[SOLVED] How do I change the CPU voltage on a Asus BIOS for Ryzen 3000?

[Kolbe Howard]

This might be a stupid question but I'm having some input issues in the BIOS and I can't seem to figure out how to manually adjust the CPU voltage. I go to the VDDR CPU voltage and change it to manual but then how do I actually enter in the desired voltage level? I click on the box, nothing, hit enter and it pulls up the option to change it to auto again.

The stock voltage is way too high at almost 1.5v so I really need to lower it.

Motherboard is Asus Tuf gaming X570 plus wifi
CPU is Ryzen 3700x.

 

[dimtodim]

This might be a stupid question but I'm having some input issues in the BIOS and I can't seem to figure out how to manually adjust the CPU voltage. I go to the VDDR CPU voltage and change it to manual but then how do I actually enter in the desired voltage level? I click on the box, nothing, hit enter and it pulls up the option to change it to auto again.

The stock voltage is way too high at almost 1.5v so I really need to lower it.

Motherboard is Asus Tuf gaming X570 plus wifi
CPU is Ryzen 3700x.

i have same board only cpu is different...i change offset voltage for -0.1 and works for me...that is under VDDR CPU voltage u will see options

 

[drea.drechsler]

This might be a stupid question but I'm having some input issues in the BIOS and I can't seem to figure out how to manually adjust the CPU voltage. I go to the VDDR CPU voltage and change it to manual but then how do I actually enter in the desired voltage level? I click on the box, nothing, hit enter and it pulls up the option to change it to auto again.

The stock voltage is way too high at almost 1.5v so I really need to lower it.

Motherboard is Asus Tuf gaming X570 plus wifi
CPU is Ryzen 3700x.

What makes you think it's way to high?

AMD's said it will go up to 1.5V when it boosts, so AMD says it's right.

 

[Karadjgne]

What makes you think it's way to high?

AMD's said it will go up to 1.5V when it boosts, so AMD says it's right.

Not the whole story. That 1.5v is for single core only use, when using multiple cores it's advised the cpu not exceed 1.26v at boost for any length of time. Early OCers found that out the hard way in early OC attempts when cpus were burning out in less than 6 months with constant voltages of 1.35v - 1.4v ish range.

 

[drea.drechsler]

Not the whole story. That 1.5v is for single core only use, when using multiple cores it's advised the cpu not exceed 1.26v at boost for any length of time. Early OCers found that out the hard way in early OC attempts when cpus were burning out in less than 6 months with constant voltages of 1.35v - 1.4v ish range.

Did you read the article I linked? The behaviour that AMD's Robert Halleck describes is pretty much that. His response to 'what's a safe voltage' has been fairly consistent: safe is to leave it in AUTO, and when in AUTO you'll see it going as high is 1.5V and in the 1.2 to 1.3 volt range during heavier loads. And more: it will drop to as low as .6V when in sleep states (which is amazingly often, even in the midst of heavy workloads).

Also, saying '1.5V is single core only use' is a bit misleading as Ryzen does not have core-specific voltage. Ryzen boosts a single core only in a lightly threaded workload to as high as it's max boost clock and might need as high as 1.5V when doing it. When it spikes the voltage, all cores will have to see it. Early on an overclocker, TheStilt, wrote that Ryzen does not have functioning LVDR's on the die. You'd need something like an LVDR on die to lower voltage from the single VCore voltage feeding the CPU. I've not read where anybody has suggested that's changed, although there was some discussion it might be different on Threadripper.

It's unfortunate but we have to live with dribbles of good stuff of the sort we get from Robert Halleck, and what little some enthusiasts (TheStilt, Buildzoid, 1Usmus, Der8auer are notables) can gleen from their analysis of CPU's, BIOS code,and leaked engineering data. It seems anybody who gets good information from AMD is also silenced with NDA's. That's been one of Buildzoid's complaints as he said he could ask AMD for more info but he'd have to sign an NDA and then he'd probably not even be able to share what he finds out with his experiments for fear of tripping the line. AMD has to loosen up a little bit and share more info with the enthusiast and DIY community so we're not all left speculating. This all goes to OP thinking he 'really needs to lower' voltage when it's perfectly normal going to 1.5V when it boosts.

 

[Karadjgne]

I can understand the NDA's. AMD has been fighting to get competitive with Intel for years, frivolous lawsuits, Intel bullying game designers etc. They finally get a break and are running with it, Ryzens maintaining high core counts, manageable temps and voltages and not needing oversized motherboard components and power supply and cooling to achieve it, all on 7nm when Intel is still struggling trying to get 10nm process streamlined.

They aren't sharing because they don't want to give Intel any more info specifics than they have to, especially when Ryzens aren't quite finished.

 

[drea.drechsler]

I can understand the NDA's. AMD has been fighting to get competitive with Intel for years, frivolous lawsuits, Intel bullying game designers etc. ....

To be sure, I can see it may be necessary even though I don't get to see how, exactly.

The thing is, Ryzen works substantially different (from Intel's boost algorithms especially) and really has even from 1st gen. People continue to think high(er) voltages are always bad. There's just terribly little coming directly from AMD to explain, not simply that it's 'safest in Auto', but exactly why the up to 1.5V spikes it sees isn't bad (by explaining something about electron migration maybe) and what the processor boost algorithm is doing that makes it safe as the processor starts to heat up in heavy multi-core loads. The demonstrations that Buildzoid has aired on his Youtube channel are really great for that (his rambling aside)...but if he was tied up in an NDA I wonder if he could have done them?

I have to think that if AMD had done something similar themselves, or even 'commissioned' something from one of the techtube enthusiasts, it could have gone a long way to help get the idea across that this is different so you have to approach it differently. But they seem way too guarded with information, even some that shouldn't really need to be a deep secret.