Screen turns dark and fans start spinning. Discord and other sounds remain audible.
Screen turns dark and fans start spinning. Discord and other sounds remain audible.
at low load conditions (about 57 degrees hotspot and 40 degrees on the gpu), it operates at 1000-1200 rpm (~50% speed).
when under load (around 80 degrees hotspot and 61 degrees on the gpu), it increases to 1720 rpm fan speed (~70% speed).
This isn't a suitable cooling balance.
It suggests the issue may be due to elevated ambient temperatures (hot air attempting to cool the card), a dusty heatsink, fan, or inadequate paste application. Or perhaps the paste and pads require replacement.
What is your complete cooling configuration?
Your case has fans installed, directional airflow, an AIO radiator in front of the case, and a custom fan speed profile (quiet mode, etc.).
All these factors influence the video card temperature.
This situation might also suggest the card is throttling to manage heat, a detail I overlooked as another sign of overheating. The in-game configurations being applied also play a role, since reduced settings lower the GPU's workload, influencing power consumption particularly when performance targets are not met at high frame rates.
If you're playing titles like Cyberpunk and your 3080 isn't reaching the maximum GPU boost of 1710 MHz or 1785 MHz for the overclocked version, and instead hovers near a base clock of 1440, overheating is likely.
Repasting the GPU is straightforward, and your 3080 model features only eight screws securing the PCB to the heatsink. I've previously replaced two GPUs using the same ASUS TUF cooler design. By removing and reattaching them carefully—avoiding damage to thermal pads—you can reuse the boards without needing to change the paste, and you'll just need to swap the thermal paste.
I recommend using a PTM 7950 phase change material instead of traditional thermal paste. Simply cut a 29mm x 25mm piece, and you'll likely avoid the need for future rework as long as you own the original card.
my main air intake is positioned at the front with three fans, while the exhaust comes out through another fan and a relatively open case frame at the back, along with the AIO at the top. this setup uses three fans for air intake and three more (one for the back of the case and two from the AIO). so far it's working reasonably well, but based on your comments and my observations, it seems my GPU is generating a lot of heat, so I might need to consider reapplying the GPU entirely.
Edit. The fans are now running at full speed at both the front and back, and the AIO is using a custom profile that maintains 40% speed at lower temperatures (with the CPU package as the main sensor) and reaches 100% at 75°C.
If I proceed with repasting and replacing the thermal pads and paste, which specific pads should I choose? I’m not sure which ones are best, but I’d prefer a thorough approach to ensure everything is done properly.
Thank you both for your feedback—I’ve been having a tough time with this issue. I hope it gets resolved!
Ok well your temps don't look awful and the power supply looks ok, has the 200% excursion/transient response.
Suspecting hw gpu fault. you rinse and rinsted software, what's left? Some obscure thing some genius knows or just the card showing symptoms. Last move is to search for similar known symptoms of gpu fault.
So looked it up here – there are a few teardowns, one from Gamer's Nexus:
ASUS RTX 3080 TUF OC Review & Tear-Down: Thermals, Noise, & Overclocking
Not an OC version?
Asus TUF 3080 thermal pad replacement
Another chance: faulty sensor causing issues
What about the cable (dp or hdmi)? Try one of the known good ones
Monitor glitch? What’s the model and make?
If it’s still under warranty, contact RMA. If not, renew pads – they might not fit.
if it’s a sensor or core that can’t be repaired, i’ll have to accept it until i upgrade to a different card later. Undervolting still appears to resolve part of the problem, so it’s not as if it’s always crashing.
Thanks for checking this out. I plan to use this and other references to replace the thermal pads and paste soon.
A monitor glitch seems unlikely because everything has been functioning well so far and the hardware isn’t old. It would also be improbable given that the GPU is running at full fan speed, which usually points to a thermal or hardware failure.
This PC was a pre-built unit purchased over five years ago, so the warranty is no longer valid. The cables have been swapped recently and the slots have been modified extensively due to the issue.
Since the graphics card isn’t overclocked, the second link is what i’ll rely on.
Undervolting the card might point to an overheating problem, which seems likely, but other factors like a faulty PSU, power delivery issues, or input problems could also be involved. I’d still consider repasting the card. This site can clarify what’s needed, and there’s a linked video on the page.