Solved

i5-6600k overclocking stress test worker failing constantly

Hi,
I just built my PC and everything runs smoothly.
I overclocked my i5-6600k to 4.4ghz with a 1.280 voltage (most recent voltage) and have tried to stress test with prime95 and every time my fourth worker fails.
My board: Gigabyte GA-Z170-HD3 LGA1151 ATX
I have tried increasing voltage several times but no luck, I haven't overclocked my RAM just stock at 2400mhz.
My CPU hasn't went over 60℃ on HWmonitor.

The exact error that I get on prime95 is:
--
fatal error: rounding was 0.488303501, expected less than 0.4
hardware failure detected, sonsult stress.txt file.
torture test completed 1 tests in 2 minutes - 1 errors, 0 warnings.
worker stopped.


Really stuck so any help would be amazing!
Thanks,
-Calvin.
9 answers Last reply Best Answer
More about 6600k overclocking stress test worker failing constantly
  1. What heatsink and thermal paste are you using, if its oem stock heatsink with preapplied grease that won't work.
    At best you could probably do 4.0ghz with the stock cooler, other than that it won't work.
    Underclock your cpu 3.9ghz and then run a stress test, if that works fine with no issues.
    Try 4.2ghz when you hit that stump you will know what it can and cannot do in its current state.

    https://communities.intel.com/thread/98882

    http://www.overclockers.com/forums/showthread.php/762912-i5-6600K-i7-6700K-OC-results/page3

    Ask overclockers for more direct help on this if it still persist to be an issue call intel for rma help.

    Phone Number: 1-916-377-7000
    Business Hours:
    Monday - Friday
    7:00 AM to 5:00 PM (Pacific Time)
  2. Best answer
    You might have a look here to see if this is the type of issue you're having.
    http://www.pcworld.com/article/3021023/hardware/how-to-test-your-pc-for-the-skylake-bug.html

    Usually a bios update will fix it via a microcode update.

    Ideally you should be running p95 v26.6 small fft's for thermal testing.

    (I've moved this thread to the overclocking forum for you)
  3. maikutech said:
    What heatsink and thermal paste are you using, if its oem stock heatsink with preapplied grease that won't work.
    At best you could probably do 4.0ghz with the stock cooler, other than that it won't work.
    Underclock your cpu 3.9ghz and then run a stress test, if that works fine with no issues.
    Try 4.2ghz when you hit that stump you will know what it can and cannot do in its current state.

    https://communities.intel.com/thread/98882

    http://www.overclockers.com/forums/showthread.php/762912-i5-6600K-i7-6700K-OC-results/page3

    Ask overclockers for more direct help on this if it still persist to be an issue call intel for rma help.

    Phone Number: 1-916-377-7000
    Business Hours:
    Monday - Friday
    7:00 AM to 5:00 PM (Pacific Time)


    I forgot to mention that I'm using the hyper 212 X, stock thermal paste.
  4. I tried the BurnInTest several times and everything was fine so I'm really not sure why its only happening for blend tests.
    So far small fft's tests are working fine.
  5. I use to use this on my lga core i3 540 and my old Phenom 2 X6 1055T.
    http://www.newegg.com/Product/Product.aspx?Item=N82E16835154012&cm_re=thermal_paste-_-35-154-012-_-Product

    When over clocking to the max, I would start out with that first.
    Then look into good 120mm or 140mm fans to keep the case chilled out and have the air push out of the case.

    Go to the blends website and see if they have a forum or ask them by email support.
    It might be a software version problem or it may not be, if every other test work good it may be that problem.
    About the cpu core clock speed, if your not gaming daily or using high end settings.
    Set the default speed to 3.2ghz, then you set the max over clock speed to 4.2ghz.
    That way if one of the games or programs you use needs more horse power to do any task it'll kick it up and then let it drop back down when its done.

    Kind of like amd's overclocking but it'll help out in the long run with windows 10.
  6. So long as speedstep is left intact no special settings should have to be done to allow the cpu to idle down when not under heavy load. It might be labeled EIST (enhanced intel speedstep technology) in the bios. The cooler doesn't seem to be the issue if temps aren't exceeding 60c. Nor will changing thermal paste.

    There are issues once in awhile with skylake cpu's and prime numbers (p95). That's why bios updates were issued with microcode to correct it. I'm wondering if this isn't what's causing the issue.

    http://arstechnica.com/gadgets/2016/01/intel-skylake-bug-causes-pcs-to-freeze-during-complex-workloads/
  7. synphul said:


    It is the issue, if his bios is not version f4 he will need to update it.
    If it is f4 or recent, he will need to contact the board manufacturer then intel for rma help.
    To narrow down the issue.
    http://www.gigabyte.com/products/product-page.aspx?pid=5495#bios
  8. This is why people seem to stay away from prime95. It's way overkill and puts such an unrealistic load on your CPU, which also gives you unrealistic temps, but hey if you're trying to make the most stable OC you possibly can then I suppose prime95 is the way to go, just not necessary. My 6600K actually does the same thing you're referring to, but it works perfectly fine in other stuff like AIDA64 Extreme, Cinebench, 3DMark Fire Strike, games, etc.
  9. Right, the 'skylake bug' regarding prime95 was an issue that only comes up in specific instances of trying to solve prime numbers. It won't affect the vast majority of users in real world use. There is a fix for it though via microcode.

    The idea behind using p95, ibt or others that put the cpu through extreme scenarios is to cover a 'worst case' scenario. If it can handle that and real world use doesn't push the cpu quite as hard then you know it will handle it. Prepare for the worst and hope for the best.

    For overclocking it's extremely important to stress test since you're venturing into the unknown realm. Not everything like cpu temps and clock speed are linear and once you're venturing into the unknown/untested by the oem's you need to do some legwork to re-verify everything is still ok. Imagine you overclock but haven't stress tested. In every game you've played so far your temps are fine, running at 65c. You get accustomed to it and aren't monitoring your temps all the time, you go to play a new game that pushes the cpu harder and without knowing it you're running at 85-90c.

    Unless you pushed your cpu hard enough to begin with you'd never know it could reach those temps with the current settings, ambient temp and cooling solutions. It's not that people need to stay away from p95, like anything else they need to understand what they're doing with it and why. Nothing I suppose is 'necessary' though when a worst case scenario comes along, some people will be prepared for it and unaffected by it, others may be in for a surprise.

    As a thermal stress test p95 works well. As a stability stress test there are better options or groups of options. A cpu may show to be fine under p95 for an hour and fail ibt within less than 5min. It's also possible to pass p95 and have a bsod during one of the tests using asus rog realbench which is a more system wide stress test that incorporates the ram, gpu and subsystems.
Ask a new question

Read More

Overclocking Prime95 Stress Test Intel i5 CPUs