No announcement yet.

1st test, >10000 errors after 8 passes. 2nd test, 0 errors after 8 passes?

  • Filter
  • Time
  • Show
Clear All
new posts

  • 1st test, >10000 errors after 8 passes. 2nd test, 0 errors after 8 passes?

    I have been having issues that I suspect are caused by memory issues. Apps crashing or closing unexpectedly, Windows BSOD for various error codes, driver issues, etc.

    I have eVGA P55 SLI motherboard with an Intel i7 2.8GHz CPU, 16 GB (4x4) DDR3 1600 MHz RAM. There is no UEFI for my motherboard, just BIOS, so the older memtest version (v4.3.7?) is ran.

    The first time testing, I ran for over 8 passes. Every pass, including the 1st, many errors were found. When I came back from overnight testing, 10000+ errors were reported.

    I decided to repeat the testing because I exited before studying the results close enough. The second time testing, I again ran for over 8 passes. However, 0 errors were reported.

    It seems very strange to me that such a discrepancy in errors could occur between test cycles. I did not touch any hardware between test cycles. No changes were made, setup was identical.

    I am running again for a third time, but this time I switched from round robin CPU mode to sequential CPU mode. However, the third attempt has still not yet detected any errors (I will let it continue for 8+ passes).

    Any reason such a drastic difference in results could occur?

  • Schumacher574
    I didn't open the case or re-seat the RAM between test cycles.

    I didn't think memtest 4.3.7 (BIOS) gave the option for parallel, unless I missed it. I thought the options were round robin (default) and sequential.

    Regardless, during my third test cycle I found 4 errors during Test #6 or 7, so I figured I might as well rule out the RAM and purchased new memory. If the issues continue, it likely is a motherboard problem.

    Leave a comment:

  • David (PassMark)
    Did you open the case at all? Maybe re-seat the RAM in the slots?

    If you get errors in pass 1, then generally there is nothing to be gained to letting it continue.

    Testing on all CPUs (Parallel) should be the most effective.

    Leave a comment: