X99 rig going under water eventually

Discussion in 'General Hardware' started by fantaskarsef, Feb 8, 2015.

  1. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    Well, Thursday no crash, Friday no crash, now I just had a lock up again. Not really sure what triggers it though... presumably it's not the drivers, and not directX. It's also no the temperatures, everything is as I expect from watercooling. CPU around 50-55°C, GPUs each around the 40°C, all VRM's are at similar temperatures.

    Could it be something with the power, in any case? Should I check the cables once again? Because I'm somewhat running out of ideas... and it does not crash with benchmarks or other games.
     
  2. mexy750

    mexy750 Active Member

    Messages:
    64
    Likes Received:
    0
    GPU:
    3x EVGA Titan X SC
    sorry to hear that, i don't think this has anything to do with HW, it's simply a bug in BF4. it might be triggered with a combination of features as many people on the net claim to have it fixed with different methods, i think that they were changing settings until a point where it's fixed. when you google GetDeviceRemovedReason you'll get tons of BF4 hits and fixes, it's just anoying that such a good game is ruined by a stupid error :bang:
     
    Last edited: Aug 23, 2015
  3. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    Yeah, that's what I found with recent research, too. Not really something that makes sense that could be done, and what I read mostly ends up in very basic troubleshooting ('repair' your game, reinstall it, run BF4 as an administrator, stuff like that mostly).

    The game was getting more on my nerves anyway, yet I would be curious if there's something I could do since I play it with friends, and second, what if it happens in the next game that's about as CPU intensive as BF4? I don't think I even have one to properly test it. Maybe realbench...
     
  4. yasamoka

    yasamoka Ancient Guru

    Messages:
    4,849
    Likes Received:
    243
    GPU:
    EVGA GTX 1080Ti SC
    Unstable CPU overclock? 4.5GHz is high, it might not be fully stable. BF4 is CPU-intensive.

    The first thing to do when you get issues with anything is to reset your overclock, before troubleshooting for other causes.

    I generally test CPU stability with Prime95, but it's a real punisher with AVX...
     

  5. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    Well I thought about the unstable overclock thing, and the only chance I've got is to put in more power, I think there's a little headroom still.

    I will save my bios's settings and then revert to defaults, as the system without any overclock did not show any crashes at all with BF4 or any other game or benchmark, thankfully :D
     
  6. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    I've been running the rig at 4.5GHz core, 4.25GHz uncore, and with those SLI 980s at 1500Mhz core lately.

    Some time ago I played with Unwinders advanced voltage support for the 980s via the thir party hardware database. I already had troubles maintaining the 1500MHz since 359 driver branch came up, but I thought to counter that. Not I've been surprised to find my system wouldn't boot up anymore once I've applied +80mv overvoltage at the GPUs.

    So I rebooted, reduced the overclock again, and am slowly working my way back up from about 1450MHz on the GPUs. Oddly enough, suddenly my CPU overclock wasn't possible anymore... booted normally on Saturday only to find my system not posting at first, then with the message of a failed CPU overclock. Followed by various error codes (B2 and others), I eventually got into the bios at stock settings.

    Reapplied the overclock I had before, and found it to be unstable. Pushed in a notch more voltage but still couldn't get 4.5GHz stable at 1.27v.

    What surprises me is that I'm not sure what to do now... did I damage the hardware accidentally when I toyed around with the GPUs? I could anyway pump in 1.3v into the core and see if I can keep the frequencies, but I'm somewhat hesitant as I'm not sure it would help that much.

    Any suggestions on what I should do next? System seems to run fine with the CPU OC reduced to 4.35GHz (35 multi @ 125Mhz strap, RAM @ 3000MHz).
     
  7. -Tj-

    -Tj- Ancient Guru

    Messages:
    17,173
    Likes Received:
    1,921
    GPU:
    Zotac GTX980Ti OC
    Try to reset uefi bios to optimized defaults, save & reboot and reapply your old OC. Sometimes it can get a bit "funky" if you have a few failed OC's
     
  8. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    So far I've gone for the save route and ran the reduced overclocks for some time to see if the mainboard acts up more, or anything else happens. Bloody rock stable since then.

    Thanks for the advice Tj, I guess I'll do that, 'reset' the bios and eventually get to where I am now, and probably I need to put in more voltage to keep 4.5 stable (still got some reserves I'd say).
     
  9. -Tj-

    -Tj- Ancient Guru

    Messages:
    17,173
    Likes Received:
    1,921
    GPU:
    Zotac GTX980Ti OC
    Well asus is sometimes known to have a bit buggy bios after it failed by bad Oc.

    Ive seen it myself and by few other owners.. Its like some bios part gets "corruption" and only bios reset fixes that. Kind of like windows sfc /scannow.


    Voltage and stability is restored afterwards.. No need to use higher volts.

    Edit: well unless your oc wasnt 1000% stable to begin with.. Maybe you need just a small 0.003v bump on your old initial oc.. try that after you reset and reapplied your old oc back.:nerd:
     
    Last edited: Dec 18, 2015
  10. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    Yeah, I will do that probably tomorrow. Also I wondered if a just very slightly unstable overclock was responsible for my crashing drivers, but that most likely was an edgy GPU overclock as well :D

    So I should just load defaults? Nothing more like maybe trying to flash the latest bios version (even if it's the current one)?

    Thanks for the help, much appreciated :)
     

  11. -Tj-

    -Tj- Ancient Guru

    Messages:
    17,173
    Likes Received:
    1,921
    GPU:
    Zotac GTX980Ti OC
    Hey np, welcome :)


    yeah could be both, but could be just a coincidence. Resetting to optimized default is usually enough
    About voltage; for example in my case 1.230v was a little unstable, 1.234v fixed it..
     
  12. Corrupt^

    Corrupt^ Ancient Guru

    Messages:
    7,045
    Likes Received:
    349
    GPU:
    Geforce RTX 3090 FE
    Not necessarily the CPU, but BF4 seems to succeed at finding stability issues better then any stress test.

    In some way, BF4 predicted my PSU failing a few months ago. I was having random crashes, etc until 1 day it got bad enough for the PC to simply... loose power and shutdown.

    My experience with BF4 has been that if it crashes, most of the times it's actually something in your system being unstable.
     
  13. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    Well, new problem as I need to go back to the start. More or less.

    As the lucky winner of Gskill's and Guru3D's contests I won the DDR4 kit, which is 2x8GB 3200MHz 16-18-18-38 (part number ending on GTZB). Thanks again!

    Now I ordered myself a second kit to make use of X99's quad channel layout, but out of my own mistake I ordered the following: 2x8GB 3200MHz 16-16-16-36 (part number ending on GTZ only).

    Now I'm unsure if I should return my RAM (I just realised my mistake now that I've got them, noob me!), or if there's a chance to run both kits in my rig. I could even live with all RAM running 16-18-18-38 as long as I could go down to 1T. Bad thing is, I know sh*t about Haswell overclocking and even less about RAM overclocking, and today it's too late for me to tinker with it.

    Any suggestions? I'd probably just put them in, set all timings to the slower module, and hope for the best (with no overclocks applied besides RAM settings and necessarily the strap at first).
     
  14. Fender178

    Fender178 Ancient Guru

    Messages:
    4,184
    Likes Received:
    207
    GPU:
    GTX 1070 | GTX 1060
    I think the best bet would be to set the timings to match the 16-16-16-36 set of RAM because I doubt there would be a difference between the numbers. To me it would be easier to lower the numbers instead of raising them. If that does not work you could try returning the RAM kit.
     
  15. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    Huh, I thought it would have been easier to increase timings as they'd be a bit more lose.
    Problem with returning an opened kit is I don't know if I can do that without losing money right away.
     

  16. Fender178

    Fender178 Ancient Guru

    Messages:
    4,184
    Likes Received:
    207
    GPU:
    GTX 1070 | GTX 1060
    You could try setting the timings both ways and see how stable your system is. Yeah there's that. Or you could try selling the kit if you can't use it.
     
  17. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    Yeah, I guess I will have to tinker with it this weekend. Hope I get to that so I could technically send back the RAM on Monday or Tuesday, should get my money back :)
     
  18. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    So a little update, weird things happening:

    Was running stock bios on my R5E, put in all the modules, and the system booted fine with basic specs, and also with the XMP applied, no CPU overclock. I was even able to run all modules as 3200MHz, 16-18-18-38-1T without any voltage inputs. Yesterday I was feeling adventurous and flashed bios 0901, and suddenly it doesn't want to do it at all. Either I run the sticks with 2133MHz, or not at all. Overclock fails each time I try to apply it, and during reboots I see the Q code 55 (no RAM installed). I even set RAM voltages to 1.35V, upped system agent voltage a bit (+0.05), and even tried to feed the CPU itself a little more, in case it would have helped, but to no avail.

    So unless some better ideas come along, I think I'll flash an older bios (will try with 0802 first, then if it doesn't help back to 0801) and see how it goes (never had compatability issues). But eventually I get the growing feeling I should just send back the wrong RAM I bought and order another kit :)

    Any input on this? Something I missed?
     
  19. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    12,181
    Likes Received:
    4,339
    GPU:
    2080Ti @h2o
    Well, status update... got my second set of 2x8GB ram, and putting them in they post and boot with standard jedec specs (2133MHz). Sadly, I can't seem to get XMP to run. I need to run memtest today... I'm really disappointed to be honest, can't even get them to run at 3200MHz... already updated to latest bios, tried applying the XMP stats manually to no avail...

    Any input?
     

Share This Page