Overclocking Frustrations

Nexillus

[H]ard|Gawd
Joined
Nov 6, 2012
Messages
1,116
Hello everyone!

This has been an on going frustration for the last week. I have been trying to overclock my 3930k (rest of build in signature).

I have read several guides and making sure I have as much informaiton before attempting the overclock.

Here's what I am trying to do. Been trying to get a 4.5GHz OC on the CPU, however have run into many frustrations and tried everything I know about OCing, and looked up on the internet on recommendations to get a stable OC using offset.

The problem is not at load, I have successfully tested Prime95 up to 8 hours at 1.35v-1.40v and been stable. It is at idle is when the problem arises, I'm not getting any type of 101 BSOD, just the OS will lock-up or freeze within ~5-10 mins of loading windows if i do not put a stressor on the CPU. I thought this maybe a vcore problem from what I have read. I have adjusted the offset vcore to go as high as .880v on idle and still receiving a lockup/freeze loading windows within ~5 mins, when I did this my vcore at load was the 1.40v, which I didn't feel safe having that high as I want this to be my 24/7 OC.

Here are my current OC on my MB.
VTT/VCCSA 1.15v
Offset: -0.10 idle .832v load 1.344-1.352
CPU LLC HIGH
CPU CAPABILITY 130%
VCCSA LLC High
CPU Power phase control Optimized
C-States all Enabled

I have tried an offset between -0.020 to +0.020, changed the VCCSA LLC and CPU LLC ranging from regular-extreme.

I have ran MEMTEST86+ just incase it is a RAM issue for ~10 hours with 4 passes and no errors.

I have tested the PSU to see if any fault or not ample load is being sent and nothing came back out of the ordinary.

I am currently at my wits end with trying to OC from everyone saying it is so easy :(

I would greatly appreciate any help ahead of time to possibly getting the OC stable, until then I'm running it sadly at its stock settings.
 
Last edited:
From my understanding C1E has to be enabled for my offset to function, are you referring to C3, C6? Which should only effect sleep "mode" ?
 
Turn off offset and try with fixed voltage.


Booted perfect and sat at idle for ~15 mins @ 1.35v (Fixed voltage) and no lock up, every time in the past it would lock up and have to do a hard reset. I dont see a point to test load, as that wasn't the problem to begin with. :/ Just need to figure what is causing the problem when using offset now.
 
Wait. So you are only having an issue when you idle? Does the mobo have some kind of power save feature?
 
Yes, correct only having an issue at idle. The only thing I'm aware of is EPU Power save mode on my board, which is disabled by default.
 
For some reason the MBO goes too low on voltage during idle but quite high on load. Does the CPU clock change to minimum during idle? What does it go down to?

Change all of the following setting you have:

Phase Control - extreme
Duty Control - extreme
VRM frequency - 350
Disable all spread spectrum options

Also, try with LLC off and a higher offset.

If it still doesn't work, try a different BIOS, both older and newer.
 
Last edited:
I tried some of the settings above to no avail :( going to go re-flash the bios. Will be back with more information!

Also it goes down to 1200MHz.
 
You're not getting enough voltage when the CPU is at idle. Every overclock I've done recently has bailed not at load but at idle when using C1E or C states. You can still use C1E with fixed voltage and your CPU will idle at 1600MHz or so. Fixed vs offset will cost you about 5 watts more power at idle, or $5 a year. Sounds like you're using an ASUS board, ASUS doesn't use C-states anywhere in the BIOS, I believe it's enabled when you set the BIOS to allow the OS to idle cores individually. This thread has a lot of information on OCing an ASUS 2011 board, stuff that's probably synonymous with other boards.

Before you go nuclear and settle on fixed voltage do like Meeho said and turn off LLC and increase offset. This will boost idle voltage.
 
You're not getting enough voltage when the CPU is at idle. Every overclock I've done recently has bailed not at load but at idle when using C1E or C states. You can still use C1E with fixed voltage and your CPU will idle at 1600MHz or so. Fixed vs offset will cost you about 5 watts more power at idle, or $5 a year. Sounds like you're using an ASUS board, ASUS doesn't use C-states anywhere in the BIOS, I believe it's enabled when you set the BIOS to allow the OS to idle cores individually. This thread has a lot of information on OCing an ASUS 2011 board, stuff that's probably synonymous with other boards.

Before you go nuclear and settle on fixed voltage do like Meeho said and turn off LLC and increase offset. This will boost idle voltage.


Currently looking into several things on that other thread, and when I get home going to try a few suggestions. Will give some feedback later today! :D
 
Hey all,

Here’s the most recent update. I looked over the information that many of you have helped me with to include looking at a few other forums. I gave everything a shot as I couldn’t get it to be stable, I did an offset ranging from -0.05 to +0.050 to no avail, changing my LLC to also match the offset to prevent high Vcore under load, however every time I would get to desktop it would lockup within 5-10 mins. I spent ~4-5 hours yesterday trying to get info on why this is happening, and about to give up after exhausting almost everything I could think of or suggested when I got an idea to just change the multiplier to 46 instead of 45 and restarted leaving everything else the same. Booted (always booted above -0.005) then got into windows. 10 mins, still idling. (now I was getting a little excited) Then I was like “alright, now a small stress test” did a LINX test for 30 mins, 25% memory, 20 passes. Passed! Then I shut it down, launched firefox to see the vcore change and frequency, no lock up yet. Was concerned with a fluke so restarted to check settings and save them to a profile. I then gave it the idle test to leave it on overnight, woke up and it was still on! :D

My theory is my chip does not like the x45 multiplier and it was locking up with a twofold problem, one not enough vcore at idle and secondly with the x45 multiplier during different VID points when changing frequency it would not get enough vcore for that frequency, simple change in multiplier changed he VID-frequnecy-vcore ratio.

Now to see if it works at prime95 8 hours!
Idle: .898v
Load: (only LINX so far) 1.388v @ 4.6GHZ
Thanks everyone for your help!
 
Interesting. Yeah, seems VIDs aren't quite right at 45x. Quirks of overclocking :)
 
Back
Top