Windows = Borked

D3K

D3K

D3K

Soldato
Joined
13 Nov 2014
Posts
3,967
Looking for some advice on a) overclocking and b) what to do about my windows install

Story of fail is as follows:

I followed the guide here relating to my mobo for overclocking:

http://www.overclock.net/t/1490835/the-gigabyte-z97x-overclocking-guide

I went for the 4.5 o/c template with the following values:

Frequency:
CPU Clock Ratio: 45x
XMP: Enable
Uncore Ratio: 40x or 43x
C1E: Disable
C6/C7: Disable
CPU Thermal Monitor: Disable
C3: Enable
EIST: Disable
Voltage:
CPU VRIN Loadline Calibration: Extreme
CPU Phase Control: Extreme Performance
VRIN Override: 2.2v
VCore: 1.275v
Ring Voltage: 1.1v

I disabled Turbo boost on top of that.

All was well. I then went for 4.6, and upped the following values. This was just my gut feeling based on the trend above:

VRIN override: 2.25v
VCore: 1.3v
Ring voltage: 1.15v

The one facepalm was that intel gfx had switched itself back on in the BIOS. No idea why, but my BIOS was having difficulty remembering settings after I rebuilt the machine to take a H110i GT.

However, I didn't notice this until running an Intel XTU bench. It showed me the graphics clock, and the penny dropped. Alas, the screen froze, and with the fans still running, I panicked and did a hard reset. Bad mistake.

I switched off Intel gfx, ran it again, and got a blue screen with:

WHEA_UNCORRECTABLE_ERROR

Rebooted, ran CMD: "sfc /scannow"

Irreparable problems found.

Ran "dism /online /cleanup-image /restorehealth"

This also failed. Tried again by adding the path to the install.wim file from Win 8.1 media, and it still failed.

Last resort I did a system restore which apparently went OK. But I've done another bench in Intel XTU and a stress in OCCT and got the same blue screen.

From what I understand, there is no other way to fix this than resintalling / restoring from backup. I have one from a few weeks ago but I'm hoping to avoid that if possible. The general running of Windows seems fine, I just can't bench at 4.6 (haven't tried 4.5 again yet).

Were my voltages bad?
Should I carry on as is and wait til the 29th to "fix" windows with win10?

TIA
 
No idea, seems to have disappeared from existence in the eventvwr.

But dmp file shows hal.dll (Hardware Abstraction Layer) as the cause. I possibly corrupted it doing the hard reset.
 
Using commands that alter system files whilst running an unstable overclock is a bad idea.

WHEA_UNCORRECTABLE_ERROR is a normal error to see when overclocking. It just means your current settings are unstable.
 
That's a hell of a lot of VRIN my friend. Most people go with 0.5v more that their vcore, (so 1.8v at 1.3v vcore)

That's also a lot of vcore for that overclock. 1.25v would most likely be more than enough for 4.6ghz, though obviously YMMV.

1.2 ring at 42x cache would be what I would try next, or maybe even just set it at 40x @ 1.2v just to keep it simple.

What speed is your ram? If its high (2400MHz or so), then set it down to 1600MHz until you find a stable OC, because you might have to tweak another voltage or two to get high speed ram working properly.
 
Last edited:
I've tried your suggestions new boy, and it didn't work. BSODs on those settings.

Since then I've tried a variety of settings across the board after watching some other people's setups, and nothing works - it's either a BSOD or a system reset 2secs into an Intel XTU benchmark (stress test is fine).

I managed to get a 4.6 bench completed with 1.325 vcore, 1.9v vrin, and 1.2 ring and thought I was making progress. Turned up my case fans and ran it again, system reset.

I've been at this all day and no further forward! It looks like 4.5 is all I can do so not much point when turbo is 4.4. Also considering I've had the RAM on 1333 it terrifies me to have to start tweaking it all again to accommodate whatever I potentially get working.

I have a feeling it's the gigabyte mobo, seen some reports that it doesn't handle voltages correctly.
 
Last edited:
Could try flashing mobo to newest bios (at stock speeds)

Voltage tweaks to get ram running higher would be voltages you don't need to mess with during a cpu overclockers like system agent voltage.

Sorry to hear about your troubles. Not really sure what else to suggest because you are already running some high voltages.

My 4790k does 4.7GHz at 1.218v vcore 1.2v ring 1.72v vrin...

You running 40x cache at 1.2 ring yeah? You could go a little higher on ring, but you shouldn't need to at 40x cache tbh
 
Last edited:
Ran cache at settings between 40x and 43x. Would I need to up the ring for 43?

Unfortunately there hasn't been a new bios for a while.

I'm gonna tinker a little more tomorrow, but failing that I'm considering a reinstall of the OS. There is corruption with the install that can't be fixed, but I'm not sure if that is having a direct effect. The system reset in XTU was fixed by upping the ring, but the BSODs keep coming.

On another note: in OCCT, is AVX Capable Linpack supposed to send temps absolutely mental? I tried that at one point and saw spikes of 85 and quickly disabled it. None of the other stress tests went as high.
 
Rodger.

The beta drivers for this mobo only list Intel gen 5 update. I'll skip them for now.

Ran a memtest over night. The Windows advanced memtest froze at 21% on both and individual sticks. Memtest86+ took 6.5hrs for 3 passes on 1 core @1333. Running individual sticks now on all cores, but it it hasn't updated past test 8 although it's still processing patterns. Having to babysit it to force the next test.
 
Ok so I've tried each stick in every slot (8 runs) at 1333, and a final run of 1 stick in 1 slot at 2400 and all results are the same:

ac8zAO1.jpg

They fail at exactly the same point each time on test 11. The results feel inconclusive. Would this indicate that both sticks are faulty in the same way, or that the mobo is at fault?
 
Ah, I see.

I'm not really familiar with memory errors, as it's not something I've had to deal with before. You say your running each stick individually, and both sticks are failing the test? Sounds strange to me.
 
Might be worth trying a +0.200 offset on both digi and analog I/O voltages too.
My 4770K is perfect at 4.5gig (40 uncore), 1.24vcore, 1.175vring, and the above I/O volts.

May not help, but you've tried most other things...

EDIT I also leave turbo, EIST and all C-states on auto- otherwise the chip is banging 4.5 all the time, with no stepdown when idle.
 
Might aswell add .03v to system agent while your at it.

Though it should be able to run at 1333MHz at stock agent / IO voltages...
 
Reset everything to stock and run your RAM through memtest. If there are any errors at all then it's faulty and needs RMA.

There's no point trying to tweak a fundamentally unstable system.
 
You must log in or register to reply here.
Back
Top Bottom