Is my 970 Dying?

Discussion in 'Videocards - NVIDIA GeForce' started by MegaFalloutFan, Nov 24, 2015.

  1. MegaFalloutFan

    MegaFalloutFan Maha Guru

    Messages:
    1,048
    Likes Received:
    203
    GPU:
    RTX4090 24Gb
    Hi,
    I have a pair of ASUS Strix GTX970 in SLI, when I game i overclock them to 1400Mhz, Min voltage to 1170mv and max voltage to 1265mv, FAN set to 60%. (Using ASUS Tool)
    The cards were stable, no artifacts, i could play MGS5 for hours in 4K and it was fine, no issues.
    So i started playing Fallout 4, same thing here, no issues.


    I updated to the latest 359 driver and the day after I started to get artifacts in Fallout 4 about 3 - 5 minutes after loading. (some patches of white squares, random red dots/squares and some white/black "sticks" sort of effect, but game wasn't freezing imidielty)
    I tested MGS5, it hanged in while sitting in a chopper ie not even a full level loaded.

    I clean installed 358.91, and nothing still the same issues, in both games, so its not the driver?

    I played fallout 4 on low overclock 1345Mhz and after about 1 hour i started to get very light random almost unnoticeable artifacts, and after 3-4 hours the game froze.

    Also 50-60Mhz less overclock had strange and severe effect on 4K SLI gaming, in places i had 50FPS i started to get drops to 40 and 35FPS.

    Note: Fallout 4 in SLi using Tomb Raider profile.

    Any ideas?
    Should I try and swap the main card with the secondary card and see if it helps? Or its enough to swap the monitor cable without swapping the cards?

    Should i just get 980ti on black Friday? (only if Pascal coming no earlier then 6 months,otherwise I just wait and if its hardware issue i try and give it to warranty, but they wont be bale to fix it if it works fine when not overclocked?!)


    Thanks
     
  2. C0D3M4N

    C0D3M4N Member

    Messages:
    30
    Likes Received:
    3
    GPU:
    Asus 1080Ti
    I would suggest disabling SLI, and removing one card to test with. Then do the same for the other. (This will help you identify if either of the cards has an issue, or if it is possibly the SLI profile.)


    Just to clarify, you said you were running 1.4ghz core w/ 1.265v , and fixing your fan speed to %60?

    What was max temp with those settings on both cards?(After 4 hrs with fallout 4?) I ask because voltage seems a bit steep, and I don't know that 60% fan on 2 sli cards that dump the heat back into the case is sufficient.

    There are people referring to artifacts due to lighting in fallout 4 while running tomb raider profile... Doesn't account for MGS TPP, but maybe you could remove everything nvidia from your programs and features in control panel, then boot to safe mode and run DDU.

    Good luck.
     
  3. MegaFalloutFan

    MegaFalloutFan Maha Guru

    Messages:
    1,048
    Likes Received:
    203
    GPU:
    RTX4090 24Gb
    Under load its 79 Celsius on one card and 68 Celsius on another in Fallout 4, probably less in MGS5 with normal SLI profile.

    I have a 140mm side fan that blows fresh air on both GPUs, 140mm rear fan also blows inside and both 140mm frontal fans too, only 240mm water cooling that installed on top with 4 120mm fans blows everything out


    I run DDU yesterday before i went to sleep gonna test again.
    what you said makes sense, to test them separately
     
    Last edited: Nov 24, 2015
  4. MegaFalloutFan

    MegaFalloutFan Maha Guru

    Messages:
    1,048
    Likes Received:
    203
    GPU:
    RTX4090 24Gb
    Its a new 860w seasonic platinum PSU and the spares I have are much weaker
     

  5. GanjaStar

    GanjaStar Guest

    Messages:
    1,146
    Likes Received:
    2
    GPU:
    MSI 4G gtx970 1506/8000
    red dots on maxwell mean not enough voltage at that core speed in my experience.
     
  6. MegaFalloutFan

    MegaFalloutFan Maha Guru

    Messages:
    1,048
    Likes Received:
    203
    GPU:
    RTX4090 24Gb
    Any ideas why? It worked fine for a year using same setting.
     
  7. mahanddeem

    mahanddeem Master Guru

    Messages:
    663
    Likes Received:
    28
    GPU:
    Asus GTX1080Ti StrX
    Overclocking a die will degrade it after sometime depending on your luck. Revert to stock and clean windows and/or driver and test again. If still have problems your gpus / or one is toasted
     
  8. GanjaStar

    GanjaStar Guest

    Messages:
    1,146
    Likes Received:
    2
    GPU:
    MSI 4G gtx970 1506/8000
    you used variable voltage and variable clocks under high load, with an OC. this can degrade chips much faster than a constant voltage and clock in my experience.

    So it's possible as stated above that your chip degraded. if there are no issues at factory clocks, you can't really RMA, but if stock clocks have issues, and you have no ethical qualms about it, you can RMA, they won't know you overclocked unless you have a custom bios and don't replace it back with the factory one.

    but definitely check both cards alone to see if any have issues by themselves, which will also give you an idea if the PSU has any issues.
     
  9. Krizby

    Krizby Ancient Guru

    Messages:
    3,103
    Likes Received:
    1,783
    GPU:
    Asus RTX 4090 TUF
    sounds more of a problem with Mem clock rather core clock, lower your Vram clock, even downclock if you have to. If you set your core clock too high most often it will crash to desktop.
     
  10. MegaFalloutFan

    MegaFalloutFan Maha Guru

    Messages:
    1,048
    Likes Received:
    203
    GPU:
    RTX4090 24Gb
    I used that driver cleaner whats its name, re-installed the driver that 100% worked for me.
    Played F4 with Default SLI profile (meaning in reality single card mode) with the overclock, no issues, played for 3 hours right now.

    Now its time to test with a SLI profile, maybe driver cleaner helped.

    Getting acceptable FPS with good graphics in 4K and Single card, 45 to 35.
    With SLI profile it was near 50 almost always (I locked it to maximum 50)
     

  11. MegaFalloutFan

    MegaFalloutFan Maha Guru

    Messages:
    1,048
    Likes Received:
    203
    GPU:
    RTX4090 24Gb
    Hi,
    How do i set Constant Voltage? Im using ASUS GPU Tweak 2, its the only one with both Minimum and maximum voltage setting, the popular MSI tweaker didn't worked for me since it only has maximum voltage and my overclock never worked.
    with asus utility it works.

    Is there any other way to force the voltage constantly to maximum value in 3D mode?

    P.S. If i find out that on card is giving up ill fry it and get a new replacement.:banana:

    Thanks
     
  12. MegaFalloutFan

    MegaFalloutFan Maha Guru

    Messages:
    1,048
    Likes Received:
    203
    GPU:
    RTX4090 24Gb
    I wish, max Min Voltage = 1175, Max max Voltage = 1213
    No way to set them to same number
     
  13. GanjaStar

    GanjaStar Guest

    Messages:
    1,146
    Likes Received:
    2
    GPU:
    MSI 4G gtx970 1506/8000
    you need a custom bios.

    I suggest you check out the thread here.

    http://www.overclock.net/t/1573308/nvidia-gtx-900-cards-custom-bios-upon-request/480#post_24491822

    Mr. Dark there will use your cards bios to disable voltage and clock throtling during games. the card will still downclock and downvolt everywhere else and conserve power.


    or if you want to do it yourself get ready for a bit of reading.

    http://www.overclock.net/t/1517316/...d-980-firmware-zosons-h2o-and-air-custom-bios
     
  14. MegaFalloutFan

    MegaFalloutFan Maha Guru

    Messages:
    1,048
    Likes Received:
    203
    GPU:
    RTX4090 24Gb
    So here is what happened.

    Short Version:
    2-3 months ago I used Liquid Metal to replace the factory thermal paste, it helped, i had better temps and better heat management, like when the fan kicked it it lowered them faster.

    One day without any notice Fallout 4 started to get Purple squares and white lines when i overclocked the cards. :uhoh:
    I tested the Factory overclocking modes and they worked for much longer before the artifact appeared like 3-4 hours. :bang:
    After talking to you guys I decided to remove both cards, and test them one by one in Single card mode and found that it was the SLI card that had this issue.
    So I was thinking of ways to solve it and remembered :idea: that i seen a video about Liquid Metal drying up (after years or year and half), in my case it was just 2-3 months but i decided why not.
    Opened it up and yep, it was dry.
    So i run Artic Clean#1 number of times and then Aertic Clean #2 number of times and put some MX-4:approval:
    :pray2pc:

    Tested the card with highest overclock that I used for gaming and everything was OK.
    So I did the same with the second card (so far working card) just in case.

    They both work, overclock and no artifacts.

    A day after I got a psychotic rage :bonk: playing Fallout 4 in 4K and dealing with the lack of proper Nvidia SLI profile and the s***y 3.5Gb of VRAM (I actually need it in 4K, it reported 3.6 and up to 3.9 being used :realmad:), I went out and got MSI 980TI:dave:.
    LITERALLY the SAME DAY AT NIGHT Nvidia :peeleft: said that Tomorrow on December 1 they releasing a new driver with the official SLI profile :crapper: :cussing:


    P.S.
    Long Version, for those who like to read LOL

    After I got my 4K monitor I needed the best possible overclock.

    I was reading online that there is real "magic" thermal paste that can get you up to 8 degrees CELSIUS lower then regular: Liquid Metal.

    So I got this syringe with liquid metal, read the instructions, cleaned the old thermal paste, STRIX cooler has aluminum heat spreader and what looks like some sort of coated/painted copper heat pipes, the part that touches the GPU is clean and copper, but it doesn't have any copper clock that combined the heat pipes and touch the GPU, all the heat pipes touch it directly and so there are gaps between and they are not shiny smooth like CPU copper heat spreaders, but sort of ruff like.
    So i cleaned everything and applied liquid metal to the GPU, spread it with cotton bud, then did the same to the heat pipes, but the liquid metal would not spread properly (it was cleaned with Artic Clean #1 about 5-7 times and then Artic Clean #2 4-8 times until it stopped giving any gray marks on the white paper).
    In any case their video says that its OK to only apply to one side, that applying to both may glue them together (not permanently, there is a video of guy removing such combination of CPU and big air cooling heat spreader, he just used a hair fan to heat them up and they separated).
    So i did my best, GPU was coated and the heat spreaders coated as best as possible.
    It says in the instructions Not to use it on or near Aluminum and thats no joke: I tested it myself, put a drop of it on the aluminum heat spreader it started to dissolve, yes dissolve not melt, the aluminum was gone.
    Put some Artic Clean #2 to clean it and it only speed up the process LOL, I wish we had some of this stuff that can dissolve steel locks :evilgrin: LOL :cops:
     

Share This Page