[SOLVED] 3600x voltage? Confusing answers.

Apr 15, 2021
16
0
10
I read a few places that going above 1.325 voltage on 3600s is considered a bad thing for the chip. Is that true and why? Currently I'm manually OC'ing to 4200 all core at 1.325v. But I'd like to go up to 4300 or even 4400 if my temps look fine with higher voltage.
 
Solution
~80C is where the cpu stops pursuing higher clocks.
Cinebench R23 is a more common scenario than the likes of Prime 95(small FFT, AVX/ AVX2/ AVX-512 off) and OCCT, which are more of a worst case scenario, but not impossible to reproduce in specific titles.
These games you play are not as heavy.
I read a few places that going above 1.325 voltage on 3600s is considered a bad thing for the chip. Is that true and why? ...

Early degradation of the CPU due to electromigration is why. The reason you get confusing answers is most people misunderstand what's happening.

Left in stock the CPU uses fairly high voltage (up to 1.5V) to boost high, single cores at a time up to rated boost clock. So people incorrectly assume you can go that high all the time, all core. The CPU (in AUTO) can do that because it monitors CPU conditions with hundreds of sensors and lowers clocks and voltage (down to 1.25V or lower) when temps and other things start to get get hot. Once voltage and clocks are fixed it can't do that so it slowly burns itself up (electromigration) in heavy work.

1.325V is kind of an un-happy medium setting. It will most likely go unstable pretty early in life if you do a lot of heavy processing (like rendering) but gaming (which is pretty light) may let it live longer. It's kind of like what Roy Batty found out:there's a price to pay for living a bright life.
 
Last edited:
  • Like
Reactions: John Chesterfield
The only worry here is from electromigration due to excessive current/heat as voltage is a factor in both. But even then, integrated IC manufacturers are well aware of this and they've designed such ICs to mitigate electromigration and run within the designed parameters for years.

Basically, don't worry about it. Just keep the processor reasonably cool and it'll last longer than you'll care about it.
 
Apr 15, 2021
16
0
10
The only worry here is from electromigration due to excessive current/heat as voltage is a factor in both. But even then, integrated IC manufacturers are well aware of this and they've designed such ICs to mitigate electromigration and run within the designed parameters for years.

Basically, don't worry about it. Just keep the processor reasonably cool and it'll last longer than you'll care about it.

That's the thing, when I use auto overclock from ryzen master, sure, the hz can shoot up to 4.4 but the voltages are crazy, anywhere from 1.2-1.45v, mostly being on the higher end. The CPU can hit in the 70c with that. Setting voltage to say 1.325v and 4200 all core (this is where I find overwatch won't crash), it's very rare I'll hit 70c while gaming. But my research is confusing, people say 1.325v or higher isn't safe and even a little bit lower isn't safe as well, so yeah not sure... I want to go higher like 1.37v 4300 all core if I can keep the temps reasonable and below 75c. Not sure if that voltage at all times is something safe though or what.
 
Last edited:

Phaaze88

Titan
Ambassador
From what I've read, 1.325v isn't a 'safe boundary' or anything. It varies from silicon to silicon. For example, some can't handle 1.25v 24/7, and some can do 1.35v 24/7.
Many older revisions could do with 1.325v, but some months later, the PC is crashing due to the OC no longer being stable, or it straight up died.

That's the thing, when I use auto overclock from ryzen master, sure, the hz can shoot up to 4.4 but the voltages are crazy, anywhere from 1.2-1.45v, mostly being on the higher end. The CPU can hit in the 70c with that. Setting voltage to say 1.325v and 4200 all core, it's very rare I'll hit 70c while gaming.
The key factor here is that the cpu is still allowed to 'amp down' in scenario 1. Sure, your OC in scenario 2 achieves lower thermals, but there's more to it than that. It's probably pulling more amps during heavier loads than scenario 1.
 
... But my research is confusing, people say 1.325v or higher isn't safe....
..The CPU can hit in the 70c with that. Setting voltage to say 1.325v and 4200 all core (this is where I find overwatch won't crash), it's very rare I'll hit 70c while gaming....
The CPU is constantly boosting and pulling back, it's very fast. Up to 100 times per second. When temps are getting to 70C and above the pulling back is to protect itself. The point is: it's not constant, as it is when you run a fixed voltage. In fact, it's pulling back to under 1.2V most of the time while I'm gaming and only boosting upwards of 1.5V a small part of the time. It's pretty dynamic, only hitting the high clocks when game processing action is intense then pulling back when it can.

You also have to use right voltage monitoring tool: HWInfo64 is the best. Look for a reading: Core voltage (SVI2 TFN).

Also, the thing you should do with Ryzen is ignore clock speed as a benchmark. It's really pointless as it is so dynamic. Instead use a decent performance benchmark like Cinebench 20 or 23. Be sure to also run the single thread bench as you might get better all-core performance but kill single core performance. Most games are lightly threaded, even the multi-threaded ones only one or two of the threads are heavily loaded.
 

Karadjgne

Titan
Ambassador
Which voltage. There's a very big difference between setting vcore, SVI2 and VID.

Vcore is what the cpu demands + LLC, SVI2 is what the cpu actually uses, VID is what the motherboard VRM's supply.

Left on auto, the cpu thinks about its loads, adds in LLC and demands higher voltages from the VRM's than it realistically needs, just to cover all the bases. The motherboard is happy to supply that, and then the cpu only uses some of that.

So limiting VID is like asking mom for $10 and she laughs and gives you $5 and tells you to make do. Overall, that's a benefit, reduces heat, reduces stress on the VRM's, keeps vcore and SVI2 at reasonable levels. Setting a solid vcore has the opposite affect, you must set it high enough to cover single core demands and also multi core demands, but then LLC adds voltage to help with vdroop and stability. So you may set 1.325v solid, but only use 1.125v but the cpu gets supplied 1.375v VID or higher to cover everything if you've got funky vdroop.

So if just setting vcore, you are setting yourself up for failure, you must take LLC into consideration as well, and current limits, load turbo timers etc.

My 3700x runs a 1.232v VID and still gets 4.4GHz boost on many cores, 4.28GHz all core, vcore running @ 1.205v and SVI2 running 1.18-1.2v
 
I disagree that 1.325v will cause early instability. Given what the stock voltage often reaches from my experience, 1.325v is an undervolt that will increase lifespan.

I have had 2 Ryzen 5 3600s. One on an MSI B350 board, one on an Asrock A520. Both have pushed well above this voltage during all core stress tests out of the box, such as Cinebench. Not a single core workload, not a light bursty workload. All core.

I will run another test here to confirm.
 
Apr 15, 2021
16
0
10
From what I've read, 1.325v isn't a 'safe boundary' or anything. It varies from silicon to silicon. For example, some can't handle 1.25v 24/7, and some can do 1.35v 24/7.
Many older revisions could do with 1.325v, but some months later, the PC is crashing due to the OC no longer being stable, or it straight up died.


The key factor here is that the cpu is still allowed to 'amp down' in scenario 1. Sure, your OC in scenario 2 achieves lower thermals, but there's more to it than that. It's probably pulling more amps during heavier loads than scenario 1.
I 100% for sure get much lower temps when setting voltage, gaming with auto overclock I've even seen the CPU gets in the 80s because it's cranking out voltage like crazy. I don't think hitting that high seems ideal, also stock basically does the same thing, very high jumps in voltage and performs bad in benchmarks compared to all core.
 
Just tried Cinebench R23 on my Ryzen 5 3600, specs in my sig. At the very start of the test, the CPU was turboing to 4.2ghz all core at about 1.39-1.4v. It soon dropped down to 4.1ghz all core and 1.34v. This was reported via hwinfo.

I fail to see how running below 100% stock voltage is going to cause the CPUs lifespan to be decreased.
 
Apr 15, 2021
16
0
10
Which voltage. There's a very big difference between setting vcore, SVI2 and VID.

Vcore is what the cpu demands + LLC, SVI2 is what the cpu actually uses, VID is what the motherboard VRM's supply.

Left on auto, the cpu thinks about its loads, adds in LLC and demands higher voltages from the VRM's than it realistically needs, just to cover all the bases. The motherboard is happy to supply that, and then the cpu only uses some of that.

So limiting VID is like asking mom for $10 and she laughs and gives you $5 and tells you to make do. Overall, that's a benefit, reduces heat, reduces stress on the VRM's, keeps vcore and SVI2 at reasonable levels. Setting a solid vcore has the opposite affect, you must set it high enough to cover single core demands and also multi core demands, but then LLC adds voltage to help with vdroop and stability. So you may set 1.325v solid, but only use 1.125v but the cpu gets supplied 1.375v VID or higher to cover everything if you've got funky vdroop.

So if just setting vcore, you are setting yourself up for failure, you must take LLC into consideration as well, and current limits, load turbo timers etc.

My 3700x runs a 1.232v VID and still gets 4.4GHz boost on many cores, 4.28GHz all core, vcore running @ 1.205v and SVI2 running 1.18-1.2v

vcore
 
Apr 15, 2021
16
0
10
Just tried Cinebench R23 on my Ryzen 5 3600, specs in my sig. At the very start of the test, the CPU was turboing to 4.2ghz all core at about 1.39-1.4v. It soon dropped down to 4.1ghz all core and 1.34v.

I fail to see how running below 100% stock voltage is going to cause the CPUs lifespan to be decreased.

That's what I'm saying, when I first got this 3600x, the turbo is non stop high 1.35-1.45v or even higher sometimes, and the temps get toasty but I understood AMD ryzen series can run pretty hot. Why would they make it that way stock if it's so bad?
 

Phaaze88

Titan
Ambassador
Where would I find that in HWinfo
That.
iu
 
That's what I'm saying, when I first got this 3600x, the turbo is non stop high 1.35-1.45v or even higher sometimes, and the temps get toasty but I understood AMD ryzen series can run pretty hot. Why would they make it that way stock if it's so bad?
What voltage does in a transistor is basically adjust how much current can pass through it. You can think of it as turning the handle on a faucet. Sometimes the processor needs more current to do things, sometimes it doesn't. Voltage has no direct impact on the wear of the part because sometimes little to no current is flowing through it and overall, as long as you're not hammering the CPU 24/7, it's going to operate at lower currents. How much current and how hot the part is over time is what determines how much wear the part experiences, not the voltage.

To take the faucet analogy further, it's like saying a faucet that's partially turned on will degrade even if it's not connected to a plumbing system

EDIT: Also note that temperature readings from the CPU are from the hottest part. There's always going to be a core that's pinged due to handling background tasks.
 
Apr 15, 2021
16
0
10
The test should be done under the defaults, that way you can see how much voltage(SVI2 TFN) your particular chip is fine with.
fnFoa5y.png

it basically started flatlining 3-4 minutes into the test because my temps were going into the 80c+. Is that honestly realistic average though? In my games I never even hit above 70c with constant 1.35v static voltage. I sit around 65c when gaming at 4250 all core, 1.35v set. It sat at about 1.26v current for pretty much 6 minutes straight. Probably because the CPU was getting so hot it was stopping itself from going any higher (just completely stopped boosting clock). As you saw from the other screenshot from previous post, with 4250, 1.35v the 10 minute test didn't even reach above 65c. I always play on lowest graphics because I prefer more FPS.
 
Last edited:

Phaaze88

Titan
Ambassador
~80C is where the cpu stops pursuing higher clocks.
Cinebench R23 is a more common scenario than the likes of Prime 95(small FFT, AVX/ AVX2/ AVX-512 off) and OCCT, which are more of a worst case scenario, but not impossible to reproduce in specific titles.
These games you play are not as heavy.
 
Solution