• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

What is your go to testing method for checking PBO + CO stability

michty_me

michty_me

michty_me

Soldato
Joined
19 Apr 2012
Posts
6,135
I'm just looking for some pointers on testing PBO and CO stability on AM5.

It has been some time since I've tried PBO and CO at the same time (approximately 2019-2020) and since upgrading to AM5, I'm wanting to get back into it a bit.

As an example, I had +100Mhz on the CPU and -25 on CO which would pass OCCT over and over (Possibly wrong instruction set?), However testing on Ycruncher failed instantly.
+100Mhz -20 would do three passes then fail.
+100Mhz -18 has made 12 passes of Ycruncher so far. Maybe after a few more passes, I go onto Prime95 or Aida64?

Can I get some further testing tips i.e what should I be running for testing CPU, RAM etc
How many passes do you go for before calling it good? What settings do you use in each?
 
No clear answer
On paper OCCT an ycruncher do what you would want, test at spectrum of loads, core counts, instruction sets. They are good for a rough check after every adjustment.
But I had configurations stable on both, but instacrashing on Cinebench (never happened before in my life, I thought my CB install was corrupted, but no) or AIDA64 heavier FPU test. And even adjusting after that, some games had occasional crash.

So various tests and time under variable load.
And as a last step after being pretty sure it's stable, still bring CO up 3-5 points for safety margin.

Btw I suggest doing +200 clock first, curve shaper second, curve optimiser per core third.
And increasing PBO power limits optional, mostly for lower end 65W ones that are power limited. Too little gain for too much heat increase.

And since my target was responsiveness in games, I used a browser test Mozilla Kraken for single/low thread performance measure. Super quick and I think reflects the sort of load that would affect 1% lows in games.
 
Last edited:
alec said:
No clear answer
On paper OCCT an ycruncher do what you would want, test at spectrum of loads, core counts, instruction sets. They are good for a rough check after every adjustment.
But I had configurations stable on both, but instacrashing on Cinebench (never happened before in my life, I thought my CB install was corrupted, but no) or AIDA64 heavier FPU test. And even adjusting after that, some games had occasional crash.

So various tests and time under variable load.
And as a last step after being pretty sure it's stable, still bring CO up 3-5 points for safety margin.

Btw I suggest doing +200 clock first, curve shaper second, curve optimiser per core third.
And increasing PBO power limits optional, mostly for lower end 65W ones that are power limited. Too little gain for too much heat increase.

And since my target was responsiveness in games, I used a browser test Mozilla Kraken for single/low thread performance measure. Super quick and I think reflects the sort of load that would affect 1% lows in games.
Click to expand...
Thanks, I could easily pass CB runs no problem at all as well as OCCT but instacrash on ycruncher.
I've got 105w mode selected in bios.
I've not even looked at curve shaper as I know nothing about it to be honest. If you can point me in the direction of somewhere to learn about it, in all ears!
 
alec said:
yeah normally CB would never crash for me even when other tools detect instability, thats why I was surprised

Skatterbencher goes really in depth on what curve shaper does https://skatterbencher.com/2024/09/06/undervolt-ryzen-9000/#Curve_Shaper_in_Practice
I was lazy and decreased every curve shaper setting by same amount. For "reasons" it was better for this CPU that a similar lowering all core curve optimiser
Click to expand...
Brilliant, thanks!
I'll give that a good read over and watch and go from there. I'm really liking his content, it's very well explained.
 
Use it and see if it works?
I ran my 7800x3d at -40 all core and it seemingly didn't bat an eyelid and whatever tests I tried.
Then I had the odd crash playing cod. Happened once after a few hours gaming....then it happened again a few nights later...then again.
Knocked the curve back a bit and it's never crashed since
 
Prime 95 Small FFT, usually for me if it's very unstable it will crash within 1 seconds

After it passes that, leave the PC doing nothing but idling on desktop for 24 hours, usually when Ryzen is unstable it will crash idling on desktop

That's all I do, in my experience with Ryzen if it can passes both Prime 95 (low voltage, high load) and desktop idling (high voltage, low load) then it's rock solid, I don't need to test anything else
 
Last edited:
Dreadi said:
Use it and see if it works?
I ran my 7800x3d at -40 all core and it seemingly didn't bat an eyelid and whatever tests I tried.
Then I had the odd crash playing cod. Happened once after a few hours gaming....then it happened again a few nights later...then again.
Knocked the curve back a bit and it's never crashed since
Click to expand...
Thats a fair test then :cry:

I'm was aiming for 100% stable and not ringing the neck of the thing really. I'm surprised you managed on -40 for some gaming to be fair!
 
Grim5 said:
Prime 95 Small FFT, usually for me if it's very unstable it will crash within 1 seconds

After it passes that, leave the PC doing nothing but idling on desktop for 24 hours, usually when Ryzen is unstable it will crash idling on desktop

That's all I do, in my experience with Ryzen if it can passes both Prime 95 (low voltage, high load) and desktop idling (high voltage, low load) then it's rock solid, I don't need to test anything else
Click to expand...
Thanks,
I had heard low load was the trigger for fails to be honest. I've read posts on here with hours and hours of passed stability tsting then crashing on idle.
 
michty_me said:
Thats a fair test then :cry:

I'm was aiming for 100% stable and not ringing the neck of the thing really. I'm surprised you managed on -40 for some gaming to be fair!
Click to expand...
I managed -50 on for some benchmarking laughs, noticed no improvement over -40. After the crashes I put it to -35 all core and has been there ever since
 
AIDA64 Extreme without a doubt.

Went through all this and multiple benchies and tests when I upgraded fairly recently and went on a major PBO CO journey - and this was the ONLY one that was a true stability test. Been rock solid ever since for months now with a good PBO CO undervolt etc on my AM5 system.
 
Last edited:
Who knows, I passed every stress test I could think of and multiple games. Yet it would crash constantly in Doom The Dark Ages without some adjustment.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom