Strange voltage issues in BIOS - Gigabyte Z170mx Gaming 5

[alegav]

hey guys!!

So this is going to be a bit of along post but I know better then to just start a thread with, "Help!! Something is wrong!" and not provide any information or testing, here we go

So I have been sitting happy at 4.5GHz for almot a month now. Last night I noticed that the OS was a bit jumpy. Mouse lag,, audio clipping when listing to music (on machine not streaming) SO I started poking around. I took a look at my I/O and everything was normal. Nothing was pegging the disk or CPU. So I restarted, no change.

I opened up event viewer and took a look at the system log and I found it riddled with the below. I'm talking every 3 seconds it was just crazy!

Event ID: 17
Source: WHEA-Logger
A corrected hardware error has occurred.
Component: PCI Express Root Port
Error Source: Advanced Error Reporting (PCI Express)
Busevice:Function: 0x0:0x7:0x0
Vendor IDevice ID: 0x8086:0x340e
Class Code: 0x30400

Event ID: 17
Source: WHEA-Logger
A corrected hardware error has occurred.
Component: PCI Express Root Port
Error Source: Advanced Error Reporting (PCI Express)
Busevice:Function: 0x0:0x7:0x0
Vendor IDevice ID: 0x8086:0x340e
Class Code: 0x30400

So I googled and came across multiple articles stating its a power issue and to check overclock. So I did.

I used CPUid, OCCT, CoreTemp and HWZ MOnitor to check what my CPU was hitting when under full load. It maxed out at 1.280. So I went into my BIOS (mind you my vCore settings has been in "AUTO" until now) and input 1.350 thinking that should be plenty be plenty based on the readings. So I reboot and get into windows and check for the error messages, those are gone! and so was the strange mouse jumpiness and other issues. BUT THEN!!! I decided to just run Prime95 and check temps and voltages.

So using the blend test it goes for about 1 min and errors out. I thought to myself that's interesting. I was able to run this for 8 hours with the auto setting for vCore and that maxed out at 1.280v... So I did it again and took a look at the max voltage in the programs mentions above and it never hits the 1.350v it gets to 1.325 and that's it. So I tested again, set it to 1.390v (I know that's super high but just for testing) Ran Prime95 for 3 hours no issues what so ever. Checked max voltage and it only got as high as 1.375v.

So here in lies my questions haha. Is there some other setting in my BIOS that could be attributing to the offset? Most everything is set to auto minus the frequency and now vcore. It just never hits the set voltage. Temps are good 55*-65* under full load (H80i v2 AIO) I can grab BIOS screenshots (later this evening) I am using F3 BIOS version but all other drivers are the latest. I have also added an SSD but I cant see that causing this issue. Other than that no other changes have been made to teh system or any new hardware/software installed.

As always thanks!
Alex

 

[Jeff G]

Do you have an option in the Bios for LLC, usually like 1-5 if I remember correctly. Set it to 4 or 5 if you do and see if that keeps a better voltage.

 

[alegav]

I do have an option for that and I did look into that but I thought that was related to Vdroop. You think that could also be causing the the issue? I may have misunderstood but I basically thought that LLC let the volts drop when the CPU was not under full load. I can certainly try it and see if it lets me hit my target voltage.

Looks like you'e probably right, based on this http://www.overclock.net/t/197804/what-is-vdroop

 

[Jeff G]

Out of the manual:

 

[alegav]

Welp Jeff, you're the man and all I had to do was RTFM!

 

[Jeff G]

Lol, glad to help!

 

[trents]

LLC was originally designed to address vdrop, I think and it is implemented in different ways on different boards and in my experience it typically affects voltages at both ends of the load spectrum.

 

[wingman99]

Out of the manual:
View attachment 181029

Like the Quote here intel made the voltage droop so it would not over shoot the voltage when the CPU load has been lifted quickly.

It is strange that your PC was running well for a month and now it needs more voltage.

 

[alegav]

Like the Quote here intel made the voltage droop so it would not over shoot the voltage when the CPU load has been lifted quickly.

It is strange that your PC was running well for a month and now it needs more voltage.

Or I just never noticed it? Not sure exactly but this does have some legs and I'll test tonight. I should be able to hit 4.5GHz with pretty low voltage just based on the "Auto" settings in the mobo. I think I'll start at 1.350v and work backwards after adjusting the LLC.

I might leave it on auto and just mess with LLC also. Hmmmm

 

[wingman99]

My peak voltage with load is 1.332v at 4.5GHz with LLC on Auto prime95 tested for 15 hours with the round off checking option enabled. I hope I don't have a problem in a month, I can't go much higher with voltage running air cooled.

 

[alegav]

You check your event logs for (system logs) the same power issue that I had? I'm going to play around with some settings and see what happens tonight. On my first OC I did a 6 hour blend test and it was fine. Hmmmm I wonder if it has anything to do with my video card since it's a PCIE error. I was running a slight overclock but GPU was not under any load when I noticed the strange stuff. I think I'm going to first test with leaving voltage on auto but up the LLC and see what happens.

 

[wingman99]

I checked and my system logs are fine for the 2 weeks that I have had a fresh install of windows 10.

 

[alegav]

I checked and my system logs are fine for the 2 weeks that I have had a fresh install of windows 10.

Well that's good news. Ok so I set LLC to high and am testing with 1.332 volts. Checked system logs and no more power issue. Checked voltage and it's now hitting what I set. Going to kick off prime blend test and see where we are at in the morning.

Last question and I'll let you guys go to bed . Is having the target voltage always running through the CPU not a recommended thing? I know with overclocking it's risk for reward just curious. I think the voltage is low enough and my temps are pretty good. If I can get 3-5 years on CPU I'm happy

As always you guys have been a tremendance help.

 

[wingman99]

I use all the power saving features after I have a stable overclock. Well you can have the CPU clocked at 4.5GHz with full voltage at Idle load, people say it should last a long time. The times I have had a CPU show trouble it was a slow degrade. I see vary few people here that have trouble after a while.

 

[alegav]

*UPDATE*

Going strong at 1.332 volts. No error messages. Its been about 9 hours so that's some pretty good news. Temps look good, I did have 1 core hit 78* but im not to worried about it since prime really punishes the CPU. Other than that temps are still 55*-65*

 

[wingman99]

Glad it went well for you, what was your Ambient temp? I really like prime95 I trust it to test all the transistors in the CPU so I don't have problems down the road, my CPU peaked one core at 84c.

 

[alegav]

Glad it went well for you, what was your Ambient temp? I really like prime95 I trust it to test all the transistors in the CPU so I don't have problems down the road, my CPU peaked one core at 84c.

I am going to assume you mean like the temp of the room if so then my office is at about 22* to 25* Celsius

 

[wingman99]

Yes I was asking about your room temperature, I'm running the same here for room temperature. Those water coolers are vary decent.

 

[alegav]

Yes I was asking about your room temperature, I'm running the same here for room temperature. Those water coolers are vary decent.

Ya they do pretty good and I dont have a crazy beefy one either. I was on the fence at first just didn't like the idea of liquid and electronic components haha. Another update, looks like we are good at 4.5GHz with 1.332 volts and absolutely no more errors.

 

[wingman99]

That is great to hear that you don't have any more problems.