[SOLVED] CPU degredation?

dralaamonirsrg

Commendable
Sep 28, 2018
44
2
1,535
Hello folks,
I got a question... I've got a ryzen 5 3600 on AIO water cooler since months and always been working on PBO or auto oc. The thing is I tried manual oc today and set multiplier to 42 but wasn't stable till I pushed voltage to 1.42. Temps during gaming was 50-55 c . Feels bad knowing almost everyone on youtube got it in less than 1.35v and what one mate said: you degraded your CPU.
What should I really do? decrease OC or revert to default? any suggestions?

Regards
 
Solution
Well at 1.4v was alt+tab within the game and run youtube and my game crashed. Below 1.4 I get frequent blue screens under load like repeating the cinabench test.
...
Pretty much impossible to tell if it's degraded from that kind of 'testing' as there's way too much going on. Memory, GPU, CPU, it's all being used heavily. And as well there is simply bad coding by game developer.

The way to test is to run an extreme CPU stability test, like Prime95, that stresses only the CPU. Set that up and run it for 20 min's to an hour on a single FFT size to find the lowest stable voltage at your overclock. Then later on, when you think you've degraded your CPU, run that exact same test, also 20 minutes to an hour. If it can't hold at...
Hello folks,
I got a question... I've got a ryzen 5 3600 on AIO water cooler since months and always been working on PBO or auto oc. The thing is I tried manual oc today and set multiplier to 42 but wasn't stable till I pushed voltage to 1.42. Temps during gaming was 50-55 c . Feels bad knowing almost everyone on youtube got it in less than 1.35v and what one mate said: you degraded your CPU.
What should I really do? decrease OC or revert to default? any suggestions?

Regards
How did you determine it wasn't stable?

All CPU's and motherboards are different so comparing yours to theirs is pointless unless you precisely duplicate their setup (motherboard, settings, cooling, ambient temp, etc). And making a confident determination you degraded your CPU needs a good baseline comparison test that it can no longer pass at the same voltage.

One of the...umm...let's say interesting things about watching Buildzoid's rambling videos is you get unfiltered thoughts. Like those as he agonizes over whether he actually degraded his 3700X. He THINKS he did but he can't be sure because he didn't get good baseline data to compare to. In the mean time, he's still happily using it just no longer in a permanently fixed overclock.

In Buildzoid's case he was running a fixed voltage at a high fixed clock speed doing very frequent extreme AVX workload processing so he has good reason to wonder about it. In your case, if you were doing PBO right you'd left both clocks and voltage in AUTO so it was able to drop voltage as needed when it got warm. And gaming is actually very light processing even if it's time sensitive.
 
Last edited:

dralaamonirsrg

Commendable
Sep 28, 2018
44
2
1,535
How did you determine it wasn't stable?

All CPU's and motherboards are different so comparing yours to theirs is pointless unless you precisely duplicate their setup (motherboard, settings, cooling, ambient temp, etc). And making a confident determination you degraded your CPU needs a good baseline comparison test that it can no longer pass at the same voltage.

One of the...umm...let's say interesting things about watching Buildzoid's rambling videos is you get unfiltered thoughts. Like those as he agonizes over whether he actually degraded his 3700X. He THINKS he did but he can't be sure because he didn't get good baseline data to compare to. In the mean time, he's still happily using it just no longer in a permanently fixed overclock.

In Buildzoid's case he was running a fixed voltage at a high fixed clock speed doing very frequent extreme AVX workload processing so he has good reason to wonder about it. In your case, if you were doing PBO right you'd left both clocks and voltage in AUTO so it was able to drop voltage as needed when it got warm. And gaming is actually very light processing even if it's time sensitive.

Well at 1.4v was alt+tab within the game and run youtube and my game crashed. Below 1.4 I get frequent blue screens under load like repeating the cinabench test.
Here I reverted to PBO and put voltage to offset - 0.0750 and here is how it looks like while call of duty running at the background, on ryzen balanced power plan
AjDVLQQ.jpg
 
Well at 1.4v was alt+tab within the game and run youtube and my game crashed. Below 1.4 I get frequent blue screens under load like repeating the cinabench test.
...
Pretty much impossible to tell if it's degraded from that kind of 'testing' as there's way too much going on. Memory, GPU, CPU, it's all being used heavily. And as well there is simply bad coding by game developer.

The way to test is to run an extreme CPU stability test, like Prime95, that stresses only the CPU. Set that up and run it for 20 min's to an hour on a single FFT size to find the lowest stable voltage at your overclock. Then later on, when you think you've degraded your CPU, run that exact same test, also 20 minutes to an hour. If it can't hold at that same lowest stable voltage then yeah, it's degraded.

Keep in mind I'm not saying you haven't degraded it, just that you can't be certain yet. After all, if you're running a fixed clock and fixed voltage over 1.2V it will almost certainly degrade faster.
 
Solution

dralaamonirsrg

Commendable
Sep 28, 2018
44
2
1,535
Pretty much impossible to tell if it's degraded from that kind of 'testing' as there's way too much going on. Memory, GPU, CPU, it's all being used heavily. And as well there is simply bad coding by game developer.

The way to test is to run an extreme CPU stability test, like Prime95, that stresses only the CPU. Set that up and run it for 20 min's to an hour on a single FFT size to find the lowest stable voltage at your overclock. Then later on, when you think you've degraded your CPU, run that exact same test, also 20 minutes to an hour. If it can't hold at that same lowest stable voltage then yeah, it's degraded.

Keep in mind I'm not saying you haven't degraded it, just that you can't be certain yet. After all, if you're running a fixed clock and fixed voltage over 1.2V it will almost certainly degrade faster.

On PBO voltage offset - 0.750, prime not crashing max voltage seen was 1.3 it was almost always 1.27v
clock was 3873mhz for all threads temp maxed at 72 c
seems that CPU runs 1.4v while it's not under load and decrease voltage with load.

what do u recommend?
 
On PBO voltage offset - 0.750, prime not crashing max voltage seen was 1.3 it was almost always 1.27v
clock was 3873mhz for all threads temp maxed at 72 c
seems that CPU runs 1.4v while it's not under load and decrease voltage with load.

what do u recommend?
Are you sure that negative offset is -0.750? could it actually be -0.0750? If it is -0.750, that's a HUGE negative offset and almost certainly causing it to pull performance even if you don't see it in clocks.

I'd put clocks and voltage to AUTO and see what it does.

At the same time, make sure the following are set to ENABLED in BIOS: CoolnQuiet, Global C States, processor CPPC, processor CPPC preferred cores.

Here's the thing about PBO: the processor's boost algorithm still tracks voltage and frequency against the processor's FIT tables as temperature gets warmer to keep it safe from degradation. But it's only able to do it correctly if you're not changing what it thinks it's doing by offsetting voltage, that's why keeping those settings in AUTO gives a better idea how it's working.

And finally: your processor's FIT values are unique to your processor, so how it performs with voltage and frequency over temperature is also unique to your processor.

And since ALL of that is sensitive to temperature, knowing the ambient temperature is CRITICAL because it changes with seasonal changes, and so also changes the voltages and frequency considered safe. That part is especially important on air cooling since liquid cooling provides more cooling margin to saturation.
 
Last edited:

dralaamonirsrg

Commendable
Sep 28, 2018
44
2
1,535
Are you sure that negative offset is -0.750? could it actually be -0.0750? If it is -0.750, that's a HUGE negative offset and almost certainly causing it to pull performance even if you don't see it in clocks.

I'd put clocks and voltage to AUTO and see what it does.

At the same time, make sure the following are set to ENABLED in BIOS: CoolnQuiet, Global C States, processor CPPC, processor CPPC preferred cores.

Here's the thing about PBO: the processor's boost algorithm still tracks voltage and frequency against the processor's FIT tables as temperature gets warmer to keep it safe from degradation. But it's only able to do it correctly if you're not changing what it thinks it's doing by offsetting voltage, that's why keeping those settings in AUTO gives a better idea how it's working.

And finally: your processor's FIT values are unique to your processor, so how it performs with voltage and frequency over temperature is also unique to your processor.

And since ALL of that is sensitive to temperature, knowing the ambient temperature is CRITICAL because it changes with seasonal changes, and so also changes the voltages and frequency considered safe. That part is especially important on air cooling since liquid cooling provides more cooling margin to saturation.
Thanks for the reply,
Tested again using these settings on prime max power test voltage was almost always 1.165v
while clock speed was 3774mhz temp was 70 c

If it can run stable under 100% on that low voltage what makes it unstable doing 4.2ghz at 1.3-1.35v
I don't get it. Yet it keeps going upto 1.47v doing boosting in gaming and light tasks.
Is this still normal?
 

86zx

Upstanding
Nov 1, 2019
484
90
290
Ryzen seems to like voltage. My 2700x doesn’t really like to be bellow 1.35 under very slight overclocks. Another factor is motherboard power delivery issues, I used to have a p5b deluxe motherboard I overclocked my q9400 to 3.6 and it was stable for quite some time then I got blue screens here and there so I downclocked it and it was stable for a while then started happening again till one day the board quit the cpu is still totally fine and I put it in another system without issue.
 
Thanks for the reply,
Tested again using these settings on prime max power test voltage was almost always 1.165v
while clock speed was 3774mhz temp was 70 c

If it can run stable under 100% on that low voltage what makes it unstable doing 4.2ghz at 1.3-1.35v
I don't get it. Yet it keeps going upto 1.47v doing boosting in gaming and light tasks.
Is this still normal?
It's well known you can undervolt Ryzen 3000 processors really far and they remain stable, but you do not realize they're actually underperforming. It's called clock or frequency compression or something like that. I remember there was one YouTube reviewer who did that at product launch and had to correct his review when he realized what was happening. So it's possible that's what's happening in your case.