New Issues with Previously-Stable 2500k

Discussion in 'Overclocking & Cooling' started by dclapps, May 4, 2019.

  1. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    Hey all,

    I've had my trusty 2500k running at 4.5ghz stable at a safe voltage for roughly 5 years. Recently, I have started to get hard crashes sporadically; no BSOD, locks up and restarts, sometimes takes a few boots to get back to BIOS and to Windows. System setup:
    • Gigabyte z77x-ud5h (seven years old)
    • 16gb (4x4gb) Corsair XMS, all matching (one pair at seven years old, second pair maybe four years old)
    • Seasonic x650 (three years old)
    • GTX960 (three years old)
    After the errors started happening, I tried upping the voltage a bit, but that didn't help. Then, I reset everything to stock (no overclock, even went and loaded optimized defaults at bios) but the problems still persist. I have tried:
    • reinstalling windows 10, twice
    • ran memtest on ram, passed a four hour test
    • flashing new bios firmware
    • cleaned case, no heat issues, all case fans run
    • ran prime95 for a few hours with no errors
    With everything stock, ran prime95 and furmark together for an hour, no issues, no errors. Stopped, went and researched this more on the web, crashed while scrolling a forum. I cannot reproduce this issue, and I haven't solved it by going back to stock. With no BSOD, I am assuming it is a hardware issue. Could it be a faulty motherboard? Not sure how to test/verify that.

    Any help is greatly appreciated, I am struggling with this one.
     
  2. CAD4466HK

    CAD4466HK [H]ard|Gawd

    Messages:
    1,222
    Joined:
    Jul 24, 2008
    Hard crashes, lockup's and reboots...could be a PSU issue. Do you have another one to swap in?
     
    lostin3d and SticKx911 like this.
  3. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    I don't, I'd have to buy a new one. I will say that the current one was purchased 2-3 years ago because the previous PSU fried in there (it was a 10 year old Seasonic). When I swapped this new one in, everything went fine for three years.
     
  4. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    If you've done a clean installation of Windows, not a plain re-install, I would start messing with hardware. Start pulling RAM modules and seeing if you have the same issues. You can also relax the RAM timings or up the memory voltage. Corsair RAM typically needs to be run a little higher than normal. Instead of 1.35v as an example, you can run 1.365 or 1.37. I'd be leaning away from GPU in this instance, and you've done a pretty good job checking that buy running Furmark. Processor's rarely fail, and when they do they normally go out in a more spectacular fashion. But you could also try down clocking it. I've seen overclocked CPU's degrade over time to a point where they won't run at stock speeds any more. Put the voltage to stock and downclock it a couple hundred megahertz and retest. Also, check the voltage readings in the BIOS. If they are more than 10% either way (above 5v or below as an example), then your PSU is out of spec and is likely a culprit.
     
  5. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    When a power supply goes, they are typically responsible for random reboots or a hard lock (not usually a BSOD type crash) but only under load. If it didn't happen with Furmark and Prime95 running, I doubt the PSU is faulty.
     
  6. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    Thank you for the comments Dan, and others. I've had the RAM running at 1.65v per the MFR, so I do change that when the BIOS resets or I load optimized defaults. I also set it to XMP, but have tried without that and still random crashes.

    Thinking back to when I install Windows, I boot to USB and load from there. I assume that is a clean install, as it formats the hard drive, but maybe Windows 10 is handling things differently?

    Edit: I will check the voltages in BIOS as well.

    Edit2: Just checked BIOS, see 3.324 on 3V leg and 12.168 on 12V leg. Apparently it doesn't show the 5V leg. Ram is at 1.656 which I set, non-XMP. I believe that the non-XMP setting down clocks the ram to 1066 instead of default 1600MHz.
     
    Last edited: May 4, 2019
  7. SticKx911

    SticKx911 2[H]4U

    Messages:
    2,125
    Joined:
    Mar 14, 2004
    I recently had an issue with a gtx 560 not liking any new nvidia driver. It took forever to troubleshoot. Anything past a certain release would cause bsods. I only found that after one update finally broke it completely. (Windows no longer booting to the nvidia driver).

    I only mention because maybe one of the newer drivers installed is part of the problem. Did you do any updates before having the problems? Try rolling back the nvidia driver. (Your 960 is considerably newer than the one I was using, so probably not, but just something to think about.)
     
  8. CAD4466HK

    CAD4466HK [H]ard|Gawd

    Messages:
    1,222
    Joined:
    Jul 24, 2008
    I've seen plenty of PSU's that were going bad, and cause issues like his while being idle, and run fine @ 100% load. That's why I said it COULD be the PSU.
    Pulling the RAM as you suggested is also a good diagnostic route to take when dealing with these types of symptoms.
     
  9. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    Just crashed again. Maybe 5 minutes up time on this thread, then loaded a game via steam and crashed. I've been back on this thread writing this response for a bit and no crash, so I am going to to see if I can replicate the crash via loading the game.

    Your comment reminded me that I forgot to add: during one crash, the monitors dropped out as if the signal was gone, but the audio played through fine, no skips/lags, and was not looping. I thought that was weird, so I let it stream blindly for a few minutes then one monitor came up and showed a garbled, 8-bit-like image. Was just random colors and lines, I had to reboot manually. That has only happened once in the 100+ crashes, but happened recently.

    I am running 419.67 drivers from Nvidia, which was their most recent at the time that I installed Windows. I see they have a new set, but read some iffy stuff so I haven't updated. I assumed I would see some BSOD if it were the GPU, but since I am grasping at straws I will try everything.

    Also, seven years is a good run, I'm starting to spend troubleshooting time on researching Zen2. Would still be nice to figure this one out. Thanks again for the time and continued support.

    Edit: Haven't been able to replicate the crash via the game, in fact there hasn't been a crash yet in 20 minutes or so. This has been the toughest part, replicating it.
     
    Last edited: May 4, 2019
  10. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    Had sporadic crashes yesterday. Noticed that there was an intermittent issue coming out of sleep where the monitors would not turn on, so I disabled C1 / sleep settings in BIOS and am so far 10+ hours of uptime with no crash yet.
     
  11. rastaban

    rastaban Gawd

    Messages:
    818
    Joined:
    Jul 30, 2011
    My current board, a Gigabyte Z170 UD5 TH also had sporadic crashes for a good 6 months before a BIOS update fixed it...it was also ridiculously tough to troubleshoot, I swapped almost everything (even RMAed the CPU) and finally came to the conclusion that the C-states set in BIOS was causing the crashes/lockups. It's not the same gen board as yours but you can try disabling all the C-state settings and see what happens.
     
    Dan_D likes this.
  12. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    C-states and various power savings features can cause issues.
     
  13. TheFlayedMan

    TheFlayedMan Limp Gawd

    Messages:
    331
    Joined:
    May 29, 2015
    Have you tried replacing the CMOS battery?
     
    pendragon1 likes this.
  14. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    I have not tried replacing the CMOS battery. I have reset it in the past, but by replace do you mean install a new battery? I could try that.

    The C-states setting has been changed, as well as any option in the BIOS that states it modifies voltage / changes clock speed for power reasons. I've restarted a few times (without crashes) for Windows settings or a BIOS check, so I will continue to track the up time. It appears the changing of the C-states has made the difference so far.
     
  15. fullvietFX

    fullvietFX [H]ard|Gawd

    Messages:
    1,785
    Joined:
    Sep 1, 2004
    I went through almost the exact same thing as you with my 2600K. 4.8GHz for years and then all of sudden random restarts and crashes. The culprit was my PSU. However it seems you've changed yours recently. I still wouldn't rule out the PSU though.
     
  16. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    Interesting. This PSU is 2-3 years old, I could look into ways to verify voltage outside of the BIOS readings.

    I assumed that PSU issues would show up when I was stressing the CPU and GPU at the same time, but that obviously could be a wrong assumption. Did you test that way as well?
     
  17. fullvietFX

    fullvietFX [H]ard|Gawd

    Messages:
    1,785
    Joined:
    Sep 1, 2004
    No I never tested anything to be honest. I had a gut feeling it was the PSU and I changed it. Zero issues now.
     
  18. somebrains

    somebrains Gawd

    Messages:
    590
    Joined:
    Nov 10, 2013
    I had the same issue when I installed a 1080ti.
    That Seasonic 760 psu is now in my buddy's case powering his Vega 64 totally fine after a 650bronze exhibited the same behavior.

    Issues happened slowly over time, not immediately.

    Go figure.
     
  19. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    Right now I am on about 4 hours of up-time, no crashes. When I wrote this thread the other day, before changing the C-states, I was crashing at least once per hour or so. Sometimes I would get 3-4 crashes in five minutes, but I was never able to sync the crashes with any specific behavior.

    Is there a generally understood way to test a PSU or is it common to just replace if you think it's faulty?
     
  20. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    Just about hit 4 hours and crashed. This time, the computer shut itself entirely, booted for a second, then shut off, then booted again to Windows. I am thinking this is PSU. SeaSonic has a warranty on this still, but I guess I am out during shipping time if they even warranty this.
     
  21. rastaban

    rastaban Gawd

    Messages:
    818
    Joined:
    Jul 30, 2011
    Keep us posted if you do RMA the PSU. I'm curious what the CS is like, and what criteria they use to authorize an RMA.
     
  22. somebrains

    somebrains Gawd

    Messages:
    590
    Joined:
    Nov 10, 2013
    I run the PSU on someone else's PCthat is known good to see if the same problems follow.

    Is there no store local to you that you can buy a PSU, then return if problems persist?

    I'm having issues with a game right now, I know it isn't my hardware.

    Another buddy plays pubg and we had to re-Tim his GPU bc it started artifacting when he kicked it off. I install my gpu in his box so it helped narrow probs down

    I rely on my friends when things get funny.
     
  23. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    I actually just spent some time with Seasonic's CS Rep via chat, who was pretty helpful. Seasonic will accept the PSU (shipping is on me), test it and will either send back if working or send a replacement if not. So for $10 bucks (shipping), I could rule the PSU out.

    My buddy's system is down for the count so I don't have the option of borrowing or using someone's rig. I would have to buy a PSU or deal with the downtime while I ship to Seasonic. Or, I could go to Best Buy and buy/return but I typically avoid doing that.

    The Seasonic rep left me with checking all cables for proper connection, some potential issue with PWDIS / hdd (??), or RMA.
     
  24. Rvenger

    Rvenger [H]ard|Gawd

    Messages:
    1,653
    Joined:
    Sep 12, 2012
    Try one more thing before calling quits. lower the RAM speed and voltage to 1.5v. The sandybridge IMC cannot sustain 1.65v without degradation. Yes it could be the PSU but if it is not, I am betting the IMC on the CPU is toast from running 1.65v on the RAM.

    I believe the intel max safe spec is 1.57v
     
  25. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    I could drop the RAM voltage down, what type of problems would occur from running it at 1.5v from the RAMs point or view? Should I try to lower the speed at the same time to accommodate the lowered voltage?

    It's also 4 sticks at 4gb per, so that may mean 1.65v tomes 4, although I'm not informed entirely about the power interactions at that level.
     
  26. somebrains

    somebrains Gawd

    Messages:
    590
    Joined:
    Nov 10, 2013
    Buddy of mine is a video editor.
    He has a Fractal R5 fully populated with spinning rust.
    He has been having heat issues bc it's starting to be a hot Spring.

    Look what I found in his case, aio was same.
    He had all intakes otherwise, which probably helped.

    Either way check everything over again at stock clocks.

    IMG_20190507_182249.jpg
     
  27. fullvietFX

    fullvietFX [H]ard|Gawd

    Messages:
    1,785
    Joined:
    Sep 1, 2004
    Were the fans spinning in opposite directions?
     
  28. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    I'm on my phone so it's tough to see details in that photo. Were cords clipped by a fan?

    Right now I am at 18 hours of uptime, haven't made any changes. I cannot figure out what causes the crashes and why they seem to be grouped together at times then two weeks of nothing then ten crashes in an hour.
     
  29. Furious_Styles

    Furious_Styles [H]ard|Gawd

    Messages:
    1,295
    Joined:
    Jan 16, 2013
    If you have no spares you should send the PSU in to rule it out.
     
  30. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    Roughly two week later update: I went a week or so after this post with no changes and not a single crash. Then, a few crashes in a row a few days ago. I noticed on some restarts that it would hang at BIOS, so I changed the switch on the BIOS to load to the alternate / back-up BIOS. Still crashed a few minutes later as I gave up again.

    I did just go read through the entire manual again, and noticed that it states it supports up to 4 sticks of DDR3 ram at 1.5v. The Corsair XMS RAM that I have is rated 1.65v, and I have 4 modules installed. While it passed hours of memtest, I'm wondering if the issue is present here somewhere. The comment above from Rvenger identified this as a potential issue, I'm wondering what is the drawback to running at 1.5v ?

    I have 4x4GB Corsair PC3-10700. XMP-1600 runs at 1.65v / 800 MHz, should I drop to 1.5v / 666 MHz ?
     
  31. Brian_B

    Brian_B 2[H]4U

    Messages:
    2,755
    Joined:
    Mar 23, 2012
    Overclocks don’t last forever. Time to re-evaluate your OC I think.

    1.65v ram is essentially a factory OC on a dimm.
     
    Ryun likes this.
  32. MMitch

    MMitch Gawd

    Messages:
    667
    Joined:
    Nov 29, 2016
    Try removing the GPU and use onboard GPU to see if games still crashes.
     
  33. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    So, if I back the voltage down, do I drop the speed down a setting as well?

    I will try the GPU as well, I've just gamed on/off and took a week or so for first crash, but crash occured at desktop. I also will try another GPU and PSU that I will be getting my hands on soon. I won't change everything out at once, I'm trying to find the problem.
     
  34. twzTechman

    twzTechman Limp Gawd

    Messages:
    224
    Joined:
    Apr 14, 2011
    I have a power supply tester that has been reliable in identifying bad power supplies. $10-30 on Amazon or New Egg. This should give you a quick easy way to check.

    Also - after years of running that OC on the CPU, it probably has a bit of degradation. The previous suggestions to lower the OC could help if it is not too far gone.
     
    Ryun likes this.
  35. SvenBent

    SvenBent 2[H]4U

    Messages:
    2,903
    Joined:
    Sep 13, 2008
    Thermal paste gone bad ?

    How are temps under prime95 load ?
     
  36. MMitch

    MMitch Gawd

    Messages:
    667
    Joined:
    Nov 29, 2016
    Lot of good suggestions here, try to clean heatsink + re-apply thermal paste. Remove OC and bumb default voltages marginally. New CMOS battery should be in the top things too, that thing is antique.
    That should be a good start, then try PSU, no GPU, new GPU...
     
  37. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    Since just prior to the reinstall of Windows a month ago, I have removed the OC. All of these crashes have continued to happen under stock/default settings. The RAM has been at 1.65v throughout though.

    All prime testing and furmark were at stock settings. I can use a multimeter on the Mobo to check voltages, and I will replace the CMOS battery as well.

    I don't remember the temps off the top of my head, will be able to check later. They weren't out of whack because I dropped the OC.
     
  38. Furious_Styles

    Furious_Styles [H]ard|Gawd

    Messages:
    1,295
    Joined:
    Jan 16, 2013
    Have you ruled out the PSU yet?
     
  39. dclapps

    dclapps [H]Lite

    Messages:
    89
    Joined:
    Jan 22, 2009
    I haven't ruled the PSU out yet. I wanted to test voltage at the Mobo where it has testing points, then, depending on that, potentially send in to Seasonic. I've needed the PSU so haven't been able to be without it.

    Is it worth testing at the Mobo, or should I just send it in when I can?
     
  40. Tweak42

    Tweak42 Gawd

    Messages:
    603
    Joined:
    Dec 1, 2010
    My strange experience with previously stable:
    I had to drop my 2500k to 2 cores to get stable at stock and overclocked speeds after becoming unstable running at 4.5ghz for years. Tested under windows and linux. I've since swapped it with a 3570k last year.