New Ryzen system very weird problems with RAM and XMP

Discussion in 'Processors and motherboards AMD' started by Serris, Jan 27, 2021.

Tags:
  1. Serris

    Serris Guest

    Messages:
    44
    Likes Received:
    0
    GPU:
    GTX 960
    Ok so I did two more memtest86 runs. Ran the memtest86 with default voltages and the error showed up again.
    Increasing the SOC voltage to 1.07V, and memtest86 passed all the tests.

    Thanks everyone for your help, I really appreciate it!
     
  2. Undying

    Undying Ancient Guru

    Messages:
    25,330
    Likes Received:
    12,743
    GPU:
    XFX RX6800XT 16GB
    Passing a memtest isnt a sign of stability. My 2x16gb 3200 dual rank pass memtest few hours at 3466 and as soon i start gaming it crashes after few min.
     
    ViperXtreme likes this.
  3. Serris

    Serris Guest

    Messages:
    44
    Likes Received:
    0
    GPU:
    GTX 960
    Passing an 8 hour memtest86 is a sign that the RAM is probably fine. It's not the only thing that affects system stability.
     
  4. INSTG8R

    INSTG8R Guest

    Messages:
    1,659
    Likes Received:
    95
    GPU:
    Nitro+ 5700XT
    Yeah 1.1.9 was considered buggy and most pulled it.
     

  5. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,011
    Likes Received:
    7,351
    GPU:
    GTX 1080ti
    lol, no.

    passing 3 days of memtest is a sign, 8 hours is nothing.
     
  6. Serris

    Serris Guest

    Messages:
    44
    Likes Received:
    0
    GPU:
    GTX 960
    I fully disagree.

    If there is a serious issue with RAM, memtest will pick it up probably before the first pass even finishes.

    RAM also isn't extremely reliable over longer periods, probably why ECC RAM was made.

    Running memtest for 3 days might eventually result in soft error.
     
  7. GREGIX

    GREGIX Master Guru

    Messages:
    855
    Likes Received:
    222
    GPU:
    Inno3d 4090 X3
    TM5 with antaextreme settings and prime95 small 112 fft in place without AVX is for fast, safe testing ram stability. Memtest nah, too long...
    Damn I hate autocorrect...
     
  8. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,011
    Likes Received:
    7,351
    GPU:
    GTX 1080ti
    No, and i'm not discussing this with you, because you clearly don't know silicon electronics.


    "Here is something to keep in mind though. Memtest will only prove that your RAM has issues, but it will not prove that it does not have issues (even after 10 hours of tests). Even if you run the test 100 times you can not be sure that it would not spit out an error after the 101th run."

     
    Last edited: Jan 31, 2021
  9. Serris

    Serris Guest

    Messages:
    44
    Likes Received:
    0
    GPU:
    GTX 960
    I assume you clearly do?

    Since you already copied a part of Superuser/LTT comment, let me fill in the rest.

    "In most cases memtest will start spitting out errors within a minute if the RAM stick is bad. If you ask me, I'd say after 1 minute without errors you can be 50% sure that the RAM is good. After 5 minutes it's 70%. After one pass it's 90%. After 3 passes it's 99.9%, and so on."
     
  10. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,011
    Likes Received:
    7,351
    GPU:
    GTX 1080ti
    Not necessarily,

    you might test for 8 hours during a cold night and have ram that only starts throwing errors on a spring day once the ic's are beyond 40c, electromigration and cell electron leakage are the common most causes of temperature related memory testing failures so having tested for a few hours at the wrong time of day can give you a false sense of security.

    when i'm testing ram i block or shutoff the fans and hotbox it to specifically test for such cases but some people really do not want to starve their machine of air.

    you might also have a motherboard fault that can survive aida64 stress testing for 5 hours, but then the next time you open a browser the system almost fully hangs because voltage instability causing non-critical driver corruption has been paged out and the next time its referenced by the graphics device it becomes unresponsive.

    multi-day memtesting is more about the ambient changes throughout the days than anything.

    I've seen a prime95 blend fail (in a case of memory instability) that memtest cleared after days of testing too, so its not ok to say dram is fine with any amount of memtesting, memtest will find the outright ****** ram for sure but these plenty of problems with DRAM that won't be picked up the way memtest works.

    In a couple of the ambient temp cases we took the dimm apart and stripped it back to find microscopic cracks in the die
     
    Serris likes this.

  11. Serris

    Serris Guest

    Messages:
    44
    Likes Received:
    0
    GPU:
    GTX 960
    You were right, I don't understand silicon electronics.

    However, temperatures (at least in our house hold) are usually somewhat constant throughout the year. Around 22-23°C in winter and 21-24°C in summer.

    Would those temperature fluctuations also impact DRAM reliability?

    Also, from what you said, supposedly even multi day memtest86 isn't good enough, are there any other utils you also use?
     
  12. Astyanax

    Astyanax Ancient Guru

    Messages:
    17,011
    Likes Received:
    7,351
    GPU:
    GTX 1080ti
    for bruteforce online testing? Prime95.

    But at the end of the day, nothing will compare to pulling dimms and using the system in the same scenarios as it crashed previously or swapping ram with assumed good.

    And at the end, you might even find the cpu imc was weak, the entire time.
     
    Last edited: Jan 31, 2021
    Serris likes this.
  13. ViperXtreme

    ViperXtreme Ancient Guru

    Messages:
    3,348
    Likes Received:
    229
    GPU:
    RTX 4070 Dual UV
    Stress test? Sometimes the instability comes out on light to medium/everyday loads and not from these stress tests.
     

Share This Page