Totally confused about my RAM

Discussion in 'General Hardware' started by shimyns, Sep 25, 2016.

  1. shimyns

    shimyns Ancient Guru

    Messages:
    1,657
    Likes Received:
    0
    GPU:
    SAPPHIRE r9 290x Tri-X OC
    Hi Guys,
    This has got me entirely confused, I'll try to be brief.

    I've had a stable, modest OC for years, with my RAM running at their specified 1600 MHz. A year ago, my PC started having intermittent black screen issues and boot problems. I got the "detecting DRAM size" message over and over. I finally found that if I just lowered my SPD from 8 to 6 (thereby reducing my RAM from 1600 MHz to 1200 MHz but not affecting my CPU speed or anything else) I resolved the issue and it's been fine. http://forums.guru3d.com/showthread.php?t=403009

    A week ago, it started happening again. I figured my RAM is deteriorating and I just need to lower it a bit. I couldn't lower the SPD so I just reduced the BLCK from 200 to 190. This worked. But the next day, I had to lower the BLCK to 180 or I'd get the "detecting DRAM size" message again and no boot. This went on until I was at 120 BLCK after several days. I tried interchanging the 3 sticks but nothing helped.

    I ordered new RAM, which will be arriving soon. But in the meantime, I noticed in CPUz that the number of channels was listed as DUAL. I am no expert, actually didn't really know for certain but was pretty sure this should be showing up as TRIPLE.

    I removed the furthest right stick (in slot #5 of six) and it booted up fine, in DUAL channel mode, at 1600 MHz. I then took the that stick and replaced one of the others (in slots #1 and #3) with it. Each pair worked fine at 1600 MHz. Finally, I just put all three sticks in, and miraculously, the PC is stable at it's original 1600 MHz, same as over a year ago before all this started happening. I checked CPUz and the memory is listed as TRIPLE channel now.

    What in the world is going on??!! I mean, I obviously tried swapping around the RAM sticks a year ago and that didn't help. (I didn't check what CPUz said for memory channels back then). So, I find it hard to believe that this was all simply due to the RAM not seated well. They were seated and reseated. Also, why then was dropping the SPD or the BLCK resulting in successful booting, and also why did I need to keep lowering it every day in order for it to boot?

    I just don't get it. And now I have $60 RAM coming in the mail...

    Please let me know what you think! Thanks!

    [EDIT]: it is not completely stable now. Sometimes, it boots up and runs fine with all three memory modules at 1600 MHz but then it might randomly freeze, and only if I remove a memory module (it doesn't matter which physical one, but of course I need to place the remaining pair in particular slots for dual channel memory) can I get it to boot.
     
    Last edited: Sep 26, 2016
  2. thatguy91

    thatguy91 Ancient Guru

    Messages:
    6,643
    Likes Received:
    98
    GPU:
    XFX RX 480 RS 4 GB
    It's either a faulty memory module, which you can test by trying that third module in the other slots, a faulty motherboard, or a faulty memory controller (inbuilt in the CPU).
     
  3. shimyns

    shimyns Ancient Guru

    Messages:
    1,657
    Likes Received:
    0
    GPU:
    SAPPHIRE r9 290x Tri-X OC
    Thank you, but your answer makes it seem like you skimmed my text and didn't read it very thoroughly. I already explained that all pairs work and that all 3 modules work together now, at 1600 MHz. I'm trying to figure out how that's possible, considering my description of events. It's clearly not a trivial answer.
     
    Last edited: Sep 25, 2016
  4. shimyns

    shimyns Ancient Guru

    Messages:
    1,657
    Likes Received:
    0
    GPU:
    SAPPHIRE r9 290x Tri-X OC
    Even though i ran furmark without issue and a sensor test with real temp with no problems, it just randomly froze.
    I popped one memory module out and it seems fine. So, i'm beginning to get the feeling that it is having trouble running triple channel memory now and it's either a voltage issue or that memory slot is an issue.

    If it's a voltage issue, could someone suggest which voltages on this motherboard could be increased that might solve a triple channel vs dual channel memory issue?
     

  5. CrazY_Milojko

    CrazY_Milojko Ancient Guru

    Messages:
    2,062
    Likes Received:
    378
    GPU:
    Asus STRIX 1070 OC
    [​IMG]As thatguy91 just suggested most likely CPU's IMC is the problem here, probably it degraded a bit and now it needs a bit more voltage to run stable again.

    One of my rigs now running X5670 @ 4.4GGz HT is GigaByte GA-EX58-DS4, simmilar to your mobo, earlier same mobo was running one 920 C0 @ 4.01GHz HT and later it was 920 D0 also @ 4.01GHz HT.

    I'm still keep screenshots of Voltage and other important settings so that you can compare it to yours, of course every chip is different and take this settings just for reference:

    920 C0 @ 4.01GHz HT

    [​IMG]

    [​IMG]

    [​IMG]



    [​IMG]

    [​IMG]




    Now setting for 920 D0 @ 4.01GHz HT...
    [​IMG]

    [​IMG]

    [​IMG]

    [​IMG]

    [​IMG]

    Both for C0 and D0 Load-Line Calibration was Enabled to keep Vdrood under control.

    I show you mine now you show me yours :D
    [​IMG]
     
  6. shimyns

    shimyns Ancient Guru

    Messages:
    1,657
    Likes Received:
    0
    GPU:
    SAPPHIRE r9 290x Tri-X OC
    Hey there, wow thanks for the shots. Here's mine. I put the 3rd module back in and booted right up. Could freeze at any moment though obviously.. Note, these are the same settings I've been running on for 5 years. There were no issues until a year ago. Please read the first post for all the details.

    What do you think is the culprit?

    [​IMG]

    [​IMG]

    [​IMG]

    [​IMG]

    [​IMG]

    [​IMG]

    [​IMG]

    [​IMG]

    [​IMG]
     
  7. -Tj-

    -Tj- Ancient Guru

    Messages:
    17,133
    Likes Received:
    1,903
    GPU:
    Zotac GTX980Ti OC
    why XMP disabled?
    Enabled should also help by possible timing issues,
     
  8. CrazY_Milojko

    CrazY_Milojko Ancient Guru

    Messages:
    2,062
    Likes Received:
    378
    GPU:
    Asus STRIX 1070 OC
    More or less voltages are OK, but I've noticed in BIOS MCH/ICH voltage settings 4 of 5 options in your case are on Auto and it's well known fact that most mobos are pushing more voltage than needed on Auto while overclocking. Hardcore OC'ers don't like that because it produces more heat and in long term could cause more or less CPU degradation.

    First step: try to set MCH/ICH voltages manualy, like on my screenshots above. Also I've noticed you RAM voltage is manaly set to 1.6V @ DDR3 1600 but actually it's running at 1.584V, look at your 4th picture. Bump a bit RAM voltage over 1.6V (save and exit BIOS, then enter BIOS again) so that in BIOS menu MB Intelligent Tweaker for DRAM Voltage it shows 1.6V or a little bit over 1.6V.

    If that doesn't help in Advanced Voltage Setting try to bump up a bit options on list below, step or two up, one option by another (not all of them at the same time) then test and if it doesn't help raise another option on list:
    1). bump up a step or two Dynamic Vcore(DVID)
    2). I don't like this: CPU Vcore @ Normal, try to set it at value (combined with DVID) a bit over your curent Vcore shown in MB Intelligent Tweaker menu.
    3). Set Load Line Calibration on next more agressive level
    4). Tried with C1E disabled? C1E is known to cause problems if Vcore is at the egde of stabillity.
    5). CPU PLL set manually to 1.88V

    Try above and come back with results, need to go to sleep now, almost 2am here, s*#t load of work is waiting for me tomorrow.

    cheers!
     
    Last edited: Sep 26, 2016
  9. shimyns

    shimyns Ancient Guru

    Messages:
    1,657
    Likes Received:
    0
    GPU:
    SAPPHIRE r9 290x Tri-X OC
    I actually tried that now and it wouldn't boot till I disabled it. Besides, it's been like that for years without issue. Thanks for the suggestion, though.
     
  10. CrazY_Milojko

    CrazY_Milojko Ancient Guru

    Messages:
    2,062
    Likes Received:
    378
    GPU:
    Asus STRIX 1070 OC
    Forget about XMP on GigaByte X58 mobo's while OC'ing, more problems than benefits. On stock CPU speed XMP would be OK, XMP + OC and: problems, here I come.
     

  11. shimyns

    shimyns Ancient Guru

    Messages:
    1,657
    Likes Received:
    0
    GPU:
    SAPPHIRE r9 290x Tri-X OC
    Thank you very much for taking the time to go over my settings and to answer so thoroughly. I will try to work through all of your suggestions.

    One question that comes to mind, though, is which of these settings are directly related to RAM? I mean, obviously the DRAM voltage is. Note, I've tried increasing it like you suggested last year and it did not help, only reducing the speed to 1200 MHz via spd 8-->6 solved the issue. I will try again, though. Are the MCH/ICH settings related to RAM? Or to the motherboard's ability to produce enough power for RAM?
    It seems like the rest of your suggestions are related to CPU stability. I know they are all valid points! But, I just don't understand how the issue I'm experiencing could be related to CPU stability (I don't claim to be very a very knowledgeable overclocker).

    - a year ago, reducing spd from 8 to 6 and not changing anything else, solved the problem.
    -this week, reducing BLCK incrementally each time it failed to boot (and showed "detecting DAM size") temporarily solved it. I know this also reduces CPU speed, of course, but when I removed one memory module and set the BLCK back to where it's been for 5 years (200), everything seems fine; i.e., the CPU seems stable. (And right now, BTW, all three modules are in and it's running fine at 1600 MHz again but who knows for how long?)

    To me, looking at this evidence means that it's memory related, not CPU related and although making the CPU voltage adjustments you suggested make sense for a more stable overclock, I'd like to first try only things that are memory related.
     
    Last edited: Sep 26, 2016
  12. CrazY_Milojko

    CrazY_Milojko Ancient Guru

    Messages:
    2,062
    Likes Received:
    378
    GPU:
    Asus STRIX 1070 OC
    How could I forgot about this, it was late I guess, quick update before going to work...

    Try to reseat CPU! Years ago friend of mine had some stabillity isues on his OC'ed LGA775 rig. Bumping up CPU & RAM voltages, changing RAM, better cooling, new PSU.. nothing helped. Finally he managed to solved problem by reseating CPU and cleaning LGA connectors on the bottom of the CPU, problems were gone like they never existed.

    Few years later, same dude, same rig. He switched CPU from C2D to C2Q, he dropped the CPU on opened CPU socket and broke completely few pins in CPU socket :stewpid: After that mobo didn't worked in Dual Channel memory mode even with all RAM slots populated, neither with C2Q or with C2D.

    Just to be sure about the RAM.... Clean both RAM pins (Alcohol 96% or some hard pencil rubber) and RAM slots on mobo and then test every single RAM module separately in it's own RAM slot, bootable Passmark Memtest86 will be perfect.

    Gonna answer on your previous questions in your last message later this day, I'm already late at work, cheers!
     
  13. shimyns

    shimyns Ancient Guru

    Messages:
    1,657
    Likes Received:
    0
    GPU:
    SAPPHIRE r9 290x Tri-X OC
    Thanks again for your advice about the CPU. I really should do that. I haven't reseated the CPU in a few years! I've avoided doing it because my temps have always been fine but it is entirely possible that I bumped my big old noctua d15 at some point and the CPU isn't seated perfectly. I have to check if I have any arctic silver left haha.
    I'll try that when I get the chance. I'm already at work now, too!

    ps- I can check the RAM one by one but only in slots#1 or #3, so I can't check the integrity of slot #5 by itself. I've checked them in pairs (all combinations) using slots #1 and #3. And when all three modules are in, they occupy #1, 3, and 5.

    Maybe the real inconsistency here is due to the damn CPU contact. Every time, I mess with the RAM modules, I have to remove the fan from the noctua for access and this does involve a certain amount of pushing against it.

    [​IMG]
     
    Last edited: Sep 26, 2016
  14. shimyns

    shimyns Ancient Guru

    Messages:
    1,657
    Likes Received:
    0
    GPU:
    SAPPHIRE r9 290x Tri-X OC
    Well, I've reseated my CPU, ran a sensor test and additional p95 for a while. I haven't made any adjustments to my BIOS settings yet. All 3 modules are in and at 1600 MHz. So far, it seems to be stable.

    EDIT: I decided to just keep the new 3x4GB RAM (doubling the total 6GB I have) although it looks like the current modules aren't the issue, and installed them yesterday w/o problems, just adjusted the timings and dropped the voltage to 1.5V and so far there haven't been issues and it's reading the triple channel at 1600MHz like it should.

    I don't know what the root of this problem was/is or for how long it will remain stable, so for now it remains a mystery.
     
    Last edited: Sep 30, 2016

Share This Page