I repasted my nearly six-year-old RTX 3080 Ti Founders Edition after it started running unusually hot and loud. Disassembly went smoothly, the thermal pads looked intact, and the old paste had poor coverage. I cleaned the die with alcohol, applied a sheet of PTM, and reassembled the card, making sure the screws and GPU spring bracket were snug.
Since then, the computer works normally for desktop use, but launching games often causes the GPU to crash and reboot the system. Occasionally I see nvcontainer or RuntimeBroker errors mentioning an inaccessible memory address. GPU statistics appear normal overall, although utilization jumps around at idle.
I'm wondering whether the issue could be incorrect spring-bracket orientation, uneven or excessive mounting pressure, poor PTM contact, or the indented thermal pads preventing proper contact. I have a spare GPU available, but I'd like to determine whether this card can be safely repaired before passing it on to a family member. What should I inspect or test first?
4 Answers
The timing strongly points to something that changed during disassembly or reassembly. I’d take it apart again and carefully inspect the die, cooler contact surface, spring bracket, screws, pads, and every component around the GPU and memory. Look for shifted pads, cracked or missing components, contamination, or anything that could prevent the cooler from sitting flat. Reassemble it methodically and test at stock settings before trying any overclock or undervolt.
I wouldn’t jump straight to replacing the PTM. PTM is commonly used on GPU dies, and the first priority is verifying flat, even contact and correct mounting pressure. The older memory and VRM pads may also be compressed enough that they interfere with the heatsink sitting properly, so inspect their thickness and placement rather than automatically replacing them with ordinary paste. Avoid heating or baking the card, since that can cause permanent damage and won’t fix the underlying problem.
Check the GPU spring bracket first and confirm it is installed in the correct orientation. It is what applies the proper pressure between the die and heatsink. Too little, uneven, or excessive pressure can all cause problems. Also inspect the cooler base for warping and make sure the PTM sheet is positioned correctly and has its protective films removed.
Because the crashes happen when a game starts using the card heavily, memory or core contact is a real possibility. A displaced thermal pad or a component that was stressed while separating the board could cause this even if nothing visibly fell off. I’d reseat everything and test with the spare GPU to confirm the rest of the system is stable. If the problem remains with this card, professional board-level diagnosis may be the safest option.
Related Questions
Lenovo Thinkpad Stuck In Update Loop Install FilterDriverU2_Reload