My PNY RTX 5070 Ti has been running normally for about six months, but today the fans suddenly became extremely loud while I was playing The Finals. MSI Afterburner showed the GPU reaching around 88°C, even though the same game and load were quiet and cool last night. I opened the case, cleaned out the dust, and confirmed that all of the graphics card fans are spinning. I may have installed a Windows update overnight, so I'm wondering whether that could have changed the GPU settings or driver behavior. Is 88°C still an expected temperature, and what should I check before considering repasting or undervolting the card?
4 Answers
Check whether Windows installed an update or whether the graphics driver changed overnight. An update can sometimes reset a driver profile, fan curve, power setting, or game-specific configuration. Compare your GPU settings with what you were using before and reinstall or roll back the driver only if the change lines up with the update.
88°C is not automatically dangerous for a modern GPU, especially if it is still maintaining its normal clocks and not reporting thermal throttling. The sudden increase in noise is worth investigating, though, since the fan curve or workload may have changed even if the temperature is technically within the card’s limits.
The temperature itself is what caught my attention, but the biggest difference is that the fans are now loud enough to hear over my headphones when they used to be nearly silent.
Use GPU-Z or another hardware monitor while playing and record the core temperature, hotspot temperature, clock speed, power draw, utilization, and fan speed. That will show whether the card is actually overheating, running at a higher power limit, or simply following a more aggressive fan curve. Also check whether another process is using the GPU in the background.
I wouldn’t repaste the card as the first step. Thermal paste usually doesn’t suddenly fail overnight, and opening the card can affect the warranty. Check the driver, fan profile, background GPU usage, room temperature, and case airflow first. Consider undervolting only after confirming the hardware and software are behaving normally.
A failed or obstructed fan is still worth ruling out, but since all of them are spinning, monitoring the fan speed and hotspot temperature should tell you more before taking the card apart.
I think Windows did install an update last night, so I’ll compare the driver and GPU settings before changing anything else.