Announcement

Collapse
No announcement yet.

Memory testing - How many passes is enough?

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

  • Memory testing - How many passes is enough?

    Kind of an open question I know. But how many passes with memtest86+ is generally considered to be enough to show memory as good?

    Here's my situation: I have F3-8500CL7Q-16GBRL on a M4A785TD-M EVO. All hardware purchased new in Jan 2011.

    I run ESXi on this box and had at least two virtual servers running, sometimes three. Never exceeded the total physical memory. About 2 weeks ago I started crashing and having problems booting. Also noticed that my BIOS settings were not "sticking", i.e., I would set the system to boot from USB, but it wouldn't find it. Located and ran memtest86+ and it showed errors in the upper memory ranges.

    Replaced CMOS battery (which seemed to solve the BIOS setting sticking issue) but was still having problems booting and crashing. Ran memtest86+, again showing errors but now all over the memory range.

    Folks at support suggested I test each stick individually, which I have been doing since yesterday. I let one stick run 10 passes for almost 9 hours. Another stick I let run for 5 passes for just over 4 hours. 3rd stick has now been testing for almost 4 hours.

    So as the topic asks, "Am I letting these run diags log enough?" If they make it through 1 pass, is that enough to declare the stick good? I am not moving sticks around in the mb slots. In other words, I found one slot that would let me boot and that's the slot all the testing is being done with.

    BTW, had to change CAS to 7-7-7-18 which seemed to make a big difference. Have not changed voltage--yet. Should I once I get all four sticks back in? and if so, by how much?

    Sorry for the long post and TIA for any help.

    Bill W

  • #2
    Do you have the latest BIOS for the motherboard?

    Make sure memory settings are DDR3-1066 7-7-7-18 1.50V.

    It should work flawlessly.

    Thank you
    GSKILL TECH

    Comment


    • #3
      Oh yeah, one of the first things I did.

      So, how many passes of memtest86+ should I do on each stick (just to be sure)?

      Bill W

      Comment


      • #4
        It's just up to you. Typically a couple passes is sufficient, then you can go on to a stress test. Stress test will test the memory itself and settings you input in BIOS, so if there is a problem, it will be even more apparent. Stress test puts the system into a live situation @ 100%, whereas the memory test only tests for errors within the memory modules (chips).

        Thank you
        GSKILL TECH

        Comment


        • #5
          That makes sense. What doesn't make sense is after successfully testing each individual stick for 4 or more passes and running for 3-4 hours each, why I would get a failure after inserting all memory back in the mb!

          Now it only happened on the first pass and subsequent passes did not generate any more errors--so far. I think I'll let it run overnight and see what happens.

          I really hope it turns out to be a fluke because I have been very happy with the price/performance of this memory. At least until recently 8-)

          Bill W

          Comment


          • #6
            That's one reason to test individually, that basically tells you the sticks are good, when tested with two or more you can get errors from improper voltages, normally the DRAM or MC (Memory controller) voltage. Might try increasing one or both, just a bit at a time ( + .01 or + .02) or I'll often pop it + .05 each and if stable take it down a bit at a time


            Pls offer comments on support I provide, HERE, in order to help me do a better job here:

            Tman

            Comment


            • #7
              I had a sneaky suspicion that might be the case, i.e., all the sticks at once drawing more power.

              Ok, I'll bump voltage (if I can figure out how to do that with this BIOS) and see what happens.

              Bill W

              PS: Why can't bios makers all agree on what to call the various settings? 8-)

              Comment


              • #8
                Forgot to mention advanced timings for multiple sticks might be off also, which is determined by the BIOS in some cases.....as to what the various voltage are call....you got me...used to be pretty stand, Intel had one set and AMD another set of names then came the 1366 that took a different approach, I've tried to cover some in the thread:

                http://www.gskill.us/forum/showthread.php?t=10646


                Pls offer comments on support I provide, HERE, in order to help me do a better job here:

                Tman

                Comment


                • #9
                  what would be the max voltage I should use for this memory? I've bumped it one "notch" already and the memory errors almost tripled. Should I try even more voltage, go down in voltage, sacrifice a rooster or what?

                  BTW, I was just venting about the BIOS naming conventions. I know Gskill has nothing to do with that.

                  Bill W

                  Comment


                  • #10
                    What are your voltages currently DRAM, vCore and CPU/NB? Also what are the advanced timings, in particular the tRFC and what is CR (Command Rate (might be Command timing or something like that at - might be at 1N (or possibly 1T) else should be 2N or 2T ( the N and T designators are another gripe of mine, used to be Intel used T and AMD used N, now it seems interchangeable)


                    Pls offer comments on support I provide, HERE, in order to help me do a better job here:

                    Tman

                    Comment


                    • #11
                      Hmmmm. Don't know all of those. DRAM is currently 1.515 (just brought that down from 1.53). THAT voltage seems to be working so far. I think I'll let memtest finish a pass and see what happens, then look up those other settings. May be later today before I can report back.

                      Thanks for the help. Not used to all these settings. Used to be just slap it in and go (whoa, beginning to sound like my dad!)

                      Bill W

                      Comment


                      • #12
                        Might jump CPU/NB + .05, that's the MC voltage and w/ 4 sticks it eases the strain on the MC


                        Pls offer comments on support I provide, HERE, in order to help me do a better job here:

                        Tman

                        Comment


                        • #13
                          Man this is frustrating and aggravating!

                          So I run my system for 4-5 hours with the settings (mis-spoke about the voltage earlier) below. Get 4 error-free passes from Memtest86+. Alright, I got this figured out finally. So I reboot to note the voltages that Tradesman asked about. Not sure where to look in this bios so I let the system boot. BAM! Errors faster than I can count. All over the place.

                          I have no idea what to do next? Does this mean MB problems? Do I RMA the 4 sticks? Keep poking voltages? Help me Obi-Wan?

                          Current Settings under the screen Advanced -> JumperFree Configuration

                          CPU Overclocking - auto
                          GPU Overclocking - auto
                          PCIE Overclocking - auto
                          Processor Frequency Multiplier - auto
                          CPU/NB Frequency - auto
                          CPU Over Voltage - auto
                          VDDNB Over Voltage - auto
                          LoadLine Calibration - auto
                          HT Link Frequency - auto
                          HT Link Width - auto
                          HT Over Voltage - auto
                          Memclock Value - 533MHz
                          DRAM Timing Mode - DCT 0
                          TCL - 7 CLK
                          TRCD - 7 CLK
                          TRP - 7 CLK
                          TRTP - 7 CLK
                          TRAS - 18 CLK
                          TRC - AUTO
                          tWR - AUTO
                          TRRD - AUTO
                          tRWTTO - AUTO
                          tWRRD - AUTO
                          tWTR - AUTO
                          tWRWR - AUTO
                          tRDRD - AUTO
                          tRFC0 - AUTO
                          tRFC1 - AUTO
                          tRFC2 - AUTO
                          tRFC3 - AUTO
                          Memory OverVoltage - 1.53000

                          Comment


                          • #14
                            VDDNB Voltage is on auto, try raising that a couple notches (in effect that's the MC voltage on your mobo or the CPU/NB that I referred to earlier


                            Pls offer comments on support I provide, HERE, in order to help me do a better job here:

                            Tman

                            Comment


                            • #15
                              How about one module at a time, any problems?

                              Thank you
                              GSKILL TECH

                              Comment

                              Working...
                              X