[SOLVED] Why is my GPU OC no longer stable?

[theoneRISA15]

I have an EVGA 1080Ti FTW3 and I had a stable overclock at +120 on the Core clock. I used that OC for a few months and there were no problems. Then one day it made everything I tried to do crash (games/benchmarks) and I have no idea why as none of my settings changed. I did some testing (3D Mark) and found that if I disable the Core clock OC then everything is fine, so it isn't the Memory OC. I also found that simply decreasing the core clock multiplier didn't help. It only increases the amount of time the program runs before it crashes. If anyone could provide some answers on why this is happening and the solution I would greatly appreciate it.

 

[feelinfroggy777]

It could be a driver issue from the crash. Try doing a clean uninstall of the drivers and download the current ones from Nvidia. Use the DDU tool below in safe mode to remove the current drivers.

https://www.guru3d.com/files-details/display-driver-uninstaller-download.html

 

[theoneRISA15]

Unfortunately even though I did a proper uninstall and reinstall I still have the same issue so it isn't the drivers. Thanks for trying to help though.

 

[damric]

Capacitors on your graphics card and in your PSU can be sensitive to higher temperatures, not filtering ripple quite as well, breaking an overclock that was once stable at cooler temperatures.

You can increase fan speeds, raise voltages, increase your air conditioning, or reduce your overclock.

 

[theoneRISA15]

Thanks for trying to help but the overclock was stable and the temps haven't increased. I was already at max voltage, the fan curve was not conservative at all ( my GPU was sitting at about 50 C under load, but for the sake of trying it I raised all of my fans to max and that didn't help, it didn't bring down the temperature by much. It was only a couple degrees lower), my room is about 21 C and I don't want to have to lower the AC just for this. Also, if you read my original post I mentioned that decreasing the overclock doesn't help. There isn't a single overclock that is stable, even small ones like +25 don't work.

 

[damric]

Thanks for trying to help but the overclock was stable and the temps haven't increased. I was already at max voltage, the fan curve was not conservative at all ( my GPU was sitting at about 50 C under load, but for the sake of trying it I raised all of my fans to max and that didn't help, it didn't bring down the temperature by much. It was only a couple degrees lower), my room is about 21 C and I don't want to have to lower the AC just for this. Also, if you read my original post I mentioned that decreasing the overclock doesn't help. There isn't a single overclock that is stable, even small ones like +25 don't work.

The PSU running hotter is the more likely to be causing the problem. Model #?

Regardless, aging capacitors cannot be avoided, but can be minimized by keeping them cool. The dielectric evaporates over time, and that is why you will see PSUs with more heat resistant caps have the longest warranties.

 

[theoneRISA15]

I'm not an expert, but I haven't had any of this computer for more than a year. I would think that components shouldn't be deteriorating THAT fast to make any overclock unstable.

 

[damric]

I'm not an expert, but I haven't had any of this computer for more than a year. I would think that components shouldn't be deteriorating THAT fast to make any overclock unstable.

You would be surprised. There are terrible PSUs out there only rated for 25C-30C by big brands like EVGA and Corsair and the caps can burn up quite quickly, never making it to the 2 year warranty. Shoddy caps can also be found on low end motherboards and graphics cards, but it's much more of a problem on PSUs.

The filter circuit is extremely important for overclock stability. You want your DC to look like this ------ not this ~~~~~.

 

[Karadjgne]

There's multiple components on a graphics card, pretty much the same as on your motherboard with 1 exception. Where a motherboard has multiple sensors for reporting temps, the graphics card only has 1. The graphics processor unit (gpu). There's no temp sensor on the VRM's, vram, pcie or USB circuitry etc. So it's quite easy to see VRM's topping 95-100°C, and yet the gpu only reporting 70-80°C. That's card killing temps that you cannot see, but will definitely mess with the cards stability.

And all it takes is a small lack of cooling on the heatsink side closer to the back edge of the gpu and prior stable acceptable voltages are now turning your vrms/vram into an oven.

Might want to break out the tiny Phillips screwdriver, pull off the shroud and give the whole hestsink/fan area a good clean. Also physically verify that any and all fans are spinning as they should be