VCore extremely low

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

redeyedskink

Proper
Jan 19, 2019
58
0
130
Hi everyone,

I posted this thread on the GPU forum because I thought it might be a gpu issue, but I may have been wrong.

My computer will randomly blackscreen and restart at times. I've tested nearly everything and then found a weird issue with my voltages.

My voltages tend to fluctuate, especially my vcore. Now i know due to speedstep that a vcore should change, but my min and max value are along the lines of .14 and 1.152. .14 seems extremely low to me and was wondering if this might be a cpu issue or a psu issue.

Thanks for your help.
Specs:
I5-6600k
gtx 960
z170 gaming 5 motherboard
2x8 2200mhz ram
seasonic 550w psu
 
Solution
Cool! No problem.

Still wonder what the problem is. Manual voltage is a good workaround, but it doesn't solve the issue completely.

But hey, at least there is a workaround LOL.


Even using FFT on prime95, my voltages don't move at all. They stay steady at 3.305, 11.88, and 4.98. could it be a bad sensor? My CPU temps don't make any sense either. They are at 38 c under load in prime95. Maybe it is the mb that is the major problem?
 


I'm using the installed version
 
Something doesn't seem right somewhere. I hate to say this, especially given the type of unit that is and the fact that we don't see this often on Seasonic units, but if were MY unit, and I was seeing less than 12v on a brand new unit, I think I'd consider returning it for a replacement given the 12v value BUT it might be a good idea to test the unit with a multimeter first if you own one, can buy one (You can get a pretty inexpensive multimeter at any home center or Harbor freight, even Walmart.) or borrow one.

https://www.youtube.com/watch?v=ac7YMUcMjbw
 


I'll see if I can go out and buy a multimeter in order to test it out. I might just use the unit in the meantime and order another one from amazon and then swap them out. Glad it's modular lol
 


What am I looking for when I test with a multimeter? To see if my 12v rail is above 12v? 5v rail above 5? and 3.3v rail above 3.3?

Do I also test just the 24pin or should I test them all?
 
It is specifically outlined in that video tutorial I linked to. Every step of what you want to do.

And yes, you generally want to see above 12v, 5v and 3v, ESPECIALLY on a brand new unit. Even more so on a very good Seasonic. There is no way I'd accept a below 12v reading on a brand new Seasonic PSU. I'd return it immediately with an explanation as to why if that turns out to be the case.

Seasonic is a very good, trustworthy manufacturer, and they ARE a manufacturer, not just a "brand" that has other companies manufacture their power supplies, like Corsair, Antec, Thermaltake, XFX, and many others, but they are still vulnerable to the occasional sub-par unit sneaking through quality assurance.
 


Alright. I just got done with the tests. Here are the results

1 3.3v- 3.31
2. 3.3v-3.31
4. 5v-5v
6. 5v-5v
10. 12v-12v
11. 12v-12v
12. 3.3v-3.31

13. 3.3v-3.31
14. -12v- -12.25
21. 5v- 5v
22. 5v- 5v
23. 5v-5v

Everything is perfect it seems
 
Yes, it would seem so. That is why I wanted you to check and verify. Seems that something is off elsewhere, so either software or motherboard is causing an erroneous reading.

At this point, given the history of this thread, I might bend gravitating a little more towards motherboard now. So if you have any further problems, I think that might be what I'd start looking at. I recently had to replace my Z170X-Gaming 5 with a Hero VIII off Amazon and my voltage issues disappeared after that. It was only in use for maybe two years when the problems began. I still have it as the problem was never a factor at the stock settings, only with the overclocked configuration. Well, that and the fact that Gigabyte Z170 boards only let you make .020v incremental adjustments to the memory voltage which was not helpful when trying to tune the memory. Other boards generally allow .005v increments to the memory DRAM voltage.
 


Sounds good. If I have any more issues then I'll get a new MB then. Thank you for the help! I learned quite a bit from all of this
 
Idk if this is helpful, but I also get super low reported Vcore values at idle. Like down to 0.012 V. Gigabyte GA-Z170M-D3H, 6700K. I don't think it's always been like that, and I have no idea when it started happening, but I've never had any actual issues so I didn't worry about it. Figured the sensor was buggy at low values or something. Load voltages look as expected.

Only happens if I leave voltage as variable, if I set a fixed voltage it will always be reported at that level. Haven't investigated whether it's related to things like speed step or c state settings in the BIOS.
 
Now that just got my attention because you and him are the first two instances of this I've heard of. I will look into this. Maybe there's something to it. Perhaps a chipset driver issue or something with the firmware. I wouldn't even be surprised to find that it was a Windows related issue due to patches affecting the various vulnerabilities that we've seen over this last year.
 
Some variation in Vcore is no issue but when you have large variations, like this https://imgur.com/a/NYuLbA9, i feel like that can cause really large issues like my blackscreen and restart issues. It's weird because the Vcore seems stable during higher loads so playing games on my pc shouldn't be an issue. I noticed that many of my blackscreen crashes happened during load screens and stuff and maybe the vcore was down volting and then when the pc demanded more cpu power, the computer would crash because there wasn't enough. Like I said, I'm going to keep seeing if this new PSU solved the issue, but I haven't had any issues (knock on wood) since installing the new PSU. I might multimeter the old PSU and see what readings I get too
 


Out of curiosity, what bios version do you have?
 

22F, the latest. Came out last spring, I think it was specifically to address spectre/meltdown vulnerabilities.

What settings were you using for the OCCT 3D test? I find that if I have my PC on the Balanced power setting, my PC will run the benchmark full speed for a bit but then the CPU/GPU will downclock as if they're reverting to a low power state and my fps will tank. In that case I did notice some major Vcore swings. Running my PC in High Performance, or running the benchmark in full screen, seemed to result in consistent performance.
 
I have also found that there are problems with the implementation of Speed shift on Z170. Speed step seems to work fine, and it's not on every system or every BIOS version either, but if you have a BIOS version that is beyond the last version which had a microcode update, it can't easily be rolled back (The microcode changes) even if the rest of the settings can, and some can't. In some cases you simply cannot roll back beyond a certain BIOS version without a complicated and risky process involved. Anyhow, I've seen some really weird unexplained behavior to Speed shift after the BIOS changes and Windows patches in December/January of 2017/18.

Not sure that it's related, but it could be to some degree. Still going to check into this low voltage drop thing because that's a problem if it's happening to many systems.
 


I always tested using high performance. I have my vcore voltage statically set at 1.11 right now and I'm not having anymore voltage issues. I don't overclock or anything, but I might consider it once the computer/parts get a little older