Announcement

Collapse
No announcement yet.

Need help interpreting memtest result

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

  • Need help interpreting memtest result

    Hi, so recently I've been getting random BSODs every few days so after some searching I found out that this usually happens when there is a bad RAM so I decided to try run this memtest and I did indeed get some fails. However, I have noticed strange things with the test results and I was hoping someone could help me out with this.

    My specs are
    Memtest: MemTest86 V7.1 Free (64-bit)
    OS: Windows 10 home edition
    RAM: 2x8GB DDR3-1600 CL10-10-10-30 1.5V 240PIN
    CPU: Intel Core i5 4690k 1150 LGA
    Mobo: ASUS Z97-A
    GPU: GeForce GTX 760
    XMP enabled

    Basically, my first test had errors around the address 0x1C5D9038C(7261MB).
    After that, I tried testing each stick individually but I found no errors.
    And then I tried switching the slot of the 2 sticks and I found out that the error was still happening near the same address around 0x1C5D9038C.
    I've also noticed that every time there is an error, it seems that it happens with CPU 2.
    Also I tried running the test on UEFI and BIOS and I noticed that I get waaaay more errors when running this test on BIOS.
    Is this really a RAM problem or is there possibility that something else might be causing these errors?
    My initial thought was that it may be CPU error instead of RAM problem as seeing it kept happening with CPU 2 and seeing how error address did not change even after swapping position, but if it truly was CPU error, wouldn't the error happening at random places and not at that specific address?
    I'm totally lost on what to make of these tests.

    Here is the imgur album with few of my test results
    http://imgur.com/a/AfT4J

  • #2
    Can you try the following:
    - Run with a set of known good RAM modules with conservative timings (or at least with XMP disabled). And/or move the RAM modules to another PC to rule out any problems with the RAM
    - Run in single CPU (0) mode
    - Check for a BIOS update. Errors at specific address ranges may indicate a bug in memory map allocation

    Comment


    • #3
      When I first started getting BSOD first thing I did was doing a clean reinstall of Windows 10 along with updating all my drivers including my BIOS so I do not think this is being caused by outdated driver or anything software related.
      I tried running the MemTest with XMP disabled and with single CPU (0) mode but I still got errors at that specific address.
      Here are the results
      XMP disabled
      CPU(0) mode
      BTW, even after I disabled XMP through my BIOS settings, the MemTest result page still said my RAM was 1600 MHz XMP. Is that supposed to be like that?

      Also I do not know if this is relevant or not as this is not MemTest related strictly speaking but I was also recommended to run Prime95 stress test.
      The Blend test for Prime95 which supposedly test both CPU and RAM failed after about an hour while Small FFTs test for Prime95 which supposedly only stress CPU for most part was able to run 3 hours without fail.
      Last edited by CallMeCrouton; Dec-01-2016, 08:25 PM.

      Comment


      • #4
        I think it would be a good idea to rule out RAM being the issue by swapping with a set of known good RAM (preferably different vendor). If you still get errors, at least you can reasonably determine the issue is further down the data path.

        BTW, even after I disabled XMP through my BIOS settings, the MemTest result page still said my RAM was 1600 MHz XMP. Is that supposed to be like that?
        MemTest86 only reports the static timing information stored in the RAM SPD; it does not detect the active configuration. So XMP will be reported whether or not it is enabled.

        Comment


        • #5
          Thank you for your replies Keith.

          So I decided to take the risk and buy a new RAM while they were still on cyber Monday sale and after running some Memtests with the new stick I got a good news and a bad news.

          Good news is that the original error no longer seems to be appearing meaning that my old RAM should have been what was causing the problem.

          Bad news is that I got a different error now but this seems to only happen in a specific scenario after multiple testing. It seems that with the new RAM, I sometimes fail test1 around 2256MB ~ 2340MB range.
          Some of these errors can be seen here.
          However, this error seems to only happen the first time I run the memtest after I turn on my computer, and even then, it only happens roughly 50% of the time and only on the first pass about 5 seconds into test starting.
          Basically, from what I observed, this error only happens when I: Shut down computer -> Turn on computer -> Go to BIOS -> boot from Memtest USB -> start test -> 50% of the time I will get test1 fail within 5 seconds -> if I stop the test and try starting again without turning off my computer this error will not happen again no matter how many time I try, only way to get this error again is to shut down computer and start from beginning again.
          It seems that this is the only possible point where test can fail as after this point, whether I got error here or not, I did not find any other errors past this point.

          Any idea why this might be happening? Would this be a problem? Did I get unlucky and buy a defective RAM or is there possibility that MemTest can return false positive when running the test on freshly booted computer as this only seems to have possibility of happening on the first test I run after starting my computer.

          Also Prime95 stress test which failed after about an hour on my previous RAM was able to run for 6 hours with new RAM without failing but I've only done this twice so far so I will need to do some more testing with Prime95 before I'm sure Prime95 failure has been fixed with new RAM.

          Comment


          • #6
            Looking at the expected and actual values from the memory errors of both the old and new RAM, it seems that the error is always in the most significant byte which suggests that the issue may not be with the RAM. I assume you are getting the same errors with MemTest86 v4 in BIOS mode as well?

            There is a possibility that it may be a bug in the BIOS, as memory errors in a specific address range may indicate that the BIOS is not properly marking the area off as reserved which prevents the range from being tested.

            Comment


            • #7
              Yes, I'm getting same error with MemTest86 in BIOS mode.
              Do you know any way I can test to see if the problem is the BIOS?

              Also, after bit more testing, it seems that I do not get the error even after fresh boot if I let it sit on MemTest settings screen for ~30 seconds before starting the test.

              Comment


              • #8
                Another update:
                After individually testing each new RAM stick, I found out that I was able to reproduce the cold boot error on one stick pretty reliably on all the DIMM slots (usually within 3 cold boots) but I could not reproduce the error on the other RAM no matter which DIMM slot I tried after something like 30 cold boots total. Seeing this, is it pretty safe to assume that I got unlucky and got bad RAM and the fact that most significant byte being wrong on both old and new RAM is just coincidence?

                The error I got from testing individual RAM stick can be seen here

                Also, I noticed that when I was testing the stick individually, the error location was around 1150MB which is roughly half of error location when I was testing both stick at the same time which was around 2300MB. I thought that when I had 2 8GB RAM sticks, 0~8GB address would be mapped to first stick while 8~16GB address will be mapped to second stick. However, this suggests that entire 0~16GB memory range is mapped to both sticks in an alternating fashion. Is this how RAM works? Cause this would also explain why when I swapped the slots of my 2 old RAM sticks, the error address stayed around the same location.

                Comment

                Working...
                X