GTX 1080 Black screen and fans 100%

[Niels G]

Hi,

Last Saturday all of a sudden, my screens went black and fans were skyrocketing for the first time in 10 months. I've never encountered any problem before. I can use my computer fine, but whenever I want to game, my screens will go black in about 30 seconds till 5 minutes, after which I can't do anything anymore. This is frustrating, so I started to explore what could be the problem. I did the following:

 

- My computer still operates though, so trying to read dump is impossible.

- Windows event viewer also does not tell me anything useful.

- The PCI-e slot where my graphics card sits in is reinforced with steel, so therefore it could not be bent (I am also using a graphics card sag bracket).

- I have uninstalled my drivers using DDU in safe mode, no effect

- When watching the data in HWMonitor, I see that it is always on 1.050 volts and about 60 degrees, no problems there

- I have not overclocked it or anything like that

- The motherboard is an Asrock Taichi Z370, which has a very solid power delivery. The PSU is an RM650x, so no cheap PSU or whatsoever.

 

I have not had this problem in over 10 months, and none of the solutions worked. I would not like to open up my graphics card to replace the thermal paste, as it will void my warranty. What more can I do?

 

PS: The card is a Gigabyte Aorus GTX 1080 (https://bit.ly/2NHGWr6)

[Guest Generallee]

This reminds me of the R9 290, it had the same exact problem and it was discovered that the stock memory clock was too high for the Elpida chips on board. A solution was to basically modify the vBios and flash the card which eliminated the problem without impacting performance.

 

I do not know if this procedure could be used on an Nvidia GPU, but you can at least try this : underclock your card both core and memory, try a stress test and see if it goes black after some time. You could also try the card on a different computer if that's possible.

[asus killer]

open the card, clean, new thermal paste. It should work by then

[Niels G]

54 minutes ago, Generallee said:

This reminds me of the R9 290, it had the same exact problem and it was discovered that the stock memory clock was too high for the Elpida chips on board. A solution was to basically modify the vBios and flash the card which eliminated the problem without impacting performance.

 

I do not know if this procedure could be used on an Nvidia GPU, but you can at least try this : underclock your card both core and memory, try a stress test and see if it goes black after some time. You could also try the card on a different computer if that's possible.

Will try

 

38 minutes ago, asus killer said:

open the card, clean, new thermal paste. It should work by then

As stated above: 

Quote

I would not like to open up my graphics card to replace the thermal paste, as it will void my warranty. 

I want to keep my warranty

[asus killer]

9 minutes ago, NielsG said:

 

As stated above: 

I want to keep my warranty

i've seen this before, it's thermal paste. why would it void warranty? does your card have some stickers or something?

 

if they have "warranty void" stickers send the card to them, RMA

[Niels G]

I have finally found the solution: uninstalling Aorus Engine software. Thanks for the tips!