Too Much Voltage, Should I Worry?

[Mike Soda]

After updating to 2.9 from 2.8 BIOS on my MSI B350M Gaming Pro I noticed something odd after trying to reapply my previous stable 3.8GHz OC. Just like before I manually input 1.350V, which the board rounded up to 1.360 which was fine. Now however it's rounding up to 1.4V that I know for 3.8 is very unnecessary. While gaming my temps aren't exceeding 55C, usually in the mid 40's although from past experience I know this voltage would result in 80C or higher when under Aid64 Extreme full load. Here's a screenshot of all the powers for my chip that HW monitor is claiming, to play it safe I'm going to revert my CPU to stock settings for now.

 

1: Is there anything I should worry about enough to justify rolling back to 2.8 BIOS?

 

2: Is there something I can tweak to get the voltage I want to actually set?

[Jurrunio]

Set a lower value like 1.3V or slightly reduce LLC if you turned it on.

[Mike Soda]

9 minutes ago, Jurrunio said:

Set a lower value like 1.3V or slightly reduce LLC if you turned it on.

Do you mean load line calibration? I just had someone else, elsewhere say that mines probably set to 1, 2, or 3 & that if I set it to 5 that should keep the BIOS from bumping my voltage up. I'm leaning more towards your suggestion though because logically lower does make more sense.

[Jurrunio]

3 minutes ago, Mike Soda said:

Do you mean load line calibration? I just had someone else, elsewhere say that mines probably set to 1, 2, or 3 & that if I set it to 5 that should keep the BIOS from bumping my voltage up. I'm leaning more towards your suggestion though because logically lower does make more sense.

Yes, I'm talking about that.

That depends on your BIOS settings and there should be caption or graph that show whether increasing or decreasing the number reduces LCC effect. In my mobo though the higher the value the less LLC effect (level 5 turns it off and is the default setting). I think this is just Asrock only thing, but I'm not sure whether the same applies to yours.

[Mike Soda]

4 minutes ago, Jurrunio said:

Yes, I'm talking about that.

That depends on your BIOS settings and there should be caption or graph that show whether increasing or decreasing the number reduces LCC effect. In my mobo though the higher the value the less LLC effect (level 5 turns it off and is the default setting). I think this is just Asrock only thing, but I'm not sure whether the same applies to yours.

I checked their site & although MSI does talk about it they're not clear on in which direction turns it up or down.

 

https://www.msi.com/blog/why-llc-is-your-friend-when-overclocking

[Mike Soda]

29 minutes ago, Jurrunio said:

Yes, I'm talking about that.

That depends on your BIOS settings and there should be caption or graph that show whether increasing or decreasing the number reduces LCC effect. In my mobo though the higher the value the less LLC effect (level 5 turns it off and is the default setting). I think this is just Asrock only thing, but I'm not sure whether the same applies to yours.

I tried changing it to 3, 5 & then 8 which I determined to be the lowest setting because there in the BIOS it made the 1.350V I set only raise to 1.376. Although that's still way higher than I think it should be for what I think is considered the off setting. I put everything back to Auto until I know otherwise what to do because if something breaks I don't want overvolting to be on me. Although even on Auto HWMonitor shows a max vcore spike of 1.440V.

[Jurrunio]

Just now, Mike Soda said:

I tried changing it to 3, 5 & then 8 which I determined to be the lowest setting because there in the BIOS it made the 1.350V I set only raise to 1.376. Although that's still way higher than I think it should be for what I think is considered the off setting. I put everything back to Auto until I know otherwise what to do because if something breaks I don't want overvolting to be on me. Although even on Auto HWMonitor shows a max vcore spike of 1.440V.

That's scary. 1.44V even on a spike could have done damage. I'd set the manual voltage even lower to compensate for LLC

[Damascus]

[Mike Soda]

23 minutes ago, Damascus said:

You mean I was looking at the wrong thing in HWMonitor? I just panic rolled back to 2.8 BIOS & it's still showing a max spike of 1.4V something. Also it still doesn't change the fact that in the BIOS while set on Auto Core voltage I was seeing it bump up to 1.4 when I specifically set 1.350. I'm going to replicate the OC I had before with this older BIOS & edit/update this post on what it does.

 

*Update: After rolling back to 2.8 form 2.9 BIOS & restoring the OC settings I used for months now all is as it was. 1.350 set results in 1.360V, although I do realize that I was looking at the wrong thing in HWMonitor. It was still showing 1.4V set on the core in BIOS when I input 1.350, until further replies are made I'm just gonna stick with what I got/had originally that I know runs fine & safe.

Edited November 4, 2017 by Mike Soda
new information added

[johndms]

4 hours ago, Mike Soda said:

*Update: After rolling back to 2.8 form 2.9 BIOS & restoring the OC settings I used for months now all is as it was. 1.350 set results in 1.360V, although I do realize that I was looking at the wrong thing in HWMonitor. It was still showing 1.4V set on the core in BIOS when I input 1.350, until further replies are made I'm just gonna stick with what I got/had originally that I know runs fine & safe.

Yeah, CPU VDD is the line to focus on.

 

If you set 1.350 in your bios, it should not be going over by much. 1.381 shouldn't happen. Your goal should be to stay as close to 1.350 during full load without going over. Do note that the highest settings of LLC are extremely dangerous. I'd stay around LLC1-5 and never use 7 or 8 regardless of what those at MSI tell you. Extreme LLC can cause a massive, temporary voltage spike (1.5v, 1.6v) when returning to idle from full load. You won't see this spike in any software monitor. I don't care how good someone says a board is, don't do it.

 

Have a look at Buildzoid's video where he details how LLC works and the dangers of Extreme LLC. He knows what he's talking about.

 

[Mike Soda]

5 hours ago, johndms said:

Yeah, CPU VDD is the line to focus on.

 

If you set 1.350 in your bios, it should not be going over by much. 1.381 shouldn't happen. Your goal should be to stay as close to 1.350 during full load without going over. Do note that the highest settings of LLC are extremely dangerous. I'd stay around LLC1-5 and never use 7 or 8 regardless of what those at MSI tell you. Extreme LLC can cause a massive, temporary voltage spike (1.5v, 1.6v) when returning to idle from full load. You won't see this spike in any software monitor. I don't care how good someone says a board is, don't do it.

 

Have a look at Buildzoid's video where he details how LLC works and the dangers of Extreme LLC. He knows what he's talking about.

 

Thank you that's what I thought as with the older currently used 2.8 BIOS I'm seeing the BIOS VCORE stay at 1.360 while CPU VDD is at 1.331V is the highest I'm seeing while gaming right now VS 1.387V on the 2.9 BIOS last night. What I'm still confused on though is the setting itself, you're telling me I should lower LLC to around 1-5 but I think on my board that means it's high. As I said in my previous post, when I manually input the voltage of 1.350V the Auto LLC setting put it at 1.4V. That didn't come down at all until I tried an LLC of 5 & it came down further to 1.376V when I tried LLC of 8. I don't know why all motherboard manufactures can't just use the same logic because I'm very confused right now. I've checked MSI's site too, they use LLC one as an example but the way they word it, they make it sound like the highest. If you could put into laymens terms or find me a better source other than what I'm gonna link below again I'd greatly appreciate it.

 

https://www.msi.com/blog/why-llc-is-your-friend-when-overclocking

[johndms]

@Mike SodaI'm going TL;DR mode here. I'm too lazy to read about LLC when I know how it works. As far as I know, ASRock is the only one to go in reverse.. 5 being lowest, 1 being highest. On your board, 1 should be the lowest setting. This is very easy to test. Enter 1.30 in bios with LLC3. How low does the vcore drop when you hit it with a load? Compare that to 1.30 with LLC6. (I'm intentionally staying away from 1 and 8 because we don't know which one is the highest). Compare the results. Which forces the vcore to stay closer to 1.30v? Ignore how HIGH it goes, you need to focus on how low it drops.

 

And by "hit it with a load" I don't mean a gentle load.. full on stress test for a few minutes.

[Mike Soda]

3 hours ago, johndms said:

@Mike SodaI'm going TL;DR mode here. I'm too lazy to read about LLC when I know how it works. As far as I know, ASRock is the only one to go in reverse.. 5 being lowest, 1 being highest. On your board, 1 should be the lowest setting. This is very easy to test. Enter 1.30 in bios with LLC3. How low does the vcore drop when you hit it with a load? Compare that to 1.30 with LLC6. (I'm intentionally staying away from 1 and 8 because we don't know which one is the highest). Compare the results. Which forces the vcore to stay closer to 1.30v? Ignore how HIGH it goes, you need to focus on how low it drops.

 

And by "hit it with a load" I don't mean a gentle load.. full on stress test for a few minutes.

Okay well here's my results & conclusion, let me know if ya agree on my decision for now until future BIOS's or if I should still change something for the long run.

 

2.8 BIOS, 3.7GHz set on 1.3V LLC 3: VCORE 1.304-1.312V, CPU VDD 1.250-1.294V

2.8 BIOS, 3.7GHz set on 1.3V LLC 6: VCORE 1.288-1.304V, CPU VDD 1.238-1.288V

2.9 BIOS, 3.7GHz set on 1.3V LLC 3: VCORE 1.304-1.352V, CPU VDD 1.294-1.338V, VCORE usually at that max while VDD 1.30-1.306V

2.9 BIOS, 3.7GHz set on 1.3V LLC 6: VCORE 1.304-1.344V, CPU VDD 1.274-1.331V, VCORE usually at 1.336V while VDD 1.288-1.306V

 

Now for the settings I'm currently using & have been for the past 2 months since I first got 2.8 BIOS, never encountered any stability issues.

 

2.8 BIOS, 3.8GHz set on 1.350V LLC Auto: VCORE 1.352-1.360V, CPU VDD 1.300-1.344V, VCORE usually at 1.360 while VDD 1.306-1.319V

 

So it looks like the higher the number on LLC the less likely it is to over-volt. Although IMO the 2.9 BIOS with any LLC settings results in greater extremes in terms of voltage highs & lows compared to 2.8. I've also tested the last setup listed many weeks ago & it passed a 12+ hour Aida64 Extreme test. Finally, I noticed that although I was setting the core clock & voltage lower on 2.9 BIOS, my temperatures were the same my higher currently used OC & voltage on 2.8. So that newer BIOS runs hotter too because it favors 0.50+ more V regardless of setting.