PassMark Software

Announcement

Collapse
No announcement yet.

Question about multiple failed tests across RAM modules

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

  • Question about multiple failed tests across RAM modules

    I ran memtest on a 2x8GB set of RAM (PNY mfd). They had 500,000+ errors in test 6 across DIMM slots and individually and in pairs. Ordered new RAM (crucial vengeance, 3000MHz, 2x8GB).

    They also failed on the same tests, ~500,000+ errors, on tests 6, 8, and 9.

    The odds of 2 brands of RAM failing in the same way seem very small to me.

    Question is then: is this pointing at my motherboard? My CPU? I'm not sure the hierarchy of hardware here and what other information I could gather. I've ordered another motherboard to test things out with, but would love more input on to what seems likely.

    Thanks for a great tool!

  • #2
    Could be many things.
    Would be good to see an extract from the log. In particular how many bits are in error.
    Would also be good to test different CPU cores to see if the result is the same.
    Could also try underclocking the RAM to see if that reduces the error count.

    Could be a free memory map BIOS bug resulting in the wrong sections of RAM being tested. (is there a BIOS upgrade available?)
    Could be a bad slot on the motherboard or even the CPU.


    Comment


    • #3
      Here's the output of a whole test. 2 sticks in the recommended slots. Anything useful gleaned here?

      Comment


      • #4
        That is a stupidly high number of errors. There really is something wrong.
        All the errors seem to be a single bit error. Which would normally point to bad RAM (and not a BIOS bug). But given the number of errors and the fact you have already swapped the RAM, maybe it is one of the bus lines that are bad, or one of the memory slots. Did you try testing with 1 stick in 2 different slots.

        Comment


        • #5
          Ok, also just tried running single cores at a time. 0, 2, and 3 all made it clear of test 6 the most troublesome one. Core 1, however, never started a timer nor began any tests, just hung memtest86 and forced a hard reboot. That seems pretty conclusive to me that somehow one core of my CPU went bad. Is that a correct interpretation?

          EDIT: Or is could the motherboard still be involved somehow? (Unsure of the physical nature of the components in play here, does each CPU get a separate wiring or something that could fail and affect communication with just one CPU, for example)

          Comment


          • #6
            Originally posted by alex-mcp View Post
            Ok, also just tried running single cores at a time. 0, 2, and 3 all made it clear of test 6 the most troublesome one. Core 1, however, never started a timer nor began any tests, just hung memtest86 and forced a hard reboot. That seems pretty conclusive to me that somehow one core of my CPU went bad. Is that a correct interpretation?
            This could be a UEFI BIOS issue. It is better to start with assuming the error is within the RAM first, then work your way down. If you get similar errors after swapping your current RAM with known good RAM, then I would start looking at the CPU/mobo.

            Did you try testing with the legacy MemTest86 v4? If you get a different result, it may indicate a BIOS issue.

            Comment

            Working...
            X