• Happy holidays, folks! Thanks to each and every one of you for being part of the Tom's Hardware community!

[SOLVED] Which is more "healthy" for Ryzen: manual vs PBO OC?

erdeicsalad10

Commendable
Nov 8, 2017
39
0
1,530
So i have an R5 3600 and i've read in a few threads that ~1.3V isn't that healthy for Ryzen CPUs in terms of longevity.
So here is 2 option:
  • Manual: 4.2GHz all-core (12 threads) at 1.315V
  • PBO: +200MHz
Based on my tests the 4.2GHz all-core performs better in most of the games(+raw performance), but runs a bit hotter. Basically with PBO it's 65-68°C, with 4.2GHz it's ~66-72°C.
Of course temp spikes happens rarely to like 80°C, but that's okay for this 12 threads as i know.
Also i don't use 100% of my CPU power 99% of the time, so stress test and rendering doesn't makes sense for me imo, but here is the temps after 3 R20 bench:
  • 82°C with 4.2GHz
  • 78°C with PBO
Or should i try messing with CPU offset, that would help the CPU for longer life, or it's okay?
 
Solution
What most don't consider is how a Ryzen works. Most assume it's an Intel, and be dead wrong. At idle, Intels drop clocks and voltages on all cores, but all cores are still active. The liw/idle workloads get split up amongst those cores, so no one sees heavy usage. All cores ending up with a low, average temp. Ryzen are different in that at idle, all the cores except one are put to sleep. Only one remains active. So the entire load of all background tasks is on one core only. Any new task added, just adds to that one core load.

You look at a singular temp reading, for Intel you see 33, and that'd be the highest of the (6) cores, the other 5 being at 32-30ish. For a Ryzen, you'd see 43, but that's only 1 core, the others at case temp...
Pbo allows the cpu to regulate itself, and it's not going to allow itself to be damaged. So overall has a better chance at being healthy.

Done RIGHT, a manual OC is no less healthy than pbo and in some instances would be healthier as pbo boosts can be harder on the VRM's using non-tailored voltages and currant usage.

Done WRONG, bottom of the barrel and by far the most unhealthy.

Just saying ~1.3v isn't healthy, isn't accurate. Ryzens can easily handle 1.5v, under very low currant and core load use. Generally 1.325v is the upper limit for heavier currant and core usage. So setting 1.325v for a long term OC, when the cpu isn't stressed isn't bad, it'll run lower voltages than the 1.4v+ that is often seen, but extreme core usage like constant stress testing, long term renders etc it's going to affect lifespan being that high.

It's a difference in point of view. Amd builds cpus for the 99%. That's the masses who just want something that works really well and can do the job. Intel builds cpus for the 1%. That's the people interested in pushing boundaries and live for benchmark results.

So decide what's more important, playing the game, or one-upping anybody and everybody on a benchmark who plays the game, because that's the difference between letting the cpu/pbo just do its thing, and you forcing it by doing it manually.
 
I would highly recommend PBO + AutoOC. That will insure the CPU monitoring software stays enabled and will insure longevity of the CPU. As said already...

However, a neat trick i heard works well, is to run a negative offset voltage. This keeps the monitoring software running, and allows you to manually adjust the voltage at the same time. This should both lower temps and allow the CPU to clock itself a little higher.
 
Pbo allows the cpu to regulate itself, and it's not going to allow itself to be damaged. So overall has a better chance at being healthy.

Done RIGHT, a manual OC is no less healthy than pbo and in some instances would be healthier as pbo boosts can be harder on the VRM's using non-tailored voltages and currant usage.

Done WRONG, bottom of the barrel and by far the most unhealthy.

Just saying ~1.3v isn't healthy, isn't accurate. Ryzens can easily handle 1.5v, under very low currant and core load use. Generally 1.325v is the upper limit for heavier currant and core usage. So setting 1.325v for a long term OC, when the cpu isn't stressed isn't bad, it'll run lower voltages than the 1.4v+ that is often seen, but extreme core usage like constant stress testing, long term renders etc it's going to affect lifespan being that high.

It's a difference in point of view. Amd builds cpus for the 99%. That's the masses who just want something that works really well and can do the job. Intel builds cpus for the 1%. That's the people interested in pushing boundaries and live for benchmark results.

So decide what's more important, playing the game, or one-upping anybody and everybody on a benchmark who plays the game, because that's the difference between letting the cpu/pbo just do its thing, and you forcing it by doing it manually.

Thanks for the replies.

I didn't mentioned it, but i have -offset in my PBO profile, because i wanted to get rid of the idle temp spikes, that's the worst thing i ever experienced with a CPU, and also produce lot less heat.
 
I would highly recommend PBO + AutoOC. That will insure the CPU monitoring software stays enabled and will insure longevity of the CPU. As said already...

However, a neat trick i heard works well, is to run a negative offset voltage. This keeps the monitoring software running, and allows you to manually adjust the voltage at the same time. This should both lower temps and allow the CPU to clock itself a little higher.
Ohh yeah, this. So i'll use my PBO profile for sure. Thanks. :)
 
What most don't consider is how a Ryzen works. Most assume it's an Intel, and be dead wrong. At idle, Intels drop clocks and voltages on all cores, but all cores are still active. The liw/idle workloads get split up amongst those cores, so no one sees heavy usage. All cores ending up with a low, average temp. Ryzen are different in that at idle, all the cores except one are put to sleep. Only one remains active. So the entire load of all background tasks is on one core only. Any new task added, just adds to that one core load.

You look at a singular temp reading, for Intel you see 33, and that'd be the highest of the (6) cores, the other 5 being at 32-30ish. For a Ryzen, you'd see 43, but that's only 1 core, the others at case temp (26?) inactive. Add a new process and that single core can peak/spike to 70, but quickly drops back to 43. And then you look at temps and see a max of 70 and freak out, not realizing that was a process startup, which put the cpu at 100%, for a single core, 0% for the other 5.

At Windows idle, fully expect temp bounces, just understand that they mean absolutely nothing and are just part of the pc working normally.
 
  • Like
Reactions: tiberium87
Solution