GPU hotspot reaches 105°C
GPU hotspot reaches 105°C
Hey there, I see you're dealing with a tricky situation with your MSI RTX 3080. When you play intense games, the hotspot temperature spikes to over 105°C, which is definitely pushing things. The fans seem to be struggling to keep up, and it feels like the system is trying to cool itself aggressively. Your main concern is whether this high temperature could damage the GPU or affect memory temperatures, especially since the hotspot is already so hot.
You mentioned adding more fans—two 140mm intake and three 120mm exhaust fans—which helped a bit, but the hotspot didn’t drop much. After some research, you found advice about replacing thermal pads and paste to lower the hotspot. However, some people reported that changing these components actually increased the hotspot temperature, which is worrying.
Since you don’t overclock and clean dust regularly, here are a few thoughts:
- Double-check your cooling setup—ensure all fans spin freely and there’s no blockage.
- Consider checking the thermal paste quality; if it’s old or degraded, replacing it could help.
- If possible, try cleaning the GPU heatsink thoroughly to remove any dust buildup.
- Monitor temperatures closely during gaming to see if adjustments are needed.
Let me know if you want help with specific steps or parts to inspect!
This same question gets asked quite often. Here is your best option.
The substantial distance between the GPU core and hotspot suggests a contact problem between the die and the cooler's cold plate, likely due to: significant GPU sag, improperly reassembled cooler, thermal paste issues, or thermal pads that are too thick or rigid. If you haven't disassembled it, ignore this note.
I've only heard about GPU sag recently after checking online and seeing the hardware. I plan to make a simple DIY support to straighten it out and test if it helps. Regarding the paste and pads, I haven't opened the card before, so I'm using the standard materials. I was thinking of getting a Noctua NT-H1 3.5g and a Thermalright Thermal Pad 12.8 W/mK. What should the pad hardness be?
NT-H1 can fail in some bare die setups, so it may not be reliable here. NT-H2, Gelid GC Extreme, Prolimatech PK3, Kingpin KPx, and Cooler Mastergel Pro V2 have proven effective on bare dies. I’m not sure if there are more options, but I’m not aware of them. You don’t have to inspect the pads unless you damage them, and even then… adjusting pads can be frustrating. You might find pads matching the standard size, but they could still cause problems since the new ones may be too soft or too rigid. "70-80C GPU Memory Junction works well, and it won’t be a problem until temperatures reach 95C or higher."
I added some NT-H2 thermal paste and swapped out four thermal pads near the chip. The temperature drop was noticeable, staying under 86°C after the adjustments.
Yes, the problem with my Zotac 3090 is clear. Core temperatures reached around 88°C, while the hotspot hit 105°C. After a repair, the core temperature dropped to 72°C, but the hotspot remained at 105°C and memory stayed unchanged.