Announcement

Collapse
No announcement yet.

Memtest only runs on 1 CPU, crashes otherwise. Also, weird memory amount displayed.

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

  • Memtest only runs on 1 CPU, crashes otherwise. Also, weird memory amount displayed.

    Starting memtest crashes virtually instantly (I see a blue screen flash and then back to the BIOS)

    Here's the last thing I see when I do the boot trace:

    http://i.imgur.com/ugKGQ3Q.png

    And here's what's weird. If I boot into single CPU mode (option 4), it actually runs fine. It's been a LONG time since I've run Memtest, but it seems like it's running REALLY fast.

    http://i.imgur.com/z2Jykkz.png

    25 iterations in 8 seconds? Is that because it's detecting so little memory? The machine is an HP Z400 work station, xeon processor with 8gb memory. Windows 8 detects the full 8gb memory fine.

    Any thoughts?

  • #2
    This is a known issue with HP Z800, Z400 workstations (might also effect Z200 and other Z models as well).

    Only 596KB of RAM is detected. Resulting the the testing being too fast (and incomplete).

    I did see 1 report saying that using on 2GB at a time worked. The suspected cause if a bad entry in the BIOS e820 memory map in BIOS. (i.e. a BIOS bug).

    Can you check if you have the latest BIOS for the machine.

    We'll have another look at it in any case.

    Comment


    • #3
      Originally posted by David (PassMark) View Post
      .....Can you check if you have the latest BIOS for the machine.

      We'll have another look at it in any case.
      Thanks for the reply - I updated my BIOS one version but there was no difference with Memtest86 - it still crashes instantly with multiple CPU's/cores enabled, and it still doesn't detect the correct amount of RAM.

      The only noted fixes in the BIOS update were:
      Originally posted by HP Support Site
      The following enhancements have been added to BIOS version 03.57:
      - Updated to latest Intel microcode.
      So, no change in my experience at all (crashing and memory discovery wise).

      Let me know if I can provide anything more.

      Comment


      • #4
        Originally posted by Chunks View Post
        Thanks for the reply - I updated my BIOS one version but there was no difference with Memtest86 - it still crashes instantly with multiple CPU's/cores enabled, and it still doesn't detect the correct amount of RAM.

        The only noted fixes in the BIOS update were:


        So, no change in my experience at all (crashing and memory discovery wise).

        Let me know if I can provide anything more.
        Thanks for that. We believe we may have determined the cause of the crash with multiple CPUs enabled. The crash occurred while attempting to boot up all processors in the system. MemTest86 detected 8 processors while there should only be 4 (not including hyperthreads) with the Xeon W3565, causing it to boot either a hyperthread or non-existent processor. We're working on a fix for that now.

        As for the memory size problem, we're also looking into adding additional boot trace statements to the code to output the memory map reported by the BIOS. In the meantime, can you send a screenshot of the memory map by pressing 'c' for Configuration -> Miscellaneous Options -> Print Memory Map. Thanks.

        Comment


        • #5
          Originally posted by keith View Post
          As for the memory size problem, we're also looking into adding additional boot trace statements to the code to output the memory map reported by the BIOS. In the meantime, can you send a screenshot of the memory map by pressing 'c' for Configuration -> Miscellaneous Options -> Print Memory Map. Thanks.
          http://i.imgur.com/UHlhW9N.jpg

          That's.... it. I hit escape, tried again, nothing more than that. I didn't give it much time, but I figured that was part of the problem. I'm happy to test more if I can.

          Comment


          • #6
            Originally posted by Chunks View Post
            http://i.imgur.com/UHlhW9N.jpg

            That's.... it. I hit escape, tried again, nothing more than that. I didn't give it much time, but I figured that was part of the problem. I'm happy to test more if I can.
            Thanks, it looks consistent with the low amount of memory we saw.

            Can you try the following debug build with boot trace enabled, at your convenience:

            (USB image) http://www.passmark.com/ftp/memtest86-usb-4.3.4e.zip
            (CD image) http://www.passmark.com/ftp/memtest86-iso-4.3.4e.zip

            You will probably need to take multiple images as the trace entries will overwrite each other. A good time to take the snapshots is right before the arrow ('>') goes back to the first line again. If possible, please grab the boot traces up to when the memory tests begin (ie. the progress bar starts moving). Thanks in advance.

            Comment


            • #7
              http://i.imgur.com/V7D9Hxc.jpg?1

              Do you still need the trace entries? Looks like the cpu issue is resolved for me. I did actually make a video of the boot traces I can throw on youtube if it'd be interesting for you. No problem at all, but maybe the fact that it loads fully is good enough for you

              Comment


              • #8
                Originally posted by Chunks View Post
                http://i.imgur.com/V7D9Hxc.jpg?1

                Do you still need the trace entries? Looks like the cpu issue is resolved for me. I did actually make a video of the boot traces I can throw on youtube if it'd be interesting for you. No problem at all, but maybe the fact that it loads fully is good enough for you
                Thanks for the quick turnaround. Good to hear the CPU problem has been fixed.

                Yes, please upload the video of the boot trace. We still need to get to the bottom of the memory size issue.

                Comment


                • #9
                  Originally posted by keith View Post
                  Thanks for the quick turnaround. Good to hear the CPU problem has been fixed.

                  Yes, please upload the video of the boot trace. We still need to get to the bottom of the memory size issue.
                  http://youtu.be/eauMJClysU0

                  If anything isn't clear from that, I can provide more (screenshots, logs, video, whatever!). I hope I went long enough - it seemed to be stepping through the individual tests, so in the end I killed the trace.

                  Comment


                  • #10
                    Originally posted by Chunks View Post
                    http://youtu.be/eauMJClysU0

                    If anything isn't clear from that, I can provide more (screenshots, logs, video, whatever!). I hope I went long enough - it seemed to be stepping through the individual tests, so in the end I killed the trace.
                    Thanks for the video, we were able to retrieve what we needed from the trace.

                    It looks like the incorrect memory size is caused by improper processing of the memory map entries returned by the BIOS. The BIOS is indeed reporting more than 596KB but the memory map entries weren't processed correctly so only the first entry (0 - 596KB) was returned.

                    Unfortunately, we need to inconvenience you once more with another boot trace video/snapshot. Can you get boot traces of the following build when you have the time:

                    (USB image) http://www.passmark.com/ftp/memtest86-usb-4.3.4f.zip
                    (CD image) http://www.passmark.com/ftp/memtest86-iso-4.3.4f.zip

                    Hopefully, this will be all we need to fix the problem.

                    Comment


                    • #11
                      It's no problem at all. Takes all of 5 minutes to do. In case it makes things easier, I'm only using the USB image.

                      http://youtu.be/-cy13KVp3RY

                      Comment


                      • #12
                        Originally posted by Chunks View Post
                        It's no problem at all. Takes all of 5 minutes to do. In case it makes things easier, I'm only using the USB image.

                        http://youtu.be/-cy13KVp3RY
                        Thanks for that. We think we may have fixed the logic error in the code. Can you try the following build:

                        http://www.passmark.com/ftp/memtest86-usb-4.3.4g.zip

                        Try it without boot trace first. Hopefully we'll get the correct memory size this time.

                        Comment


                        • #13
                          Originally posted by keith View Post
                          Thanks for that. We think we may have fixed the logic error in the code. Can you try the following build:

                          http://www.passmark.com/ftp/memtest86-usb-4.3.4g.zip

                          Try it without boot trace first. Hopefully we'll get the correct memory size this time.
                          That did it! I'm replying from my mobile, because it doing its job and hopefully I'll figure out if its the ram or something else causing my system to crash every day.

                          30 minutes, 2ish iterations. Much better, it's chugging away.

                          Thanks again for doing all this work. If there's anything more that it would be beneficial to check, I'm more than happy help. I work in IT and my real job has been quite unfulfilling lately - helping you guys and actually seeing progress and results makes me happy! otherwise, have a good one!

                          Comment


                          • #14
                            Originally posted by Chunks View Post
                            That did it! I'm replying from my mobile, because it doing its job and hopefully I'll figure out if its the ram or something else causing my system to crash every day.

                            30 minutes, 2ish iterations. Much better, it's chugging away.

                            Thanks again for doing all this work. If there's anything more that it would be beneficial to check, I'm more than happy help. I work in IT and my real job has been quite unfulfilling lately - helping you guys and actually seeing progress and results makes me happy! otherwise, have a good one!
                            That's good news. Thanks to your quick turnaround, we were able to get a fix up in relatively short time. Usually it's hard to track down traces and logs for issues like these as they are hardware/platform related. We wouldn't be able to reproduce it on our systems without the specific hardware. But I'm sure there are other users that are vulnerable to the same issue.

                            You can continue to use this build until we get an official build released. Thanks for all your assistance.

                            Comment

                            Working...
                            X