Messing with voltages makes the OC less stable and even causes BSOD.

surensm said:

Not yet, won't be back home until tomorrow.



Very possible, with my luck. I'll try and mess with DVID, but I think the setting in my BIOS was greyed out.

Click to expand...

If you want to use DVID set Vcore to normal. In the right panel of Bios it tells what to do.

I did not see any mention of what BIOS you're running. If you have an early on BIOS it may not fully support DDR4 3200. Try updadating the BIOS to at least version F3 if you're not already on that version of higher. If you notice on that version they improved DDR XMP compatibility.

wingman99 said:

If you want to use DVID set Vcore to normal. In the right panel of Bios it tells what to do.

Click to expand...

I left it all in Auto (see pic below).

Mandrake4565 said:

I did not see any mention of what BIOS you're running. If you have an early on BIOS it may not fully support DDR4 3200. Try updadating the BIOS to at least version F3 if you're not already on that version of higher. If you notice on that version they improved DDR XMP compatibility.

Click to expand...

I believe it is F3, I'll be restarting soon anyway to bump the multiplier to start stress testing again, so I'll double-check.

@trents, Memtest86 is running as I write this. Currenty 1 pass in and no errors.

So here is a dumb question: I can't mess with DVID just yet, probably a stupid mistake I'm making somewhere. I can, however, mess with the loadline calibration. I'm guessing it's an either/or matter? I would have to pick between either setting as far as adjustment goes? Or would it be ok to mess with both at the same time down the road?



Also, my DRAM voltage is set to auto, and it says 1.2V on the BIOS Settings screen. Whenever I used to try and mess with this setting, it would only let me change it with 0.02V increments (as in I could never set it to 1.35V, either 1.34V or 1.36V), and my RAM is rated at 1.35V as seen in the last pic. Or should I just leave DRAM voltages alone altogether?



P.S. yes, my monitor is absolute crap for the GPU I'm running, and no, there is no logical reason for this besides me being too stingy to buy a full on 4K display atm.

If the memory passes memtest86 just leave it at stock XMP default settings. You can use whatever settings that make the total Vcore with load is what counts in the end of tinkering. The total amount of Vcore with load is the only thing that counts when overclocking the Cores with mild Intel overclock.

wingman99 said:

If the memory passes memtest86 just leave it at stock XMP default settings. You can use whatever settings that make the total Vcore with load is what counts in the end of tinkering. The total amount of Vcore with load is the only thing that counts when overclocking the Cores with mild Intel overclock.

Click to expand...

Got it. Well, it's about to complete pass 3. I'm going to leave it overnight, and over the work day tomorrow. That gives me another 20 or so hours of testing on top of the 7 and a half hours that it has done so far. Speaking of, is 7+ hours for 4 passes sound about right to everyone here?

Well, some news:

Memtest86 gave me a black screen. Last I checked it was at Pass 5 with no errors, but I might just run it again. I'll double check to make sure I wasn't running Memtest86+. Currently running prime95 and HWMonitor. Hasn't been 5 minutes yet, but it hasn't crashed either. Multiplier is set to 43 (4.3 Ghz), XMP is set on, and everything else is at Auto for testing purposes. I may just have to learn more about loadline calibration and DVID (which is still grey even though main voltage is set to AUTO) to get this OC to run properly stable. In the meantime, here are some values from the last 5 or so minutes.

Normally, Memtest86+ report lines turn red on blue from white on blue when it encounters errors. It won't always find them on the first pass so it's a good idea to let it run long enough to complete 3 or 4 passes. Make sure your sleep/hibernation stuff is disabled in bios. Your're running it from a self-booting CD/DVD disk or from a self-booting USB flash drive, right?

trents said:

Normally, Memtest86+ report lines turn red on blue from white on blue when it encounters errors. It won't always find them on the first pass so it's a good idea to let it run long enough to complete 3 or 4 passes. Make sure your sleep/hibernation stuff is disabled in bios. Your're running it from a self-booting CD/DVD disk or from a self-booting USB flash drive, right?

Click to expand...

Yes, I was running it from a USB flash drive. It ran for 5 passes with no errors to report, specifically for around 12 hours. I woke up at around 2am and it was still at pass 5. It wasn't until I woke up again at 4:30am that I saw the black screen, so it could've been at any point in those 2 and a half hours.

Well, three hours in, and here are some HWMonitor results. Prime95 still hasn't crashed. Last time OTTC crashed at 3 hours and 30 minutes, so let's see if prime95 beats that time.

Conditions: Multiplier is set to 43 (4.3 Ghz), XMP is set on, and everything else is at Auto for testing purposes.

4 hours in, so far so good on the prime95 side of things. Also, found this article about LLC for those interested:


Next up, 4.4GHz on Auto, and see where that goes. The voltages are quite high for 4.3GHz as seen in the screen shot above, but the temps aren't horrible. But what are the chances that both 6700k CPUs I got are falling short on the silicone lottery?

Edit: Well, not good. At 4.4, Prime95 crashes just 5 seconds in. At 4.5, the computer threw a BSOD before even reaching the log-in screen. I guess it's time to start messing with the voltages manually.

surensm said:

Well, some news:

Memtest86 gave me a black screen. Last I checked it was at Pass 5 with no errors, but I might just run it again. I'll double check to make sure I wasn't running Memtest86+. Currently running prime95 and HWMonitor. Hasn't been 5 minutes yet, but it hasn't crashed either. Multiplier is set to 43 (4.3 Ghz), XMP is set on, and everything else is at Auto for testing purposes. I may just have to learn more about loadline calibration and DVID (which is still grey even though main voltage is set to AUTO) to get this OC to run properly stable. In the meantime, here are some values from the last 5 or so minutes.

Click to expand...

The Vcore looks good also the Temperature looks fine.

- - - Updated - - -

surensm said:

4 hours in, so far so good on the prime95 side of things. Also, found this article about LLC for those interested:

https://www.reddit.com/r/overclocking/comments/28qlv4

Next up, 4.4GHz on Auto, and see where that goes. The voltages are quite high for 4.3GHz as seen in the screen shot above, but the temps aren't horrible. But what are the chances that both 6700k CPUs I got are falling short on the silicone lottery?

Edit: Well, not good. At 4.4, Prime95 crashes just 5 seconds in. At 4.5, the computer threw a BSOD before even reaching the log-in screen. I guess it's time to start messing with the voltages manually.

Click to expand...

What does CPU-Z show for Vcore?

wingman99 said:

The Vcore looks good also the Temperature looks fine.

- - - Updated - - -



What does CPU-Z show for Vcore?

Click to expand...

Here's the most recent update:

I'm about to pull this motherboard out and set it on fire. I switched the voltage to manual, set it to 1.425, then went ahead and tried starting it at 4.6 GHz. BSODs up the behind. Then I went and set the VCore to Normal, and then went ahead and switched DVID on and set it to 0.085V (I figured 1.3+0.085V = 1.385, not a bad place to start, plus accounting for whatever issues might come up down the road). I then dropped the multiplier to 45, and restarted the computer. The first time around, it just failed to POST and wanted to restore the BIOS to factory settings. Apparently, all I had to do to remedy this is re-enter the bios and restart the system. Well, here is the most recent HWMonitor test of Prime95 (which lasted less than 5 seconds).



Note the voltages. Why in the hell is it drawing nearly 1.43V at 4.5 GHz? Supposedly it's not supposed to go over 1.385? At 4.3 it peaked at 1.368? How does that make sense? This is driving me downright nuts at this point, I'm just going to leave it alone and come back tomorrow after work and mess with it some more.

Here are some shots of my BIOS settings:

Are you running LLC on Auto? if your not I would try that it will lower the Vcore under load. For DVID it is VID (1.363v) + 0.085V

What does CPU-Z show for Vcore? I think HWmonitor is reading incorrectly.

Mine adds up correctly 1.33v.

133v. That's kind of high. Did you mean 1.33v?

wingman99 said:

Are you running LLC on Auto? if your not I would try that it will lower the Vcore under load. For DVID it is VID (1.363v) + 0.085V

What does CPU-Z show for Vcore? I think HWmonitor is reading incorrectly.

Mine adds up correctly 1.33v.

View attachment 188118

View attachment 188119

Click to expand...

How come yours is showing 1.26V for Normal, while mine shows 1.3V? Or is it dependent on the voltage that was registered upon boot?

surensm said:

How come yours is showing 1.26V for Normal, while mine shows 1.3V? Or is it dependent on the voltage that was registered upon boot?

Click to expand...

I think it depends on the CPU. How does CPU-Z Vcore compare?

wingman99 said:

I think it depends on the CPU. How does CPU-Z Vcore compare?

Click to expand...

I'm using HWMonitor, and I'm not sure if anyone saw the most recent update, since I haven't gotten feedback for that yet.


Here's the most recent update:

I'm about to pull this motherboard out and set it on fire. I switched the voltage to manual, set it to 1.425, then went ahead and tried starting it at 4.6 GHz. BSODs up the behind. Then I went and set the VCore to Normal, and then went ahead and switched DVID on and set it to 0.085V (I figured 1.3+0.085V = 1.385, not a bad place to start, plus accounting for whatever issues might come up down the road). I then dropped the multiplier to 45, and restarted the computer. The first time around, it just failed to POST and wanted to restore the BIOS to factory settings. Apparently, all I had to do to remedy this is re-enter the bios and restart the system. Well, here is the most recent HWMonitor test of Prime95 (which lasted less than 5 seconds).



Note the voltages. Why in the hell is it drawing nearly 1.43V at 4.5 GHz? Supposedly it's not supposed to go over 1.385? At 4.3 it peaked at 1.368? How does that make sense? This is driving me downright nuts at this point, I'm just going to leave it alone and come back tomorrow after work and mess with it some more.

Here are some shots of my BIOS settings:

I think HWmonitor is reading incorrectly, software does not always read correctly, try CPU-Z.

wingman99 said:

I think HWmonitor is reading incorrectly, software does not always read correctly, try CPU-Z.

Click to expand...

Will do. My biggest issue is the lack of stability from the current OC, as it won't last 5 seconds in prime95. Should I try bumping the DVID to 0.100?

What clock are you trying? if it is 4.5GHz you could just give 0.100 a try, it would not cause any harm.

What version of prime95 are you using?