Announcement

Collapse
No announcement yet.

Defective Mainboard or not?

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

  • Defective Mainboard or not?

    Hi

    I suspect I have a defective Mainboard but I'm not really sure. The mainboard in question is a Gigabyte X79-UP4 rev1.0 with a Core-i7 3930K CPU. I have used different sorts of RAM (all tests were run with stock speeds, no OC):

    - 2 different sets of G.Skill 64GB F3-17000CL11Q2-64GBZLD (RipJawsZ)
    - One set of 2* 2GB OCZ OCZ3P1600LV12GS (two out of six modules used; this RAM is known to be good). These two modules were used in RAM slots DDR3_1 and DDR3_3.

    With Memtest86 test #6 six always fails in address ranges < 1MB, regardless of RAM used.

    Here is a screenshot for the first set of G.Skill RAM


    Here's a screenshot for the second set of G.Skill RAM



    And here's a screenshot for the OCZ RAM



    After these tests I returned the mainboard to the shop where I bought it. They sent it back to me, claiming that the board was not defective because it did not show any errors with Memtest+.

    What do you think? Are these real errors or false positives? Any comments are greatly appreciated.

    Yours


    Paul

  • #2
    What version of MemTest86+ were you using?

    In my opinion any consistent memory error that is limited to a few bytes in the 0MB to 1MB range is suspect and maybe a false error.

    The reason is that there are different devices that are commonly memory mapped into this range. (meaning the devices I/O ports overlay some of the main RAM). The range of addresses used by these devices should be excluded from testing via the BIOS memory map. But it seems this doesn't always happen (BIOS bugs maybe). So instead of reading and writing main RAM, we end up reading and writing to some random hardware device, with unpredictable results.

    We are discussing internally what to do about this. Systematically excluding the first 1MB from testing is one solution under consideration.

    In your case I suspect you have some device mapped a the 0x000348F0 - 0x000348F8 range.

    Can you boot into Windows and check the memory map in device manger (select view resources by type from the menu). As per the screen shot below to see if the 348F0 - 348F8 is a mapped to anything (or just post a screen shot like the one below).

    Comment


    • #3
      I don't know which version of Memtest+ was used by the shop where I returned the board to, they did not specify. When I got the board back, I tried Memtest86+ V4.20 but only ran test #5 (block move) there for about half an hour without errors.

      After that I tried Memtest86 V4.2.0 (test #6; block move) and restricted testing to ranges <1M. A couple of secs later I got a bunch of errors. I can run all of the tests successfully for 12h+ if I restrict testing to ranges >1M.

      IO map for Linux (cat /proc/iomem) is

      00000000-0000ffff : reserved
      00010000-0009e7ff : System RAM
      0009e800-0009ffff : reserved
      000a0000-000bffff : PCI Bus 0000:00
      000c0000-000dffff : PCI Bus 0000:00
      000c0000-000cc3ff : Video ROM
      000e0000-000fffff : reserved
      000f0000-000fffff : System ROM
      00100000-bca45fff : System RAM
      ...


      Here's the windows screenshot

      Last edited by paul.pech; Apr-15-2013, 12:21 AM.

      Comment

      Working...
      X