1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Overclocking Need help figuring out my partial Instability

Discussion in 'Tech Support' started by 7hr08ik, 18 Jun 2012.

  1. 7hr08ik

    7hr08ik Kill all Humans!

    Joined:
    13 Mar 2010
    Posts:
    53
    Likes Received:
    0
    Hey guys,

    I'm having a rather interesting issue. I've OC`d my system and I`m getting weird instability.
    At first I OC`d to 4.6GHz and the PC ran prime95 64bit on small FFT for 4 hours with no issues. The next day started up the PC and it froze on windows boot. So i restarted, same problem. I couldn`t be bothered trying to faff about with voltages so i restored CMos and left it for a day.

    Now i have my system OC`d to 4.4GHz and im getting the same issue. It ran prime95 64bit for 6 hours small FFT`s no issues at all. Followed by a good few hours of skyrim, without any problems.

    I went to bed, woke up the next day, and BANG! Froze on windows boot again. But once i restart, after ignoring the annoying windows recovery crud. It boots with no problem at all. Even runs primne95 again, without issue.

    The only thing i can think of is that the mobo cools down overnight, and a fresh boot causes some instability in the VRMs? So i had a play with the mobos VRM settings but nothing i can find or think of can fix it. Ive already put more voltage than was required for the OC, but not helping.

    My system is as follows-

    • i5 2500k (step 7 revision D2)
    • Asus Maximus IV Extreme-Z (Bios 3304 - Latest as far as i can tell)
    • Corsair Vengeance LP (8GB)
    • Dark Rock Pro
    • Asus Geforce 275
    • Crucial M4 SSD 64GB (Set in Raid as Cache for main HDD)
    • Samsung Spinpoint F1 1TB
    • Enermax Galaxy 1000W


    My Bois settings for OC-
    • Blck - 100
    • Turbo Ratio - 44
    • Ram Speed - 1600
    • CPU Volt - 1.320
    • DRam Volt - 1.5
    • Spread Spectrum - OFF
    • iGPU - OFF
    • Intel Virtualization - ON
    • Sata - RAID

    Digi+ VRM Settings-
    • VCore PWM - Extreme
    • Load Line - 75%
    • Phase COntrol - Extreme
    • VCore Switching Freq - 500
    • Overcurrent Protection - 150%


    All other settings are on Auto

    My GPU is OC`d a little bit, but not changing any voltages its upto 670MHz from 633Mhz.
    I`ve racked my brain and i can`t think of anything that could be causing this. I know that ASUS boards, are slightly different with overclocking i5/i7`s in that they still gear back the CPU, (for instance mine is currently running at 1.6 while i type this) TBH though, i like that, it means the PC isn`t running overboard just to check my e-mail. And saves me some electric. But if that was the reason for the freeze, then surely it would happen every time i boot the PC, not just when its been off for more than a few hours? Maybe i need to OC the Blck a smidge?

    Any help would be greatly apprecieated
     
  2. bulldogjeff

    bulldogjeff The modding head is firmly back on.

    Joined:
    2 Mar 2010
    Posts:
    8,403
    Likes Received:
    634
    I've had a similar problem, on a totally different set up so it will interesting to see whats said. The only thing I could pin mine down to was a possible ram fault.
     
  3. 7hr08ik

    7hr08ik Kill all Humans!

    Joined:
    13 Mar 2010
    Posts:
    53
    Likes Received:
    0
    I didn`t think of ram.

    But if it was it would happen randomly and during use. Not at specific time. Its bloody annoying, it happens in the same way that it does when OCing. When you OC to far, you get the windows logo, then it freezes. but once ive rebooted, not a problem. Been trying to figure it out for a few weeks now. The only thing i can come up with is either the VRMs are too cool to handle the extra voltage and heat. Or i need to knock up the blck a bit to compensate. But im a bit nervous about tweaking that
     
  4. noizdaemon666

    noizdaemon666 I'm Od, Therefore I Pwn

    Joined:
    15 Jun 2010
    Posts:
    6,090
    Likes Received:
    794
    If it's an instability somewhere in the overclock then Intel Burn Test set to 15 runs on maximum stress setting should find it. Well, not find, but fail. P95 small FFT is good at dishing out heat but not brilliant at stressing the entire processor. Whereas IBT is. I'd give that a go first.
     
  5. 7hr08ik

    7hr08ik Kill all Humans!

    Joined:
    13 Mar 2010
    Posts:
    53
    Likes Received:
    0
    I`ve never actually used the intel burn test. always used prime95.

    I`ll get hold of it, give it a go and report back
     
  6. 7hr08ik

    7hr08ik Kill all Humans!

    Joined:
    13 Mar 2010
    Posts:
    53
    Likes Received:
    0
    FAIL!!!

    Well i just tried IBT. Maximum stress, 15 runs, and it failed.

    Left it running while i went for a shower......came back and the pc had rebooted.

    ----------------------------
    IntelBurnTest v2.53
    Created by AgentGOD
    ----------------------------

    Processor: Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz
    Clock Speed: 3.30 GHz
    Active Physical Cores: 4
    Total System Memory: 8159 MB

    Stress Level: Maximum (5000 MB)
    Testing started on 19/06/2012 10:51:49
    Time (s) Speed (GFlops) Result
    [10:53:57] 92.481 113.2041 4.306579e-002
    [10:55:52] 91.738 114.1209 4.306579e-002
    [10:57:47] 91.749 114.1072 4.306579e-002


    So what the hell is it? Im already pushing 1.32V through the CPU, everythingelse is set on auto. Im guessing its either, blck that needs tweaking or maybe the CPU PLL needs a bit more through it? But ive already tried tweaking the PLL and i cant find a stable voltage.....all the way upto 1.9!!!
     
  7. noizdaemon666

    noizdaemon666 I'm Od, Therefore I Pwn

    Joined:
    15 Jun 2010
    Posts:
    6,090
    Likes Received:
    794
    Ok, it might be a memory controller issue as IBT stresses it more than P95 small FFT. Go here: http://forums.bit-tech.net/showpost.php?p=2994378&postcount=18. Do a custom test in P95 using those values (min/max at 1344 for 25 minutes and then min/max to 1792 for 25 minutes) and run it. If either test fails and gives x124 / xD1 / x9C or x0A errors, it's a memory issue.

    At that point into BIOS and up the VTT (VCCIO it might be called) to 1.2V. Then re-run IBT.

    Edit: Or just up the VTT/VCCIO and re-run IBT depending how impatient you are lol
     
    Last edited: 19 Jun 2012
  8. Jester_612

    Jester_612 "Jammy..."

    Joined:
    14 Nov 2011
    Posts:
    1,139
    Likes Received:
    30
    Increasing VCCIO or QPI are required, something like 1.12v. The running at 1.6Ghz is the CPU power state C2/C3.
     
  9. kenco_uk

    kenco_uk I unsuccessfully then tried again

    Joined:
    28 Nov 2003
    Posts:
    10,104
    Likes Received:
    682
    Possibly a little more vcore or LLC needed, if it runs fine all day until it's stressed with IBT. Assuming your temps are okay, try vcore at 1.35 and if it gets through a quick run of IBT fine, lower it a little and check again. Keep going until it bsods, then raise it back up one notch and test.

    Atm, I have my 2500k running at slightly lower vcore and it hits a wall at x43. I'm pushing for a little bit of efficiency on mine and whilst it hits 1.28 while running IBT, it idles at 0.9. Mine seems to need 1.35vcore to get to x45 and then up to 1.45+ to get to x50 - it gets rather toasty though, hitting 90c+!
     
  10. noizdaemon666

    noizdaemon666 I'm Od, Therefore I Pwn

    Joined:
    15 Jun 2010
    Posts:
    6,090
    Likes Received:
    794
    I'd say that given it passes a standard P95 small FFT run, vcore should be fine. It doesn't seem to be the speed at which the cores are running which is the issue (I know vcore doesn't just govern this), more the other stuff now on the CPU which need a bit more to be fully stable. I'd only up the vcore if VTT/VCCIO make no difference.
     
  11. 7hr08ik

    7hr08ik Kill all Humans!

    Joined:
    13 Mar 2010
    Posts:
    53
    Likes Received:
    0
    Cheers for the replies.

    I agree that it can`t be VCore issue, because i had it running prime95 stable at lower vcore. I just initially tried upping that to see if it was the issue. Just couldnt be bothered to lower it back down.

    Plus according to the geeks over at Asus ROG (Here) They say i should be able to get 4.4GHz with a vcore of 1.25?!?! Leaving everything else at auto. Im not to good with the Digi+ settings but i think i got them how i should do. To improve stability and keeo the temps low. Im going to have to have a full day of playing around with those settings i think.


    Seems deffinatley worth checking, i know that im running a deecent CPU with an awesome mobo, but the only place i skimpt a bit was with the ram. Its only 1600 at 1.5v but the mobo can take (and i would preffer 1800 1.6v) I`ll be giving this test a check next.

    I know this sounds like a tard response but there is no seperate VCCIO volt choice. Only VCCSA/IO combination. For some reason ROG decided to NOT give me all the individual choices available.....im sure theres a point to it. But im not that much of a geek to understand why.
    [​IMG]
    p.s NOT MY PICTURE

    My temps arnt to bad either.....In prime they max out at about 65-70.....un IBT they got to 75 but thats not much of an issue...they never get that high...even after hours of gaming. I do need to lap the CPU though....10`C diffs between some of the cores
     
  12. noizdaemon666

    noizdaemon666 I'm Od, Therefore I Pwn

    Joined:
    15 Jun 2010
    Posts:
    6,090
    Likes Received:
    794
    Hmmmm I didn't know boards combined VCCSA and VCCIO, they require different voltages so I'm not sure how that works.

    Anyway, let us know how the customer P95 run goes. I'd treat VCCSA/IO as VCCIO just for now if you need to up it. Note that 1.2V is probably the max you will ever need. It's likely you'd be able to lower it down to 1.12V once you're stable.
     
  13. 7hr08ik

    7hr08ik Kill all Humans!

    Joined:
    13 Mar 2010
    Posts:
    53
    Likes Received:
    0
    Will do, im going to run a full memtest first, then run those prime95 tests.

    I`ll report back with what i can find. Hopefully we can figure this out, and i might be able to push it further :rock:
     
  14. kenco_uk

    kenco_uk I unsuccessfully then tried again

    Joined:
    28 Nov 2003
    Posts:
    10,104
    Likes Received:
    682
    Aye, post results. I'd be interested to see what comes of tinkering with other voltages than vcore to get more out of a cpu. I must admit, I just do things the quick and dirty way although with the MSI Z77A-GD65 board I have now, I got to tinkering with the intel speedboost tech and use that for overclocking rather than the usual way of setting the ratio.
     
  15. 7hr08ik

    7hr08ik Kill all Humans!

    Joined:
    13 Mar 2010
    Posts:
    53
    Likes Received:
    0
    well, as a thought i checked and found new bios....and guess what asus have done....Now its called VCCIO.....ONLY!!! No combined VCCSA/IO.....but now the F***ING board wont take my previous settings....tried VCCio at 1.2, even bumbed up the vcore to 1.35 but nothing.....thinkimjust going to have to start again from scratch...but im still going to run those RAM tests...

    ROG recons i can get 4.4 without touching anything but vcore....i had to have my vcore .05 more than they suggested and still didnt have stability. So ive got an issue somewhere. Still gonna keep you upto date with it though...Im not too good with OC`ing the i5...i went from a Q6600 to this...so im learning all over again. Love it...

    Just to check....
    Bump ratio > test > bump up vcore > test > then VCCIO if not stable > test > goto 10

    is that about right? possibly tweak PLL voltage at higher speeds?
     
  16. noizdaemon666

    noizdaemon666 I'm Od, Therefore I Pwn

    Joined:
    15 Jun 2010
    Posts:
    6,090
    Likes Received:
    794
    That method would work but IMO is quite clumsy.

    What I'd do is this: Set vcore to 1.4V, PLL to 1.88V, VCCIO to 1.2V, PCH to 1.06V and DRAM to 1.5V. Then set memory to 1600 and cpu multi to 45. Run IBT on max.

    1. If it passes lower vcore by a couple of steps. Re-run IBT. If it passes, repeat this line until unstable, then go back up a couple of notches. Then do the same for VCCIO, PLL and PCH.

    OR

    2. If it passes up the cpu multi by one. Re-run IBT. If it passes either go through 1. or go through 2. again until you're happy.

    Note: definitely run through 1. at some point as those voltages are nearing the limit of what some would consider safe for every day use.
     
  17. 7hr08ik

    7hr08ik Kill all Humans!

    Joined:
    13 Mar 2010
    Posts:
    53
    Likes Received:
    0

    Sounds like a plan.....Just finished memtest 2 passes all fine.
    Got the bios at factory settings now, apart from memory is on XMP settings. gonna run that prime95 memory test then crack on with the OC.
     
  18. noizdaemon666

    noizdaemon666 I'm Od, Therefore I Pwn

    Joined:
    15 Jun 2010
    Posts:
    6,090
    Likes Received:
    794
    If you're at stock there's no need to run the P95 test. That was only a suggestion to help stabilise the overclock. If you're starting again, no need to run it beforehand :)
     
  19. 7hr08ik

    7hr08ik Kill all Humans!

    Joined:
    13 Mar 2010
    Posts:
    53
    Likes Received:
    0
    LOL!!!

    Its frozen!!! Hahahaha

    My cpu isnt OC`d, but the RAM currently is. Its rated as 1333 but has its built in XMP profile that boosts it to 1600.

    Been running the Custom RAM test for 3 mins, and the PC has frozen....LOL
    Turns out corsair can`t supply a factory overclock that works

    Got a couple o little settings i can tweak to test it, but if not, i`ll be fine just leaving the ram at stock for now. I need to invest some decent ram anyway.
     
  20. noizdaemon666

    noizdaemon666 I'm Od, Therefore I Pwn

    Joined:
    15 Jun 2010
    Posts:
    6,090
    Likes Received:
    794
    What corsair RAM is it? Generally Corsair stuff is good. I'd check rated timings against actual timings in BIOS though, something may have gone wrong there.
     

Share This Page