Page 3 of 3 FirstFirst 1 2 3
Results 21 to 26 of 26
  1. #21
    ROG Enthusiast Array
    Join Date
    Aug 2014
    Reputation
    46
    Posts
    41

    Quote Originally Posted by lpiotrowski211 View Post
    Have an real pain in a.. with stability in Real Bench. Have 9900K oced to 5ghz. Ram: Gskill 2x8GB 3000MHZ XMP

    Voltage 1.300v manual,Load Line 6 level,Sync cores 50,all power limits maxed.



    So i dont know how to be sure if my system is stable. Sometimes after 15 minutes realbench throw whea 19 cache hierarchy error, in event log.Sometimes it will pass 2 hours. So how i can be sure all is stable then? !!!! Using 16GB ram in stress.



    One time it pass 2 hours fine, i am thinking its fine. Next day after 15 min whea 19.
    I replied to your post on reddit.
    Open the side or front of the case or your house door and drop the ambient temps and see if you pass 2 hours, then get back to me.
    But tl;dr: Increase vcore 15mv.

  2. #22
    ROG Guru: Yellow Belt Array
    Join Date
    Jan 2015
    Reputation
    10
    Posts
    237

    its BSoD.

    0x0000003b

    SYSTEM_SERVICE_EXCEPTION

    Its random.
    Sometimes it pass 2 hours without problem,sometimes it not pass on the same voltage in REALBENCH. Any ideas? Maybe i should bump to 1.310v?


    Temps are fine . 90C max peak. 80-90C


    In UEFI voltage prediction for AVX 5000MHZ is : 1.324V. Maybe listen to this and just enter 1.324V?
    Last edited by lpiotrowski211; 03-05-2019 at 08:56 AM.

  3. #23
    ROG Guru: Yellow Belt Array
    Join Date
    Jan 2015
    Reputation
    10
    Posts
    237

    Ok i found stable voltage 1.310V , but...

    I run REALBENCH with full amount memory 32GB RAM. It passed.But when i clicked run again just before test start ,said" Instability Detected". I check event log and it said something about memory pagefile etc.

    So should i worry about that one fluke? It passed before 2 hours,but next click it crashed with this.

  4. #24
    ROG Member Array Jane Doe PC Specs
    Jane Doe PC Specs
    MotherboardAsus Maximus XI Hero
    ProcessorIntel i5-9600k
    Memory (part number)Kingston 2x8gb 4000 (b-die)
    Graphics Card #1Asus Strix GTX 1070ti
    MonitorDell U2414H
    Storage #1Samsung 860 evo
    Storage #2Western digital black
    CPU CoolerNoctua NH-U14S
    CaseFractal define R4
    Power SupplySeasonic 760w platinum
    Keyboard Logitech G105
    Mouse Logitech G400
    Headset/Speakers Creative 5.1
    OS Windows 10 pro

    Join Date
    Aug 2017
    Reputation
    10
    Posts
    15

    Yes, whea is a pain, and I think Realbench is not enough to test stability Someone told me to test stability with new prime95 bcs many new
    games and emulators as prime95 using avx2 and fma3 instructions. So, if your OC cant handle new prime95 with avx instructions, we can said that
    system is not stable. Your system is not stable if he can handle aida64 or realbench. Try new version Prime95 and use first option small blend or what
    and leave them to work about 2 hours. If pass you are 99.9% stable,offcourse no whea or bsod. New prime require higher voltage than cinebench, realbench, old prime or aida64. So try to get stable 4.8 avx for beginning and after that if temperature allow you go for more. In my case I got stable 4.8ghz new prime avx for about 2 hours and voltage 1.278v. Before, for stable cinebench, aida64 fpu or old prime my system required only 1.215v or less. And my new parameters for Prime95 4.8 Ghz are: avx=0, adaptive 1.270v + offset 0.020v LLC= 5. Stable without any whea after 8 hours of play Metro exodus. If you try Prime95 with your last parameters, I'm 100% positive that prime will stop workers or maybe you will get BSOD until bump your vcore on 1.370v or higher for 5.0ghz so I'm suggesting you to make stable 4.8Ghz for beginning. Metro is a good game for testing stability and whea. Someone will tell you that railbench or Aida64 is enough but it is not,and I learned that on my own skin. Wish you luck man.
    Last edited by Jane Doe; 03-06-2019 at 07:17 PM.

  5. #25
    ROG Enthusiast Array
    Join Date
    Aug 2014
    Reputation
    46
    Posts
    41

    Quote Originally Posted by Jane Doe View Post
    Yes, whea is a pain, and I think Realbench is not enough to test stability Someone told me to test stability with new prime95 bcs many new
    games and emulators as prime95 using avx2 and fma3 instructions. So, if your OC cant handle new prime95 with avx instructions, we can said that
    system is not stable. Your system is not stable if he can handle aida64 or realbench. Try new version Prime95 and use first option small blend or what
    and leave them to work about 2 hours. If pass you are 99.9% stable,offcourse no whea or bsod. New prime require higher voltage than cinebench, realbench, old prime or aida64. So try to get stable 4.8 avx for beginning and after that if temperature allow you go for more. In my case I got stable 4.8ghz new prime avx for about 2 hours and voltage 1.278v. Before, for stable cinebench, aida64 fpu or old prime my system required only 1.215v or less. And my new parameters for Prime95 4.8 Ghz are: avx=0, adaptive 1.270v + offset 0.020v LLC= 5. Stable without any whea after 8 hours of play Metro exodus. If you try Prime95 with your last parameters, I'm 100% positive that prime will stop workers or maybe you will get BSOD until bump your vcore on 1.370v or higher for 5.0ghz so I'm suggesting you to make stable 4.8Ghz for beginning. Metro is a good game for testing stability and whea. Someone will tell you that railbench or Aida64 is enough but it is not,and I learned that on my own skin. Wish you luck man.
    You're missing out on something here.
    You can NOT use prime95 to test AVX/AVX2 stability at high overclocks. The power draw will be too massive, which will make the voltage "vdroop" too high (Vdroop = Amps * loadline resistance in mOhms), so you will crash because your load voltage will drop too low if you are pulling 180 amps of current directly to the CPU.

    If you raise voltage high enough to not drop below the required true load voltage "VMIN" for AVX small FFT's, you're going to reach 115C VERY quickly. See?
    So therefore you must use Realbench 2.56.

    If you really want a worst case prime95 AVX test that will NOT drop your "Voltage VMIN too low from excessive current draw, you can try this test, since there are only two memory channels for AVX threads.

    Using prime95 29.6 build 6: installed in two different folders, run TWO stress tests at once, together:

    1) Smallest FFT AVX disabled/AVX2 disabled, # of threads: 14 (for 9700K, set this value to 6 threads).
    2) Custom test (click blend then click custom): 512K-8192k FFT range, test time=0, AVX2 disabled, AVX enabled, Memory size to test: 5648 MB (or something close), # of threads to test: 2 (both on 9900K and 9700K).

    Start test #1 then start test #2.
    Any errors will happen on test 2, not test 1.
    If 2 AVX threads do not crash, then no games should crash and realbench 2.56 should pass an 8 hour stress test also.

  6. #26
    ROG Member Array Jane Doe PC Specs
    Jane Doe PC Specs
    MotherboardAsus Maximus XI Hero
    ProcessorIntel i5-9600k
    Memory (part number)Kingston 2x8gb 4000 (b-die)
    Graphics Card #1Asus Strix GTX 1070ti
    MonitorDell U2414H
    Storage #1Samsung 860 evo
    Storage #2Western digital black
    CPU CoolerNoctua NH-U14S
    CaseFractal define R4
    Power SupplySeasonic 760w platinum
    Keyboard Logitech G105
    Mouse Logitech G400
    Headset/Speakers Creative 5.1
    OS Windows 10 pro

    Join Date
    Aug 2017
    Reputation
    10
    Posts
    15

    Thank you for advice Falkentyne. I will try your method of testing, but my cpu is 9600k and it is much colder than 9700k or 9900k.

Page 3 of 3 FirstFirst 1 2 3

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •