Did my GPU just craped out?

Discussion in 'Videocards - AMD Radeon' started by Messerschmitt, Sep 15, 2013.

  1. Messerschmitt

    Messerschmitt Master Guru

    Messages:
    465
    Likes Received:
    0
    GPU:
    ASUS 7970 DCU II
    It started suddenly in one of my gaming session. All of the sudden redish rainbow artifacts shooting from the center of the screen. Did that a couple of times. I decided to close the game.
    All of the sudden my screen goes gray, with what it appeared to be squares.
    Then my PC would have the black screen of death after the windows logo. It would just hang in there. It would go fine in safe mode tho.

    I then decided to unplug one of my monitors (have 2). Windows boots up fine. However, if I tried to open HWiNFO64, or even catalyst drivers center, the screen would go blue. Not the BSOD, but just blue screen. It looks like trying to access any software that is related to the graphic card no longer works.

    Also while I had my second monitor connected, in BIOS, the primary monitor would no longer turn on, and the second monitor had 2 vertical wide stripes going from top to bottom. This is the primary reason I would think is dying.

    I did OC'ed it. I managed to get a nice 1125/1500, and it ran fine while benchmarking it in Heaven 4.0. Temps never went over 75c.

    I did manage to get the windows going with one monitor and uninstall the Catalyst drivers. With no drivers, it doesn't act up, but as soon as I re-install catalyst, same problems occur. I cannot boot with 2 monitors, and if I try to open catalyst, it gives me the blue screen.
     
    Last edited: Sep 15, 2013
  2. IcE

    IcE Don Snow

    Messages:
    10,693
    Likes Received:
    79
    GPU:
    3070Ti FE
    Yep, sounds dead. RMA
     
  3. thatguy91

    thatguy91 Guest

    Was it the standard (925 MHz/5400 MHz RAM) or GHz edition (1000 MHz/5500 MHz RAM) GPU?

    In any case, the highest overclock you get without artifacting or stability issues doesn't necessarily mean it's the safest maximum overclock. I would have knocked that back to say, 1050 MHz on the GPU and say, a limit of a 300 MHz (so 75 MHz) overclock on the memory. Any more than this is more about bragging rights, as it won't make the difference between the game being jerky and smooth. In fact, you would have much more gain from changing over to 2x8GB DDR3-2133 (Ivy-Bridge/Z77 and Haswell/Z87 respond very well to DDR3-2133) and tweaking your bios settings that what you would get from that extra overclock. At a complete guess, it's probably the memory that failed. Different brands and models can be quite finicky when it comes to safe overclockability
     
  4. Messerschmitt

    Messerschmitt Master Guru

    Messages:
    465
    Likes Received:
    0
    GPU:
    ASUS 7970 DCU II
    Most reviews showed that RAM clocks are irrelevant for games. a DDR-1333 vs DDR-2333 clock was 2 FPS difference in a game where the avg FPS was over 100, and graphics turned to a minimum. With graphics pumped up it was non-existent.

    OC-ing CPU or RAM gives minimum gain for gaming.

    In regards to my card, it was the standard 925Mhz. From my understanding OC'ing it shouldnt reduce the life of the card very much. Definitely not crap out in 2 days. Most often when the OC is too much you only get a few artifacts or crashes. I didn't get any of those until the magic show of lights a few minutes before getting busted.

    I didn't even modified the voltage. It was the stock 1.17v. Reviews and other users say they got it running at 1.25 or 1.3v, with ~1200 clocks.

    Is it a bad assumption that by leaving the voltage to stock and finding the max OC capability, I can expect the card to last at least half of it's MTBF (which should still be several years)
     

  5. Pill Monster

    Pill Monster Banned

    Messages:
    25,211
    Likes Received:
    9
    GPU:
    7950 Vapor-X 1100/1500
    What software did you use to overclock?
     
  6. thatguy91

    thatguy91 Guest

    Forget the specific argument of DDR3-1333 vs DDR3-2133, although I think people would be idiots buying DDR3-1333 for anything but an extreme budget non-gaming rig, I still think if you're going to the extent of eeking the absolute maximum performance out of what is considered a high-end card that you would opt for something a little better than DDR3-1333 ;).

    Anyways, I still think those overclocking amounts are a little high (the ROG version that does 6600 MHz memory doesn't count for comparison as it's different RAM and different binned GPU) it does sound strange that it crapped out only after two days. That would suggest that it was faulty to start with, which can happen.
     
  7. ---TK---

    ---TK--- Guest

    Messages:
    22,104
    Likes Received:
    3
    GPU:
    2x 980Ti Gaming 1430/7296
    what am I reading here? a guys gpu crapped out and your recommending a system ram upgrade from 16gb of 1600 to 16 gb of 2133?
    OP sounds like you need to rma that video card I`m afraid
     
  8. Loophole35

    Loophole35 Guest

    Messages:
    9,797
    Likes Received:
    1,161
    GPU:
    EVGA 1080ti SC
    OP have you tried to reseat the card by chance the DCU2 is a heavy card.
     
  9. ---TK---

    ---TK--- Guest

    Messages:
    22,104
    Likes Received:
    3
    GPU:
    2x 980Ti Gaming 1430/7296
    you think the pcb might be bending?
     
  10. Loophole35

    Loophole35 Guest

    Messages:
    9,797
    Likes Received:
    1,161
    GPU:
    EVGA 1080ti SC
    It's a posibilty.
     

  11. Undying

    Undying Ancient Guru

    Messages:
    25,339
    Likes Received:
    12,747
    GPU:
    XFX RX6800XT 16GB
    It's a shooting in the dark.

    He fried vram so card gone nuts.
     
  12. Loophole35

    Loophole35 Guest

    Messages:
    9,797
    Likes Received:
    1,161
    GPU:
    EVGA 1080ti SC
    Most likely but it's a three second procedure just to check.
     
  13. ---TK---

    ---TK--- Guest

    Messages:
    22,104
    Likes Received:
    3
    GPU:
    2x 980Ti Gaming 1430/7296
    since when is basic troubleshooting a shot in the dark?, its not like Loopy recommended staring at a heaven 4.0 run for 12 hours straight. looking at the pcb is a very minimal procedure
     
  14. Messerschmitt

    Messerschmitt Master Guru

    Messages:
    465
    Likes Received:
    0
    GPU:
    ASUS 7970 DCU II
    I'm already going to RMA it on Monday. There was nothing to reseat, tower didn't move and it was pinned down by 2 screws.

    Anyway, so in your opinion just OC'ing the clocks (without modifying the voltage) can also damage the card?

    What OC's would you personally put if you had this card (provided that OC is stable)?

    And I have no idea if I'm wrong or not, but wouldn't slightly increase the voltage "take away" the stress of running those higher OC?
     
  15. yasamoka

    yasamoka Ancient Guru

    Messages:
    4,875
    Likes Received:
    259
    GPU:
    Zotac RTX 3090
    Just OCing the clocks cannot damage a card that easily, and it's mostly the side effects of the extra frequency which causes damage, i.e. running the card over a cliff when it's already heating up badly.

    Voltage is what cuts down lifespan mostly. Voltage adds stability to higher clocks. It does not take the stress away.

    Did you check what the other guys have suggested? Tried reseating, tried on a different system, tried a different Windows installation?

    If it's a hardware issue, then it doesnt act up without drivers because it's not being used to the point where those issues manifest themselves.

    Troubleshoot before doing any RMA and save yourself the hassle.
     

  16. thatguy91

    thatguy91 Guest

    No, re-read that post :). I was just saying to overclock the card to the verge of artifacting makes no sense over using say, 50Mhz less to be a bit 'safer'. If that difference is important to you, then DDR3-2133 or faster would be a better suit.
     
  17. Messerschmitt

    Messerschmitt Master Guru

    Messages:
    465
    Likes Received:
    0
    GPU:
    ASUS 7970 DCU II
    I didn't OC'ed the card on the verge of artifacting tho. I never had any artifacts, crashes, driver errors, etc until the show of lights 2 minutes prior of it collapsing.

    And I did monitored the temps very closely. I can't remember when and if it ever went over 75c. I created a custom fan profile too making sure of that. On most games it stayed around 70c, which is a very reasonable temperature when GPU's are usually made up for 90+ (I would never run one above 80 personally), but what I'm trying to say is it definitely wasn't the temperatures that collapsed it.

    I have already exchanged the card. However, not sure if this time maybe as someone suggested, I will just leave it at 1050 max on core and maybe 1500 on memory, no matter how much further it can go without artifacting/crashing during benchmarks/games.

    Will keep this updated for whomever is curious.
     

Share This Page