PassMark Software

Announcement

Collapse
No announcement yet.

NEW G.Skill RAM, 8 hours test 1 error, bad ram?

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • NEW G.Skill RAM, 8 hours test 1 error, bad ram?

    Hello there again, after i returned my previous ram because it was faulty (https://www.passmark.com/forum/memte...c-pass-nr-6-7v) i bought another ones, now G. skill flare X and again got error somewhere between 6-8 hour mark.

    Could you guys PLEASE help me identify what kind of error it is? And should i return the ram again?
    I have to say, this was with XMP mode enabled and set to its highest frequency, would it make any difference to try and tweak the settings of the ram a little and run the test again (Like lower latency or higher Vcore and so on? (Does memtest take into consideration BIOS settings of the ram? or it automaticly test it with its highest settings?)

    So could this possibly be error due to wrong memory settings (like i mentioned above with Vcore, too high frequency and so on) or if it finds error it automatically means the ram is bad.

    Thanks for your help again guys.

  • #2
    Last time you had 371,000 errors. This time just 1 error.

    Having just 1 error makes it harder to test for and fix. Some memory errors are to be expected when you run a computer for a long period of time. So maybe you were just unlucky.

    I would repeat the test overnight to see if the problem can be reproduced before taking any action.

    Then if it can be reproduced you could try different slots or testing just 1 stick at a time. Then as you suggest trying different BIOS settings.
    If is can't be reproduced, then you have no way of knowing if your changes fix anything and I would be tempted to ignore it.

    If this was mission critical work you are doing, then consider using a system with ECC ram instead.

    Comment


    • #3
      Thanks for reply, David.

      So i left memtest run since morning till i got from home, and errors again.

      Maybe the language is a little barrier for me to explain what i mean.

      1. Is it possible that unstable overclock (or XMP/DOCP mode that i have set) is causing those errors? Because i have read many people complaining about unstable XMP/DOCP systems @ high RAM frequencies.

      2. For example, if i set the RAM to run at 1.2 V (default) and "underclock" it for say 2000MHz, will the memtest program run the test with this specific settings i'm gonna apply in BIOS?

      I will continue testing with 1 ram in a slot only and see what i got.

      EDIT: about mission critical, i just finished my new PC build so all i want from the PC is to run stable, nothing else, thats why i test memory, shouldn't newly purchased RAM pass the memtest without problems?

      Thanks again for your time and help!

      Comment


      • #4
        Your past run showed 371,000 errors in the 3MB to 4MB address range. This is really low is memory and typically there are many other devices using memory in the low addresses ranges.
        The run from today showed 18,000,000 errors in the 2MB to 3MB range. This is a large number of errors in a small address range with a large number of bits in error (all 32bits in fact).

        So as I already pointed out, I suspect this is a BIOS bug. But you seemed to have ignored that advice and instead replaced the RAM with new RAM. But then continued to get the same errors.

        XMP/DOCP isn't really overclocking as the manufacturers has tested and support these speeds. MemTest86 never changes any RAM settings. So it just uses whatever is set in BIOS.

        Comment


        • #5
          Thanks for your reply again, David!

          Yes about last time, there was not enough time for proper testing, all i did was just that one test and upon that i had to decide if should i return the memory and got my money back or keep it, do all the tests i need and then if i find out its faulty it would have to be RMAd what would make it slow process.

          I did try to enable lower frequencies now, and tested each RAM module separately, and after 8 passes for each module, 0 errors (about 5 hours 30 minutes each) (at slots they were in the first time i tested them both together, so i guess i will try another test with them both together in there with this lower frequency settings of 2933 as i tested those separately?)

          Thanks a lot about clarifying!

          Comment


          • #6
            I would just like to add a little update to my previous post.

            I let the memtest run with both RAMs in their previous possitions but @2933MHz 14-14-14-34 1.35V and it did run for 11.5 hours with zero errors. (116 test in total, dont remember number of passes but more than 10)

            Could that be considered stable?

            Comment

            Working...
            X