7700K OC'in time - voltages?

Hmm the only thing I've done is played with the CPU multiplier and the Vcore in manual mode and I was stable after 24 hours in Realbench @ 1.325v

 

I suspect I may be stable under XTU / Prime95 @ 1.350 Manual (Actual is around 1.355-1.36). Prime95 passed an hour - going to do more this evening.

So it could of been because I was using offset(?)

I doubt all my changes are required - I might just revert back to Changing Memory, Vcore and LLC only - (LLC if left on auto puts 1.350 to about 1.396 (LLC 7).

We'll see...

 

Get a rock solid OC with just the basics 1st.

 

It may be rock solid with manual vcore @ 1.35

IF we ignore Realbench -

I know realbench is a go-to for many, but honestly I've always had the lux mark section of it crash / crash my GPU drivers :/ - no instability GPU wise otherwise.

Previous 4970k / 980Ti's also had the same problem with Realbench :/

We'll see

 

Hi there

If RB crashing during the OpenCL I woud think or thought so is can be down TDR, did you check Event Viewer why is crashing during the OpenCL compute?

Hope this helps

Thanks, Jura

 

Will try Thanks,

----------

Currently running a stress using Adaptive voltage set to 1.350 / LLC 5 - offset -00.5 (as that keeps it in the 1.355/1.36 area)

- Quick tip (though this may just be for me being an idiot) - watch out for thing's you wouldn't nec. believe to use AVX instructions - Firefox does, and it's crashed me with AVX to 2 - so I've set it to 3 and it seems OK.

 

Faulting application name: LuxMark-x64.exe, version: 0.0.0.0, time stamp: 0x5392d759
Faulting module name: MSVCR120.dll, version: 12.0.21005.1, time stamp: 0x524f83ff
Exception code: 0xc0000409
Fault offset: 0x0000000000074a30
Faulting process ID: 0x24b4
Faulting application start time: 0x01d2767b9e8dd412
Faulting application path: C:\Users\Downloads\RealBench_v2.44\RealBench\luxmark\LuxMark-x64.exe
Faulting module path: C:\WINDOWS\SYSTEM32\MSVCR120.dll
Report ID: 70600252-2141-4203-9488-423ca28108dd
Faulting package full name:
Faulting package-relative application ID:

---

Oh and I've just tested without an OC (Auto settings, no OC) - Luxmark still crashed >.>

 

that msvcr120 seems to be visualc++ redist issue, did you update it recently or maybe try to update with that visualc+ redist aio installer v48.

 

Hi there

As above - TJ- pointed out this can be down to MS redistribute or you can try this get msvcp120.dll and msvcr120.dll from Realbench folder or where are you downloaded Realbench and a make a copy to Realbench directory and system32 folder.

Hope this helps

Thanks, Jura

 

Tried both - same issue :/

O'well - Guess I should just test in Prime95 / OCCT / HWbot x256 and alike?

 

Prime95 BSOD

Re-testing now with slightly higher offset.

http://imgur.com/a/GwF5i

 

Man that chip doesn't wanna play nice!!!

Good luck!

 

Haha true, that and I'm very "average" at doing this in the first place!

I managed to run a Prime95 1344 FFT last night for 1hr 30 - without issue, obviously more testing required but that's the best so far. This was with a slight increase in offset, normal load (e.g. gaming is about 1.355 / 1.36 but prime95 1.376)

I'll be leaving it going tonight for at least 4 hours to test.

 

& We appear stable

Real-world usage still to come, but initial results are promising.

• Vcore @ 1.320 (Adapative)
• Offset + .001
• LLC Level 4
• XMP On

Essentially the issue appears to of been the LLC. I've reduced my vcore setting to 1.320 as LLC4 with this set to 1.350 would see it shoot up to 1.396 (or more) under load - which isn't required.

With these settings, the max I seem to see is 1.360 in prime, other apps 1.344 or less.

More testing is required - but Prime was stable for 8 hours last night. That's a first

 

Damn, just had my first BSOD since thinking I was stable.

BSOD Page fault in non-paged area, I've put the voltage up by .005 to see if that helps, not sure if that's what I should be doing though!

 

What exact error code is it?

 

Isn't Page Fault In Non Page Area usually memory issue?

 

Not necessarily.
Low voltage(or too high) on cache or system agent can cause such memory errors.

 

SA is currently on auto, which as per usual is over-zealous so I should rain that in manually sometime.

I added .005 to the turbo voltage, which so far is yet to crash.

I won't say it's stable as everytime I do, it crashes soon after!

 

Check the BSOD code with bluescreenview, will help at pinpointing the issue.