Throttling? Insufficient VRM cooling on 780 ti?

Discussion in 'Videocards - NVIDIA GeForce' started by pangeltveit, Dec 12, 2013.

  1. pangeltveit

    pangeltveit Member

    Messages:
    35
    Likes Received:
    0
    GPU:
    EVGA 780ti SC AC Hybrid
    Greetings,

    I put on a Arctic cooling Accelero Hybrid on my GTX 780 ti yesterday and did a few quick tests before I called it a night.

    in Unigine Heaven, it was running beautifully, with only a few artifacts, was reaching 1340mhz on the core with stock volts.

    it was not throttling down though, but then I launched a game, and after 5 minutes the Mhz dropped to 550-650 on the core and the game crashed and I had to reset my PC.

    The bigger heatsinks that came with the set did not fit on the card so I had to go for the "smaller" ones, but with direct airflow over them I would think it enough?
    [​IMG]

    Does it sound like a "too hot VRM issue"? The GPU Core did not go above 44'C
     
    Last edited: Dec 12, 2013
  2. Extraordinary

    Extraordinary Guest

    Messages:
    19,558
    Likes Received:
    1,636
    GPU:
    ROG Strix 1080 OC
    GPU core didn't go above 34c during gaming / benching ?

    Find that hard to believe
     
  3. pangeltveit

    pangeltveit Member

    Messages:
    35
    Likes Received:
    0
    GPU:
    EVGA 780ti SC AC Hybrid
    Apologies, it was 44'C,

    [​IMG]
     
    Last edited: Dec 12, 2013
  4. Pill Monster

    Pill Monster Banned

    Messages:
    25,211
    Likes Received:
    9
    GPU:
    7950 Vapor-X 1100/1500
    Say what??? lol

    Artifacts appear when your clocks are too high. They are not a good thing, artifacts are very bad. Reduce the OC and your problems will no doubt disappear.


    lmao.
     
    Last edited: Dec 12, 2013

  5. pangeltveit

    pangeltveit Member

    Messages:
    35
    Likes Received:
    0
    GPU:
    EVGA 780ti SC AC Hybrid
    Yes ofcourse, I will do that, but I would much rather want to overvolt it so that it is stable and no more artifacts, but if it is throttling at stock volts, then I would be forced to lower my Core clock.

    Which is what I am wondering, if you think that the picture in my original post shows that my VRM has sufficient cooling?
     
    Last edited: Dec 12, 2013
  6. Pill Monster

    Pill Monster Banned

    Messages:
    25,211
    Likes Received:
    9
    GPU:
    7950 Vapor-X 1100/1500
    Overheating pwm's will throttle back the clocks before overheating, but they won't cause artifacts. Anyway, at 44c the GPU isn't working hard enough to heat them up.
    Vrm fets can handle about 100c before throttling anyway.

    Once again, reduce your overclock.
     
    Last edited: Dec 12, 2013
  7. pangeltveit

    pangeltveit Member

    Messages:
    35
    Likes Received:
    0
    GPU:
    EVGA 780ti SC AC Hybrid
    how does my GPU temperature affect my VRM temperature?

    They do not share the same Cooler. I have a Closed liquid cooler on the GPU, and Aluminum heatsinks on the VRM's, with a fan above them.

    Would not a overclocked GPU, make the VRM's work harder and therefore produce more heat? even if the GPU is not overvolted?

    and if I can overvolt the GPU, with my VRM staying cool, why should I not do that instead of tuning down my overclock?

    edit: and the GTX 7xx series dont allow you to monitor the VRM temperatures, as far as I know, so I cant know if I am reaching that 100'c line.
    Also, ive read that it is 75'C for the GTX 700's, might be wrong about that though.
     
  8. Pill Monster

    Pill Monster Banned

    Messages:
    25,211
    Likes Received:
    9
    GPU:
    7950 Vapor-X 1100/1500
    Forget it.
     
  9. southamptonfc

    southamptonfc Ancient Guru

    Messages:
    2,620
    Likes Received:
    646
    GPU:
    Zotac 4090 OC
    It's most likely that the GPU speed dropped and the game crashed because your overclock was unstable. That is EXACTLY what happens with an unstable overclock on these gpus.

    As Pill said, sort out your overclock and your problems will be gone.
     
  10. pangeltveit

    pangeltveit Member

    Messages:
    35
    Likes Received:
    0
    GPU:
    EVGA 780ti SC AC Hybrid
    I might not have many posts on this Tech forum, doesnt mean I dont have many on others ;) .

    I dont know you, but I see you have been here a while and have alot of posts, but dont tell me you give up when not listened to or not responding as expected? :confused:


    That is something I did not know, thanks for that piece of intel :)

    Today's plan consists of Starting my OC-ing with overvolting as well, might borrow my fathers Thermal monitor "gun" to keep an eye on the VRM's
     

  11. Netherwind

    Netherwind Ancient Guru

    Messages:
    8,821
    Likes Received:
    2,401
    GPU:
    GB 4090 Gaming OC
    I'm amazed that you managed to reach over 1300 on stock volts. Artefacts are afaik only visible if the memory is clocked too high but I might be wrong here. The benchmark/game crashed when the core was too high and gave me artefacts (checkerboard) when the VRAMs were too high.

    If your temps are as low as you claim there should be headroom for some volt modding as long as you also increase your power limit. You may have to flash a custom BIOS allowing you to go up to 200% Power limit.
     
  12. Shadowdane

    Shadowdane Maha Guru

    Messages:
    1,464
    Likes Received:
    91
    GPU:
    Nvidia RTX 4080 FE
    Your GPU clocks will drop if the GPU or driver crashes. Which can happen if your overclocking with not enough voltage.

    My GTX680 will crash the driver and drop to low power 3D clocks if I go beyond approximately 1100Mhz on stock voltage. Sadly my card sucks and even overvolting I can't get beyond 1200Mhz without a driver crash or hard lock.
     
  13. Fox2232

    Fox2232 Guest

    Messages:
    11,808
    Likes Received:
    3,371
    GPU:
    6900XT+AW@240Hz
    Higher gpu clocks = higher current draw.
    Current goes through vrms.
    Higher current = more heat generated.
     
  14. eclap

    eclap Banned

    Messages:
    31,468
    Likes Received:
    4
    GPU:
    Palit GR 1080 2000/11000
    what were your top vrm temps? monitor with gpu-z and post back.
     
  15. Agent-A01

    Agent-A01 Ancient Guru

    Messages:
    11,628
    Likes Received:
    1,119
    GPU:
    4090 FE H20
    OP those aluminum heatsinks are garbage. I bought an accelero xtreme for my GTX 470. GPU temps max 40c and it would crash under load. VRM temps made it crash. I reused the PCB heatsink plate and it fixed my issue. Youre better off to stop your OC until you can buy some copper heatsinks, because aluminum sinks arent worth a damn
     

  16. pangeltveit

    pangeltveit Member

    Messages:
    35
    Likes Received:
    0
    GPU:
    EVGA 780ti SC AC Hybrid
    I just now overclocked only my GPU and left memory at stock, I still got artifacts. But I think I will flash my card to get more out of it :)


    I have been getting some driver crashes indeed, I think its time to up my volts ;)


    exactly as I though :)


    Sadly, I cant get any temp readings off of these cards, not on GPU-Z anyway.

    [​IMG]


    Here are some pictures I took with an optical temperature reader, at directly on the heatsink of the VRM, I get ~50'c, on the back of the PCB right over the VRM's, I get aproximately 60-65'c.

    [​IMG]

    [​IMG]
     
    Last edited: Dec 12, 2013
  17. Fox2232

    Fox2232 Guest

    Messages:
    11,808
    Likes Received:
    3,371
    GPU:
    6900XT+AW@240Hz
    From what range you are taking reading? If from more than 5-10cm, you are actually measuring pretty large area and thermometer display average infrared radiation.

    Your thermometer should have somewhere written distance vs area ratio.
     
  18. pangeltveit

    pangeltveit Member

    Messages:
    35
    Likes Received:
    0
    GPU:
    EVGA 780ti SC AC Hybrid
    aproximately 25cm from target, 7,5cm dot o_O, will re-test at closer range.
     
  19. Pill Monster

    Pill Monster Banned

    Messages:
    25,211
    Likes Received:
    9
    GPU:
    7950 Vapor-X 1100/1500
    Artifacts can be due to either GPU or VRAM or even both at once.

    Artifacts appearing as squares or blocks are always, always VRAM related, since the monitor is really a huge checkerboard, with each square mapped to a storage cell in vram.
    Memory cells are laid out in rows and columns on the chip kinda like a naughts and crosses (X's & 0's) game.

    Now, if data stored in a block of memory cells gets corrupted by high OC or whatever, the data stored in that block of cells will often be corrupted when it is sent to X's and 0's address on the monitor.

    Fried or damaged vram shows up as a checkerboard pattern in the BIOS for the same reason, like with all those GeForce 8800's that had to be baked. :nerd:

    Just some fyi....


    Btw can someone please resize their screenshots to 1q280x1024....thx
     
    Last edited: Dec 12, 2013
  20. pangeltveit

    pangeltveit Member

    Messages:
    35
    Likes Received:
    0
    GPU:
    EVGA 780ti SC AC Hybrid
    Nice to know, thanks :)

    The artifacts I have been getting is more like Lines and stretched textures going across the screen.

    Not had them for a while now, currently reached max 1299Mhz on the core


    New Temp readings, meassured at 2-3cm (0.75cm dot) and got max reading of 68'c above the VRM (back of PCB) and 52'c on the heatsinks.

    This is in Unigine Heaven though, not Furmark which I am not planning on running.
     

Share This Page