues at all. You can only play about with different values to see what you get. My maximums are in brackets for R23. You might find that more EDC just gives your more heat. I used R23 to set my curve optimiser and my EDC, PPT, TDC values. I did a run there and got 21955 with a few windows open and it was just one run.

 

[heatonpkmassive]

Hello, my question is after more than a year with these values, have you had any problems with the CPU? Would you recommend the same values today (May 2024) and with the latest updates to agesa 1.2.0. California. ? Since for me, even if you put your values, my maximum values in the r23 test are PPT 178w TDC 123 A AND EDC 140 A with an average of 4500 MHz and a score of 21500 points. I suppose it will depend on the silicon lottery, one thing that seems curious to me is that when you put PBO activated on the board, that is, it is managed by the same motherboard, the EDC value rises to 190 A and the TDC to 135 A, that's why my question is whether you continue to recommend your values, or whether a higher EDC is currently supported and more stable. I hope you understand my question, thank you very much.

And even though you asked Ross, like him I set these values ages ago and system has been rock solid since. These are great chips.