PassMark Software

Announcement

Collapse
No announcement yet.

MemTest86 ver 4.1.0 and Intel Z68A chipset

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

  • MemTest86 ver 4.1.0 and Intel Z68A chipset

    I have a Celeron G540 running on an MSI Z68A-G43 (G3) motherboard with BIOS 7750v43 (newest version compatible with Sandy Bridge, 3rd newest overall) and 4 x 4GB of Samsung DDR3 memory. This setup has always passed MemTest86 ver. 3.5b perfectly, but with memTest86 ver. 4.1.0 it crashes near the start of test #3, even when only one 4GB memory module is installed. It doesn't matter which DIMM socket or which DIMM is used. MemTest86 ver. 4.1.0 may freeze, reboot, or show a green screen partly or full of zeroes. The motherboard's built-in video was used during testing. I'm sorry for the pictures being fuzzy:

    http://img841.imageshack.us/img841/4507/1002714r.jpg

    http://img268.imageshack.us/img268/9349/1002708p.jpg

  • #2
    One of your screen shots shows thousands of memory errors on Test #3
    Your other screen shot shows an Unexpected Interrupt of type Inv_Op.

    Inv_Op means the CPU was asked to execute an invalid CPU instruction (Invalid Operation). Which causes the CPU to generate an interrupt and then halt. The instruction that the CPU has to execute also comes from the RAM. So bad RAM can also provoke this type of failure.

    There is surely something wrong with your hardware. Maybe a fault has developed since you were using V3.5, or maybe the environmental conditions are now different (e.g. summer vs winter).

    Do you have any RAM from another vendor to try?
    Can you check the RAM timings in BIOS & the voltage levels, especially if you have been overclocking?
    Trying turning off the RAM's XMP profile if you have it on?
    How long has this RAM been working OK? What prompted you to pull out V4.1 and start testing again if this isn't a new machine?

    If you had a spare motherboard, or CPU, I would then suggest swapping them out and retesting if all else fails.

    Comment


    • #3
      Similar errors here if i press any key during test. Therefore i am unable to show configuration screen after pressing the "c" key. Version 3.5b is working fine.

      Lenovo ThinkPad T420 - Intel Core i5-2520M 4GB RAM
      Last edited by bomberman; 02-18-2013, 10:07 AM.

      Comment


      • #4
        Originally posted by bomberman View Post
        Similar errors here if i press any key during test. Therefore i am unable to show configuration screen after pressing the "c" key. Version 3.5b is working fine.

        Lenovo ThinkPad T420 - Intel Core i5-2520M 4GB RAM
        I looked at the screen shots and it's not clear what the issue is. There is a good chance that this is a legitimate hardware error that is only exposed when using more than one core. This seems especially likely since one 2 bits are in error.

        Try running version 4.1.0 with only one CPU enabled (boot option 6). It would also be very helpful to have a screenshot of the boot traces. Follow the instructions found at: http://www.memtest86.com/support/index.htm

        Chris

        Comment


        • #5
          The memory is Samsung brand 1600 MHz modules with actual 1600 MHz chips (K0 suffix), and the results were the same with each module and regardless of which of the 4 DIMM slots was used. The 4 modules have 2 different date codes, a few weeks apart, and had been in use since December and had passed several days of testing with ver. 3.5b, MemTest86+ ver. 4.2.0, and Gold Memory vers. 5.07 and 6.92, at 1.35V while overclocked to 1866 MHz (I normally don't overclock). I also tried MemTest86 ver. 4.1.0 with the RAM at 1.50V, just in case that would change anything.

          I ran ver. 3.5b again, and no errors were reported.

          I also tried MemTest86+. No errors with ver. 4.2.0, but beta vers. 5.00v1 - 5.00v4 gave errors on test #3 (moving inversions) if parallel core testing was enabled but no errors with round robin or sequential core selection.
          Last edited by larrymoencurly; 03-11-2013, 06:31 PM.

          Comment

          Working...
          X