PassMark Software

Announcement

Collapse
No announcement yet.

Memtest freezes with quad channel

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

  • Memtest freezes with quad channel

    I built a new computer for video editing, some new parts and some reused from old system.
    Hardware spec is as follows:
    Intel 6800k
    NZXT Kraken X61
    ASUS X99-E
    32GB Corsair LPX 3200 (8x4), in A1,B1,C1,D1 slots like manual suggest.
    EVGA Supernova G2 750W
    Nvidia GTX 670 (from old system)
    500GB Samsung 850 EVO (from old system)
    256GB Samsung 950 PRO m.2
    10Gbit NIC to storage server.

    I manually set the memory speed to 3200mhz on first statup, left most of the other stuff at default. I didn't reinstall windows 10 from old system, booted and played around a bit installing drivers and other stuff, got bsod. Then I couldn't log into windows any more, safe mode worked but was unable to fix windows. So I thought it must have been some driver issue or something else. Reinstalled windows clean and reverted the manual memory change back to auto(2133mhz). Windows installed fine, system seemed to work ok but as soon as I tried to encode a video with Adobe it used all memory and then I got a bsod and windows rebooted. Tried this a couple of times with the same result.

    So then I restarted and booted Memtest86 7.1 free(single cpu as bios is not compatible with the other modes), ran test for about 8h and then canceled without any errors(memory at 2133mhz). Started windows and ran Prime95 for 6h with small FFT and another 16h in blend mode, no errors. Adobe still fail.
    Frustrated about buying 3200mhz memories and running them at 2133mhz I went into bios and activated XMP profile, making them run at specified speed(3200mhz).
    I again started a memtest run and this time memtest froze/locked up completely after about 2,5h(pass 2), no errors reported. CPU at 41 degrees at lockup. Hard reboot and restarted the test, this time it froze after just under 2h at the end of pass 1, no errors this time either. Shut down the computer and removed memory sticks from C1 and D1, reran the test with only two 8gb sticks in A1 & B1. It completed all four passes without errors or lockups. Removed the "good" memory sticks and installed the other two from C1 and D1 in A1 and B1. These also made it through all four passes without lockups or errors.

    With only two sticks in the system I dont get any bsod in windows encoding with Adobe even though the encoding fail for running out of memory it doesn't bsod at least, but this is a separate issue with adobe. Before at 2133mhz and all four sticks adobe would cause a bsod, and as the four sticks make memtest lockup at 3200mhz but not with two modules make me believe the memory sticks are good. ASUS mb or cpu seems to have issues with it running in quad channel mode. Would this assumption be correct?
    Any way to rule out either mb or cpu without actually buying a second set of hardware to test with?
    Is it common that memory issues like this locks up memtest?

  • #2
    I can add that memory kits are two 2x8GB kits if that makes a difference.

    Comment


    • #3
      Yes, bad RAM can lockup MemTest86 (and Windows, and Linux, and any other software you run).

      Check for updated BIOS, sometimes they improve RAM compatibility with new BIOS releases.

      It is also isn't unheard of that some systems are stable with 2 sticks of RAM, but become unstable when adding more RAM. More RAM means more current draw and more voltage drop. In some cases manually bumping the voltage levels very slightly in BIOS can help.

      Comment


      • #4
        Problem was solved by returning the two dual-channel kits and getting a matched 64gb(4x16) quad kit. Ran memtest for about 22h and it was at the end of pass 4 at that time when I canceled it.

        Comment


        • #5
          Glad you solved it.
          Things should, in theory, work with unmatched RAM. But in the real world there are occasional compatibility issues.

          Comment


          • #6
            I have also quad core memory on x99 board computer.
            V7.1 is freezing at test 2 (test 2 displayed but it does not start).
            V6.3 and V7.0 are OK.

            Comment


            • #7
              Originally posted by w64bit View Post
              I have also quad core memory on x99 board computer.
              V7.1 is freezing at test 2 (test 2 displayed but it does not start).
              V6.3 and V7.0 are OK.
              This is likely due to the issue described here:
              http://www.passmark.com/forum/memtes...election-modes

              As a workaround for the UEFI bug, select single CPU mode before running the tests.

              Comment


              • #8
                Sorry for the kick, but this topic came up on top with a Google search for Memtest-freezes on X99-systems.

                Originally posted by keith View Post

                This is likely due to the issue described here:
                http://www.passmark.com/forum/memtes...election-modes

                As a workaround for the UEFI bug, select single CPU mode before running the tests.
                Just now I tried to run Memtest86 on my X99-era system, freeze after 15 seconds, every time I try it happens at the same moment. My motherboard is already listed there so I guess it's that bug I ran into.

                Any idea when it will be fixed? Tried with the most recent Memtest (7.2). It's now running in single CPU mode, which seems to go okay.

                Comment


                • #9
                  Yes, it is a known UEFI BIOS bug in some motherboards. (The ASUS range is the main problem). See the link above for more details. We have contacted ASUS a few times. They fixed it on a couple of their boards, but not the majority. They don't seem to care much about it.

                  In the next MemTest86 release we are going to black list some motherboards from using multi-cpu mode. So at least they won't crash.

                  Comment


                  • #10
                    Originally posted by David (PassMark) View Post
                    Yes, it is a known UEFI BIOS bug in some motherboards. (The ASUS range is the main problem). See the link above for more details. We have contacted ASUS a few times. They fixed it on a couple of their boards, but not the majority. They don't seem to care much about it.

                    In the next MemTest86 release we are going to black list some motherboards from using multi-cpu mode. So at least they won't crash.
                    Hmm, okay.

                    I've got the X99-A II and I just realized I didn't check for BIOS-updates, if it's fixed that way. Well, that's something for later, memtest has been checking 4*16GB for almost 22hrs now and is 25% away from (hopefully) pass no. 5. It's been a habit to run memtest on new hardware for at least 8 passes, but with this amount of RAM, it takes quite some time...

                    Comment


                    • #11
                      V7.3 will be slightly faster than V7.2, maybe saving a couple of min per pass.

                      Comment

                      Working...
                      X