A general observation...

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • Roxxors
    Member
    • Mar 2025
    • 57

    #1

    A general observation...

    I have been testing my system for 2 weeks now. I have Memtest configured to run 36 runs skipping test 10. I have found that the system runs just great when I use it in normal operations, but before bed I boot and run Memtest overnight. In the morning I find 5 or 6 errors, so I boot windows and the machine runs fine all day with no issues till I reboot and run Memtest over night get up and see that 24 runs netted 10 errors, reboot to windows and run all day and no crashes or anything.

    Did this for like 10 days... Memtest has errors but the machine runs just great, soo...

    Conclusion from my testing...

    Memtest overstresses the system (a good thing) and reports errors (that's a good thing) but under normal operation and load those errors never manifest themselves and the system runs just fine.

    Having said that, you milage may vary depending on your hardware...
  • David (PassMark)
    Administrator
    • Jan 2003
    • 11092

    #2
    Memtest overstresses the system (a good thing) and reports errors (that's a good thing) but under normal operation and load those errors never manifest themselves and the system runs just fine.
    You can't generalize the results from testing one computer to all computers.

    Not all memory errors result in a crash.

    You just got lucky and the byte(s) that are being corrupted aren't being used for anything critical. Or at least nothing that you have noticed so far.
    For example they could be silently corrupting your disk as you save files. They could be changing the color of pixels on the screen. They could be changing the results in your Excel spreadsheet just slightly. Adding a few spelling errors into your Emails. Causing errors in your internet connection that then get detected with checksums and re-transmitted in the background, resulting in a slightly slower connection. Or a million other subtle things that you might not immediately notice.

    Comment

    • Roxxors
      Member
      • Mar 2025
      • 57

      #3
      "Having said that, you milage may vary depending on your hardware.."

      This is why I said this... DOH. I know that. Ya, it could be a million things, but in the end Memtest is not the defining test. Whether the system works reliably is the final defining test...

      Comment

      • David (PassMark)
        Administrator
        • Jan 2003
        • 11092

        #4
        It's been 3 days now. My cancer hasn't killed me yet. So any diagnosis of cancer can safely be ignored.

        Comment

        • Roxxors
          Member
          • Mar 2025
          • 57

          #5
          You are not grasping simple concepts here. You run Memtest in one environment, the OS, Windows runs in another. The two don't equate. What is an error in your little Memtest world does not necessarily apply to the real world, although it could. Hardware varies, mine runs just fine for weeks despite Memtest errors. I am sure that the errors are real under the Memtest environment, but not on the real world where Windows operates.

          It's not a Memtest issue. I love Memtest, but I have also found it's limitations...

          Comment

          • Roxxors
            Member
            • Mar 2025
            • 57

            #6
            Would help a ton if Memtest displayed the correct CPU speed (poll it like the temps) and a CPU utilization percent. I know the voltage, it's fixed in the BIOS, but what I don't know when an error occurs is the current CPU speed and the CPU core utilization to match to a heat curve, so I have to guess. I am pretty sure that the CPU speed/utilization to temperature is not linear, but without that info I am stuck plotting a curve based on a guess. Under Windows I have all this info, but under Windows the Temps are 30c cooler and the CPU runs at about 10% utilization, under Memtest the temps are 30c hotter and I have no idea what the CPU speed is or the core utilization, so I can't even attempt to match things accurately.

            Comment

            • David (PassMark)
              Administrator
              • Jan 2003
              • 11092

              #7
              CPU clock speed can vary up and down each millisecond. Far faster than it can be sampled and displayed on the screen. So anything displayed on the screen would only be a rough indication (not really "correct"). We are also very limited on screen space for new data as some older systems are stuck at low screen resolutions in BIOS.

              Comment

              • Roxxors
                Member
                • Mar 2025
                • 57

                #8
                Just poll the CPU speed like every 10 seconds. I know it changes, but not all that fast. I think Windows polls it and the utilization every 5, but that's a changeable setting. as are my other troubleshooting utilities.

                Comment

                Working...