Announcement

Collapse
No announcement yet.

Errors in test 6 or 9 after cold boot (ie. memory training)

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

  • Errors in test 6 or 9 after cold boot (ie. memory training)

    Hi. Just a heads up, but you probably know this already:
    because memory training happens on
    a) cold boot (ie. poweroff/shutdown, then turn on PC from power button)
    or
    b) on warm boots(ie. Restart from within Windows) if MRC Fast Boot BIOS settingis Disabled (or, if on Auto but you just Save&Exit-ed from BIOS)
    then MemTest86 v8.4(at least), sometimes, correctly(I think), shows errors in test 6 and/or 9 because the motherboard memory trained the DRAM wrongly(it seems to me)
    ie.
    Click image for larger version

Name:	booboo.jpg
Views:	1957
Size:	141.7 KB
ID:	48859

    The problem is that apparently memory training seems dependent on DRAM temperature thus it just trains badly when temperature is higher than a certain threshold(ie. if I've just ran 2 passes of just test 6 and 9 before turning off then on again the PC). The other possibility, which is worse, is that it doesn't depend on temperature and it just randomly trains badly.

    I'm noticing this mainly due to my overclocking this RAM. (but I haven't tried to repro. this when RAM is not overclocked)

    So, without changing any BIOS settings, after stress testing it in Windows via Prime95(multiple tests of 40mins each), TM5 extreme1 profile (6+ hours), RealBench(a few runs), all passing ok, rebooting(warm boot, ie. Windows Restart), booted in Memtest86 (from USB stick) it failed only on test 9 (so all prior tests including 6 passed, on first pass - didn't try more passes). This was the first time I noticed the issue.

    Now I can simply press power button to turn off PC while Memtest86 just passed test 6 & 9 two times (two passes of just test 6 and 9 selected) which means the RAM is like 38-40 Celsius (if I just reboot into Windows I see 38 Celsius), and then I press power button to turn on PC after like 2 seconds of being off, and that Memory Training that happens during POST(before anything is shown on screen, and you can see those yellow/red and eventually white then green leds on the motherboard while training happens) can do 2 things:

    A) take like 2 minutes and fail by showing me the safe mode screen:
    Click image for larger version

Name:	safemodepost.jpg
Views:	1598
Size:	89.9 KB
ID:	48860
    or
    B) memory train just fine(even if it takes 20 seconds or 1 minute before showing successful POST screen)

    B usually happens only after: A happened once and then I do two more off-then-on the PC, probably until RAM cooled off some more. Unless it's just random, which is worse, imho.

    But when B) happens, two things can happen:
    a) it will pass memtest86 test 6 and 9 without any errors, and thus will also pass all Memtest86 tests if I run them afterwards!
    b) it will error on test 6 (in which case I usually don't let it try test 9 too)

    Remember, this happens without even bothering to enter BIOS, so no settings are changed! Some settings are still left on Auto though.

    The point is: memory training cannot be trusted it really seems(to me) to depend on DRAM temperature, if it's say 38+ Celsius it trains badly and memtest86 test 6 and/or 9 will error. If it's below 38 (let's say, guessing though), then it trains good and it won't error in those tests, but as soon as I power off then on again it will train badly and if it does POST it's usually that test 6 errors again.

    I'm on Asus Prime Z370-A motherboard (not the II variant), using Intel Core i7-8700K CPU at 4.7Ghz(when turbo), and using 4 sticks of 16GB RAM each made of two RAM kits: one is GTZ and one is GTZKW , here's one kit's name: G.Skill F4-3200C14D-32GTZ Trident Z DDR4-3200MHz CL14-14-14-34 1.35V 32GB (2x16GB)

    Here's my BIOS .txt saved settings that I'm using:
    Code:
    [2020/10/24 17:48:58]
    Ai Overclock Tuner [Manual]
    BCLK Frequency [100.0000]
    ASUS MultiCore Enhancement [Disabled]
    SVID Behavior [Best-Case Scenario]
    AVX Instruction Core Ratio Negative Offset [3]
    CPU Core Ratio [Sync All Cores]
    1-Core Ratio Limit [47]
    2-Core Ratio Limit [47]
    3-Core Ratio Limit [47]
    4-Core Ratio Limit [47]
    5-Core Ratio Limit [47]
    6-Core Ratio Limit [47]
    BCLK Frequency : DRAM Frequency Ratio [100:133]
    DRAM Odd Ratio Mode [Enabled]
    DRAM Frequency [DDR4-3600MHz]
    TPU [Keep Current Settings]
    Power-saving & Performance Mode [Performance mode]
    CPU SVID Support [Enabled]
    CPU Core/Cache Current Limit Max. [171.00]
    CPU Graphics Current Limit [45.00]
    Ring Down Bin [Enabled]
    Min. CPU Cache Ratio [8]
    Max CPU Cache Ratio [44]
    Max. CPU Graphics Ratio [24]
    Extreme Over-voltage [Disabled]
    BCLK Aware Adaptive Voltage [Enabled]
    CPU Core/Cache Voltage [Offset Mode]
    - Offset Mode Sign [+]
    - CPU Core Voltage Offset [0.050]
    DRAM Voltage [1.3794]
    CPU VCCIO Voltage [1.15000]
    CPU System Agent Voltage [1.15000]
    CPU Graphics Voltage [Offset Mode]
    - Offset Mode Sign [-]
    - CPU Graphics Voltage Offset [0.120]
    PCH Core Voltage [Auto]
    CPU Standby Voltage [Auto]
    DRAM CTRL REF Voltage on CHA [0.50000]
    DRAM CTRL REF Voltage on CHB [0.50000]
    DRAM DATA REF Voltage on CHA DIMM0 Rank0 BL0 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank0 BL1 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank0 BL2 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank0 BL3 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank0 BL4 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank0 BL5 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank0 BL6 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank0 BL7 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank1 BL0 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank1 BL1 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank1 BL2 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank1 BL3 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank1 BL4 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank1 BL5 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank1 BL6 [Auto]
    DRAM DATA REF Voltage on CHA DIMM0 Rank1 BL7 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank0 BL0 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank0 BL1 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank0 BL2 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank0 BL3 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank0 BL4 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank0 BL5 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank0 BL6 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank0 BL7 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank1 BL0 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank1 BL1 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank1 BL2 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank1 BL3 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank1 BL4 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank1 BL5 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank1 BL6 [Auto]
    DRAM DATA REF Voltage on CHA DIMM1 Rank1 BL7 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank0 BL0 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank0 BL1 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank0 BL2 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank0 BL3 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank0 BL4 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank0 BL5 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank0 BL6 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank0 BL7 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank1 BL0 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank1 BL1 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank1 BL2 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank1 BL3 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank1 BL4 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank1 BL5 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank1 BL6 [Auto]
    DRAM DATA REF Voltage on CHB DIMM0 Rank1 BL7 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank0 BL0 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank0 BL1 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank0 BL2 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank0 BL3 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank0 BL4 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank0 BL5 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank0 BL6 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank0 BL7 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank1 BL0 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank1 BL1 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank1 BL2 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank1 BL3 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank1 BL4 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank1 BL5 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank1 BL6 [Auto]
    DRAM DATA REF Voltage on CHB DIMM1 Rank1 BL7 [Auto]
    DRAM CAS# Latency [15]
    DRAM RAS# to CAS# Delay [16]
    DRAM RAS# ACT Time [33]
    DRAM Command Rate [2N]
    DRAM RAS# to RAS# Delay L [4]
    DRAM RAS# to RAS# Delay S [4]
    DRAM REF Cycle Time [288]
    DRAM Refresh Interval [65024]
    DRAM WRITE Recovery Time [10]
    DRAM READ to PRE Time [8]
    DRAM FOUR ACT WIN Time [16]
    DRAM WRITE to READ Delay [1]
    DRAM WRITE to READ Delay L [8]
    DRAM WRITE to READ Delay S [4]
    DRAM CKE Minimum Pulse Width [7]
    DRAM Write Latency [13]
    tRDRD_sg [6]
    tRDRD_dg [4]
    tRDWR_sg [10]
    tRDWR_dg [10]
    tWRWR_sg [6]
    tWRWR_dg [4]
    tWRRD_sg [27]
    tWRRD_dg [23]
    tRDRD_dr [6]
    tRDRD_dd [7]
    tRDWR_dr [10]
    tRDWR_dd [10]
    tWRWR_dr [7]
    tWRWR_dd [8]
    tWRRD_dr [6]
    tWRRD_dd [6]
    TWRPRE [27]
    TRDPRE [8]
    tREFIX9 [127]
    OREF_RI [64]
    MRC Fast Boot [Auto]
    DRAM CLK Period [20]
    Memory Scrambler [Enabled]
    Channel A DIMM Control [Enable both DIMMs]
    Channel B DIMM Control [Enable both DIMMs]
    MCH Full Check [Enabled]
    Training Profile [Standard Profile]
    DLLBwEn [2]
    DRAM SPD Write [Disabled]
    DRAM RTL INIT value [67]
    DRAM RTL (CHA DIMM0 Rank0) [58]
    DRAM RTL (CHA DIMM0 Rank1) [58]
    DRAM RTL (CHA DIMM1 Rank0) [58]
    DRAM RTL (CHA DIMM1 Rank1) [58]
    DRAM RTL (CHB DIMM0 Rank0) [58]
    DRAM RTL (CHB DIMM0 Rank1) [58]
    DRAM RTL (CHB DIMM1 Rank0) [58]
    DRAM RTL (CHB DIMM1 Rank1) [58]
    DRAM IOL (CHA DIMM0 Rank0) [14]
    DRAM IOL (CHA DIMM0 Rank1) [13]
    DRAM IOL (CHA DIMM1 Rank0) [13]
    DRAM IOL (CHA DIMM1 Rank1) [13]
    DRAM IOL (CHB DIMM0 Rank0) [14]
    DRAM IOL (CHB DIMM0 Rank1) [14]
    DRAM IOL (CHB DIMM1 Rank0) [14]
    DRAM IOL (CHB DIMM1 Rank1) [14]
    CHA IO_Latency_offset [14]
    CHB IO_Latency_offset [14]
    CHA RFR delay [14]
    CHB RFR delay [14]
    ODT RTT WR (CHA) [80 DRAM Clock]
    ODT RTT PARK (CHA) [80 DRAM Clock]
    ODT RTT NOM (CHA) [40 DRAM Clock]
    ODT RTT WR (CHB) [80 DRAM Clock]
    ODT RTT PARK (CHB) [60 DRAM Clock]
    ODT RTT NOM (CHB) [40 DRAM Clock]
    ODT_READ_DURATION [2]
    ODT_READ_DELAY [0]
    ODT_WRITE_DURATION [1]
    ODT_WRITE_DELAY [0]
    Data Rising Slope [Auto]
    Data Rising Slope Offset [Auto]
    Cmd Rising Slope [Auto]
    Cmd Rising Slope Offset [Auto]
    Ctl Rising Slope [Auto]
    Ctl Rising Slope Offset [Auto]
    Clk Rising Slope [Auto]
    Clk Rising Slope Offset [Auto]
    Data Falling Slope [Auto]
    Data Falling Slope Offset [Auto]
    Cmd Falling Slope [Auto]
    Cmd Falling Slope Offset [Auto]
    Ctl Falling Slope [Auto]
    Ctl Falling Slope Offset [Auto]
    Clk Falling Slope [Auto]
    Clk Falling Slope Offset [Auto]
    CPU Load-line Calibration [Level 2]
    CPU Current Capability [140%]
    CPU VRM Switching Frequency [Auto]
    VRM Spread Spectrum [Disabled]
    CPU Power Duty Control [T.Probe]
    CPU Power Phase Control [Extreme]
    CPU VRM Thermal Control [Enabled]
    CPU Graphics Load-line Calibration [Level 1]
    CPU Graphics Current Capability [100%]
    CPU Graphics VRM Switching Frequency [Auto]
    CPU Graphics Power Phase Control [Standard]
    Intel(R) SpeedStep(tm) [Enabled]
    Turbo Mode [Enabled]
    Long Duration Package Power Limit [150]
    Package Power Time Window [2]
    Short Duration Package Power Limit [154]
    IA AC Load Line [0.01]
    IA DC Load Line [0.01]
    TVB Voltage Optimizations [Disabled]
    PCI Express Native Power Management [Enabled]
    Native ASPM [Enabled]
    PCH DMI ASPM [Enabled]
    ASPM [Auto]
    L1 Substates [L1.1 & L1.2]
    PCI Express Clock Gating [Enabled]
    DMI Link ASPM Control [L1]
    PEG - ASPM [Auto]
    Hyper-Threading [Enabled]
    Active Processor Cores [All]
    Intel Virtualization Technology [Enabled]
    Hardware Prefetcher [Enabled]
    Adjacent Cache Line Prefetch [Enabled]
    SW Guard Extensions (SGX) [Disabled]
    Tcc Offset Time Window [3 sec]
    Intel(R) SpeedStep(tm) [Enabled]
    Turbo Mode [Enabled]
    CPU C-states [Enabled]
    Enhanced C-states [Enabled]
    CPU C3 Report [Enabled]
    CPU C6 Report [Enabled]
    CPU C7 Report [CPU C7s]
    CPU C8 Report [Enabled]
    Package C State Limit [Auto]
    CFG Lock [Disabled]
    Intel(R) Speed Shift Technology [Enabled]
    VT-d [Enabled]
    Above 4G Decoding [Enabled]
    Primary Display [CPU Graphics]
    iGPU Multi-Monitor [Disabled]
    RC6(Render Standby) [Auto]
    DVMT Pre-Allocated [64M]
    DMI Max Link Speed [Auto]
    PCIEX16_1 Link Speed [Auto]
    PCIEX16_2 Link Speed [Auto]
    IOAPIC 24-119 Entries [Enabled]
    PCIe Speed [Auto]
    SATA Controller(s) [Enabled]
    SATA Mode Selection [AHCI]
    Aggressive LPM Support [Disabled]
    SMART Self Test [Enabled]
    SATA6G_1(Charcoal Black) [Disabled]
    SATA6G_1 Hot Plug [Disabled]
    SATA6G_2(Charcoal Black) [Enabled]
    SATA6G_2 Hot Plug [Disabled]
    SATA6G_3(Charcoal Black) [Disabled]
    SATA6G_3 Hot Plug [Disabled]
    SATA6G_4(Charcoal Black) [Disabled]
    SATA6G_4 Hot Plug [Disabled]
    SATA6G_5(Charcoal Black) [Disabled]
    SATA6G_5 Hot Plug [Disabled]
    SATA6G_6(Charcoal Black) [Disabled]
    SATA6G_6 Hot Plug [Disabled]
    TPM Device Selection [Discrete TPM]
    VccCore Override Enable [Disabled]
    VccST Override Enable [Disabled]
    VccSA Override Enable [Disabled]
    VccSFR_OC Override Enable [Disabled]
    VccSFR Override Enable [Disabled]
    VccIO Override Enable [Disabled]
    Core PLL Voltage Offset [0]
    GT PLL Voltage Offset [0]
    Ring PLL Voltage Offset [0]
    System Agent PLL Voltage Offset [0]
    Memory Controller PLL Voltage Offset [0]
    Thunderbolt(TM) Support [Disabled]
    Thunderbolt(TM) PCIe Support [Disabled]
    Windows 10 Thunderbolt support [Disabled]
    Security Device Support [Enable]
    SHA-1 PCR Bank [Enabled]
    SHA256 PCR Bank [Enabled]
    SM3_256 PCR Bank [Disabled]
    Pending operation [None]
    Platform Hierarchy [Enabled]
    Storage Hierarchy [Enabled]
    Endorsement Hierarchy [Enabled]
    TPM2.0 UEFI Spec Version [TCG_2]
    Physical Presence Spec Version [1.3]
    ErP Ready [Disabled]
    Restore AC Power Loss [Power Off]
    Power On By PCI-E [Disabled]
    Power On By Ring [Disabled]
    Power On By RTC [Disabled]
    Hyper M.2X16 [Disabled]
    HD Audio Controller [Enabled]
    DVI Port Audio [Disabled]
    Depop [Enabled]
    M.2_1 Configuration [PCIE Mode]
    M.2_2 PCIe Bandwidth Configuration: [X2][X4] [X2 Mode]
    Asmedia USB 3.1 Controller [Enabled]
    When system is in working state [Off]
    When system is in sleep, hibernate or soft off states [On]
    Intel LAN Controller [Enabled]
    Intel PXE Option ROM [Disabled]
    USB Type C Power Switch [Auto]
    Serial Port 1 [Disabled]
    Network Stack [Disabled]
    Legacy USB Support [Enabled]
    USB Keyboard and Mouse Simulator [Disabled]
    Generic STORAGE DEVICE 9404 [Auto]
    Generic Storage Device 0.00 [Auto]
    USB3G1_1 [Enabled]
    USB3G1_2 [Enabled]
    USB3G1_3 [Enabled]
    USB3G1_4 [Enabled]
    USB3G1_5 [Enabled]
    USB3G1_6 [Enabled]
    USB9 [Enabled]
    USB10 [Enabled]
    USB11 [Enabled]
    USB12 [Enabled]
    USB13 [Enabled]
    USB14 [Enabled]
    CPU Temperature [Monitor]
    MotherBoard Temperature [Monitor]
    PCH Temperature [Monitor]
    T_Sensor Temperature [Monitor]
    EXT_Sensor1  Temperature [Monitor]
    EXT_Sensor2  Temperature [Monitor]
    EXT_Sensor3  Temperature [Monitor]
    CPU Fan Speed [Monitor]
    CPU Optional Fan Speed [Monitor]
    Chassis Fan 1 Speed [Monitor]
    M.2 Fan Speed [Monitor]
    Chassis Fan 2 Speed [Monitor]
    Extension Fan 1 Speed [Monitor]
    Extension Fan 2 Speed [Monitor]
    Extension Fan 3 Speed [Monitor]
    AIO PUMP Speed [Monitor]
    CPU Core Voltage [Monitor]
    CPU Graphics Voltage [Monitor]
    3.3V Voltage [Monitor]
    5V Voltage [Monitor]
    12V Voltage [Monitor]
    PCH Core Voltage [Monitor]
    CPU System Agent Voltage [Monitor]
    CPU VCCIO Voltage [Monitor]
    DRAM Voltage [Monitor]
    CPU Standby Voltage [Monitor]
    DMI Voltage [Monitor]
    Chassis Intrusion Detection Support [Disabled]
    CPU Q-Fan Control [DC Mode]
    CPU Fan Speed Lower Limit [200 RPM]
    CPU Fan Profile [Manual]
    CPU Upper Temperature [50]
    CPU Fan Max. Duty Cycle (%) [100]
    CPU Middle Temperature [45]
    CPU Fan Middle. Duty Cycle (%) [60]
    CPU Lower Temperature [36]
    CPU Fan Min. Duty Cycle (%) [47]
    AIO PUMP Control [Disabled]
    Chassis Fan 1 Q-Fan Control [DC Mode]
    Chassis Fan 1 Q-Fan Source [Multiple Sources]
    Temperature Source 1 [CPU]
    Temperature Source 2 [MotherBoard]
    Temperature Source 3 [PCH]
    Chassis Fan 1 Speed Low Limit [200 RPM]
    Chassis Fan 1 Profile [Manual]
    Chassis Fan 1 Upper Temperature [61]
    Chassis Fan 1 Max. Duty Cycle (%) [100]
    Chassis Fan 1 Middle Temperature [61]
    Chassis Fan 1 Middle. Duty Cycle (%) [80]
    Chassis Fan 1 Lower Temperature [40]
    Chassis Fan 1 Min. Duty Cycle (%) [30]
    Allow Fan Stop [Enabled]
    M.2 Fan Q-Fan Control [DC Mode]
    M.2 Fan Q-Fan Source [Multiple Sources]
    Temperature Source 1 [MotherBoard]
    Temperature Source 2 [PCH]
    Temperature Source 3 [CPU]
    M.2 Fan Speed Low Limit [200 RPM]
    M.2 Fan Profile [Manual]
    M.2 Fan Upper Temperature [55]
    M.2 Fan Max. Duty Cycle (%) [100]
    Chassis Fan 2 Middle Temperature [50]
    Chassis Fan 2 Middle. Duty Cycle (%) [80]
    M.2 Fan Lower Temperature [40]
    M.2 Fan Min. Duty Cycle (%) [30]
    Allow Fan Stop [Enabled]
    Chassis Fan 2 Q-Fan Control [DC Mode]
    Chassis Fan 2 Q-Fan Source [Multiple Sources]
    Temperature Source 1 [CPU]
    Temperature Source 2 [MotherBoard]
    Temperature Source 3 [PCH]
    Chassis Fan 2 Speed Low Limit [200 RPM]
    Chassis Fan 2 Profile [Manual]
    Chassis Fan 2 Upper Temperature [55]
    Chassis Fan 2 Max. Duty Cycle (%) [100]
    Chassis Fan 2 Middle Temperature [50]
    Chassis Fan 2 Middle. Duty Cycle (%) [80]
    Chassis Fan 2 Lower Temperature [40]
    Chassis Fan 2 Min. Duty Cycle (%) [30]
    Allow Fan Stop [Enabled]
    Extension Fan 1 Q-Fan Control [Disabled]
    Extension Fan 2 Q-Fan Control [Disabled]
    Extension Fan 3 Q-Fan Control [Disabled]
    Fast Boot [Disabled]
    Boot Logo Display [Disabled]
    POST Report [5 sec]
    Boot up NumLock State [Disabled]
    Wait For 'F1' If Error [Enabled]
    Option ROM Messages [Force BIOS]
    Interrupt 19 Capture [Disabled]
    Setup Mode [EZ Mode]
    Launch CSM [Enabled]
    Boot Device Control [UEFI and Legacy OPROM]
    Boot from Network Devices [Ignore]
    Boot from Storage Devices [UEFI driver first]
    Boot from PCI-E Expansion Devices [Legacy only]
    OS Type [Other OS]
    Setup Animator [Disabled]
    Load from Profile [1]
    Profile Name [23oct3600]
    Save to Profile [1]

  • #2
    If you are overclocking your RAM, then getting errors isn't that unusual.
    We don't know if in your case those errors are due to sub-optimal memory training or just that the RAM is bad (or was never designed to run at that speed). The training process is out of our control in any case.

    Comment


    • #3
      Thanks for your reply. I'm not good at explaining things. I guess that was just my try of saying, if memtest86 fails test 6 or 9, it may not fail again after a cold boot.

      Currently, I'm in a memory trained session such that memtest86 will give only 2 errors(both in one of the passes) in test 6 after all 4 passes are finished. I can warm boot (reboot, without memory training) and enter Windows, run some TM5 extreme1 profile or some Prime95 and they won't error, at least not yet anyway(maybe I need more hours to test, this time too). Then warm boot into memtest86 again and still get 2 errors, not the same addresses though, and possibly the same bit, but always just two errors (in this memtrained session), after 4 passes of test 6. But if I turn off PC then on again, I'll have a newly memtrained session, and I doubt I can get the same amount of errors again.

      In other memtrained session(s) I get no errors in memtest86 test 6 or 9.
      In other memtrained session(s) I get 50+ errors in test 6.
      But they all only ever change after a memory train.

      I'm currently trying to see if I can get any Prime95, TM5, GSAT to give me an error. After that I'll just warm boot again(to avoid any memtraining) and enter memtest86 to reconfirm that indeed test 6 will give me 2 errors after all 4 passes are done.
      It's kinda curious that none of the Windows programs were able to find anything, the first time I discovered that only test 9 would error afterwards(ie. rebooted from Windows into memtest86), and yet memtest86 does.

      I had posted more screens here.

      While I'm convinced that it's the memtraining, it's possible that it could've been a coincidence even when test 6&9 both passed 4 times and that maybe another run of 4 passes would show errors. I will try more passes next time when I re-test a memtrained session that gives no errors.


      Do you by any chance know some values or range of values for any of the following(from BIOS) because I'd like to move them out of Auto presumably to bypass memory training assigning the wrong values(but I'm sure there are hidden values which memtraining will still affect):
      Code:
      Data Rising Slope [Auto]
      Data Rising Slope Offset [Auto]
      Cmd Rising Slope [Auto]
      Cmd Rising Slope Offset [Auto]
      Ctl Rising Slope [Auto]
      Ctl Rising Slope Offset [Auto]
      Clk Rising Slope [Auto]
      Clk Rising Slope Offset [Auto]
      Data Falling Slope [Auto]
      Data Falling Slope Offset [Auto]
      Cmd Falling Slope [Auto]
      Cmd Falling Slope Offset [Auto]
      Ctl Falling Slope [Auto]
      Ctl Falling Slope Offset [Auto]
      Clk Falling Slope [Auto]
      Clk Falling Slope Offset [Auto]
      I mean, hey I figure if it memtest passes and then only after the next memtrain it doesn't, it can't really be something wrong with the RAM.
      Last edited by hackasynthetic; Oct-24-2020, 11:50 PM. Reason: actually it may be the same bit, it's the same byte at least

      Comment


      • #4
        It passed Prime95 Large dataset for 1hr 2mins, then restarted into memtest86 and this time I got more than 2 errors:
        Click image for larger version  Name:	_IMG_20201025_023022.jpg Views:	0 Size:	135.7 KB ID:	48873
        EDIT: no errors when trying only 1 CPU (cpu0), all 4 passes of test 6 were ok.

        MemTest86 still better at detecting memory errors. Odd because for example TM5 would detect errors pretty fast when memtest wouldn't at all. I'll do TM5 longer than 22mins soon.

        Last edited by hackasynthetic; Oct-25-2020, 01:22 AM.

        Comment


        • #5
          Do you by any chance know some values or range of values for any of the following(from BIOS) because I'd like to move them out of Auto
          Sorry no. We've never gone down the rabbit hole that deep.

          There is a known issues with a Gigabyte board that gives different results after a cold boot, compared to a warm boot after loading Windows. See,
          https://forums.passmark.com/memtest8...t-on-cold-boot
          We didn't attribute that to training issues however, but never really got to the bottom of the issue either.

          Comment


          • #6
            I've seen that one before, thanks. My issue is a bit different. For example, I cold boot straight into memtest86 and get 2 errors in one of the four passes of test 6. Then I can warm boot into Windows and throw virtually-any stress test at it and they will always pass ok. Then warm boot into memtest86 again and get the 2 errors (or sometimes 6 as seen above).

            I'm convinced it's just because of how it memory trained. On next cold boot it can actually work just fine and pass 4 passes of test 6 twice, and two full passes of memtest86 (except test 13 hammering, which I've disabled).

            I've actually tested the above two cases with:
            Data Rising Slope 8
            Data Rising Slope Offset 1

            but all the rest on Auto for now. If I just blindly put them all to 8-1 or 7-1 it will simply just lock up at the white led stage, or sometimes when no led is on, even before POST screen is shown, sometimes it can show POST screen.

            Made some screens and showed BIOS settings here
            Last edited by hackasynthetic; Oct-25-2020, 09:51 PM. Reason: fixed the text formatting a bit (copy/pasting messed it up and it was all on one long non-wrapping line)

            Comment


            • #7
              I've never done it (play with these particular very low level settings). But I would have assumed that the correct way to do it would be to have oscilloscope connected to the memory bus. Then make fine adjustments until the waveform meets the specs.


              Click image for larger version

Name:	image20520ddr.png
Views:	1183
Size:	111.2 KB
ID:	48883

              Comment


              • #8
                Nice, wish I had that capability to use an oscilloscope like that!

                Well, there must be some weird BIOS bug, or hidden BIOS settings or something, because now with the exact same BIOS settings that would err-fast before, now it just wouldn't err nearly as fast anymore: after 45 (11x4+1) passes of test6 I've only gotten 1 pass to give 2 errs! No amount of turn off/on would make it less stable than this, it's now far more stable than before, without any apparent BIOS settings changes. It's true that I've tried more settings since the OP, eg. tried 4000Mhz with some DLLBwEn and DRAM CLK Period settings and other voltage settings, but I've since reverted the settings to what they were before (not by loading the prev. BIOS config though, but just by setting the settings manually, and then saving them as .txt then comparing to prev. using total commander to see any diffs - no diffs! but maybe not all settings are saved as .txt, still I'm convinced that all settings related to memory were as they were before)

                I'm inclined to distrust this mobo/BIOS because of one more issue I've encountered like 3 times: sometimes during AIDA64 Benchmarks(all of them) report generation(which runs them all), it would shut off (turn off) my PC (then back again, automatically). I would try to change BIOS settings left and right and would eventually not shutoff anymore, however, setting the BIOS settings as they were before(manually though), would not replicate the turnoff issue. I doubt it was the power supply though, Corsair HX850i (and I don't have a graphics card), so the most it would use would be less than 260W (out of 850W that power supply can give). One of the times, I just didn't even bother changing anything in BIOS, just did Benchmarks after it turned itself back on, and surprise surprise it eventually stopped auto-turning off after a few times. It's almost as if some Auto setting set itself to a higher value (I don't know maybe one of those with 140% CPU current for example(CPU Current Capability [140%]), which I've now manually set to that value - no issues since this was set). But this just makes me wonder...

                I've since, Loaded Optimized Defaults, rebooted, re-loaded my config, still very memory stable in test 6. It's still very hard to even get the 2 errors! I can't help but think there are some hidden settings which got saved in the .CMO file but which I cannot set through BIOS GUI and those are the ones having this effect on stability. For example, I know of two settings which GUI doesn't expose, but I don't know if they are saved in .CMO either, they are: tXP=8 and PPD=1 and one with tAON__ something which is sometimes 9 sometimes 10 (seen in Mem TweakIt). So I'm thinking settings like these, or maybe other hidden ones that stick better, got pushed to other values during my attempts to use different settings (like the DDR 4000 Mhz one, which didn't POST btw), and because those settings are now different, when switching back to my prev. DDR 3600 settings, manually, the whole system is now more stable. Maybe these hidden settings aren't even stored in BIOS/CMOS, even though I've also unplugged PC several times, but hmm, I haven't cleared CMOS yet... hmm...

                I'll let you know here if I do eventually figure out what's going on... I'm just a lil baffled about this behaviour. Also I'm sure it's not a memtest86 issue.

                Comment


                • #9
                  So it turns out, that even when I thought it was stable, it wasn't really stable, as further passes(like 40+) of test 6 would eventually reveal 2 errors(only 2, and so late, because Memory Scrambler being Enabled means it makes things more stable and less number of errors will show up too).
                  So it wasn't really stable when I thought it was stable, but instead it was just more stable than in other the other boot sessions. A boot session meaning: a cold boot(ie. when memory training (re)sets some (hidden?) memory settings; at this point I'm not sure if there are some hidden memory settings or it's just simply those visible DRAM REF BL0 to BL7 settings which get set differently on each cold boot due to memory training - it's practically impossible to set those manually and no get tons of errors, apparently, tho I've only tested them at 3600Mhz iirc, so they could be the culprit messing with stability, but even so, I can clearly say it can't get it stable at this freq of 3600 or 3500 even!).

                  So the solution is apparently to just lower the DRAM Frequency. I'm now down to 3400Mhz. It was 3600Mhz and with Memory Scrambler on Disabled(because I do want to catch the errors sooner rather than later, while testing on test 6) I was able to get(recently anyway) 12 errors in first or second pass of test 6, all the time. But with 3500Mhz, the errors were delayed to like after pass 50 or something like that. And fiddling with the timings, eg. when at 3600Mhz, only delayed the errors. So I've noticed that timings won't get rid of the errors, neither for 3600, nor for 3500Mhz. So at 3400Mhz now... much time is needed to test.

                  Maybe could a test 6 mod be done where the block size (of 64 bytes) be random ? I think it may detect the errors sooner. I'd be willing to test it, if I'm still here

                  Well, anyway, if I have any new developments or solution(s), I'll let us know.

                  Comment

                  Working...
                  X