Shortly after upgrading my PC to a Ryzen 7 9800X3D, a new motherboard, and DDR5 memory, my RTX 3080 Ti Founders Edition began behaving strangely. The GPU, CPU, and overall resource usage appear to stay around 15–30% while playing War Thunder at 1440p and a locked 60 FPS, but the GPU hotspot sometimes climbs above 100°C and the fans suddenly run at full speed. The game itself remains smooth.
The GPU, power supply, case, and case fans were not changed. I have updated Windows, the BIOS, and NVIDIA drivers, and restarted several times. The card is several years old and has previously suffered from overheating and black-screen crashes. About two years ago, I disassembled it, replaced the thermal paste and pads, and improved the cooling with additional case fans and an undervolt through MSI Afterburner.
Could the new CPU or motherboard somehow be causing this, or is the timing just coincidental? Could an old undervolt profile have been lost? I am also wondering whether the card needs another repaste or whether incorrectly sized thermal pads could be preventing proper cooler contact. I want to determine whether the GPU is defective before my return window expires.
3 Answers
A hotspot over 100°C at a locked 60 FPS is not typical, especially if the card is not under a sustained heavy load. The most likely cause is poor contact between the GPU die and cooler. Since this Founders Edition card has already had its thermal pads and paste replaced, the pads may have compressed, degraded, or been replaced with incorrect thicknesses that keep the cooler from sitting flat. A careful teardown, cleaning, and repaste with the correct pad thicknesses would be a sensible next step.
The upgrade itself is unlikely to damage the GPU thermally. The new system may have changed airflow, driver settings, or removed your old Afterburner undervolt, but it should not make a healthy cooler produce an extreme hotspot at this workload. Confirm that no fan curve or overclock is active, then test with the side panel open and compare core, hotspot, fan speed, clock, and power readings. If the hotspot remains abnormally high, redo the thermal interface and verify the exact pad thicknesses for the Founders Edition cooler.
Do not rely on Task Manager alone for GPU readings. Check GPU-Z, HWiNFO, the NVIDIA overlay, or RivaTuner for GPU power, clocks, core temperature, hotspot temperature, and whether the card is actually throttling. Also check the difference between the core and hotspot temperatures. A very large gap usually points to poor die-to-cooler contact rather than the new processor or motherboard.
The 60 FPS cap explains why the GPU usage can be relatively low. A newer CPU may simply finish each frame quickly, leaving the GPU lightly loaded. That low usage does not make a 100°C hotspot normal, though; the thermal behavior still points toward the card's cooling system.
I did the paste and pads about two years ago, so it is possible the card simply needs the work done again. I was mainly concerned that the timing with the new build indicated a faulty new component.