What caused the sudden high GPU temperatures?
What caused the sudden high GPU temperatures?
Hi, I own an RTX 5070 Gigabyte gaming OC model. I was quite satisfied with it. I acquired it five months ago and played some games on PubG and a few single-player titles for about two months. From what I recall, the temperatures never exceeded 61°C and the fans stayed below 30 percent. For the last three months, I didn’t play any games because I was busy with other tasks, but I still used the computer for browsing and office work.
Today, I decided to play Warframe. I noticed that my GPU fans ran at very high speeds. After checking MSI after burner, my GPU temperatures were between 78-83°C even without missions. When I looked at the fan speed in the MSI afterburner, it showed 30%, but that seemed incorrect. Another point is that when the GPU was warm and the room temperature was high, I played those games while the ambient temperature rose. Now it’s almost winter, and the room feels much cooler. The temperatures have stayed consistent, not spiking.
I uninstalled the graphics driver using DDU and reinstalled it. Now the fans don’t run at full speed continuously. They stay quiet most of the time, but occasionally they reach full speed for a few seconds every couple of minutes, especially when I switch tabs in the game.
Is there any explanation for this?
Pc specs:
CPU: i7 14700KF.
RAM: Gigabyte B760 gaming DDR4 with the latest F12 BIOS update.
Memory: 128GB DDR4 3600MHz Adata Gammix 4x32GB.
PSU: Seasonic G12 GC 750W Gold+
Storage:
-2x 2TB Adata Legend max M2,
-1x Seagate Barracuda Q5 2TB M2 SSD
-2x Adata SU650 1TB SATA
(all storage units are under 50% capacity)
Peripherals:
-Mouse: Asus Rog Gladius 3
-Keyboard: Genius Slimstar (a 12-year-old black slim office keyboard, exact model unknown)
-Webcam: Microsoft LifeCam HD 3000
-Speakers: Two Genius speakers connected via a single cable. Model is Genius 2.0 SP-UDL. They are linked through one USB and one jack connection.
Monitor: LG UltraGear 32GB QHD 32-inch.
Operating System: Windows 11 Pro x64 24h2.
Everything is connected via cables, including the internet. No Wi-Fi used.
All components are less than a year old and were purchased new by me.
PC case: Quiet silent base 802 (non-window version)
the problem with these cards is the gpu paste is usually terrible, and there are other issues if you search for thermal paste problems in the 5000 series gigabyte you'll find they have issues with older batches of cards leaking thermal paste or gel.
I examined the GPU from every angle and confirmed there are no leaks. The GPU was mounted in its usual position, not tilted vertically. Could there be a chance the paste between the GPU and CPU deteriorates over time and requires reapplication?
Please update your post with complete hardware details and operating system information. Add PSU specifications including make, model, wattage, age, and condition (original, new, refurbished, used). Specify disk drive details such as make, model, and capacity, along with current usage. Provide a list of all connected peripherals. Indicate whether the network connection is wired or wireless. There could be additional concerns, so understanding the full build and setup is essential.
Check if custom configurations are active in Afterburner and confirm temperatures/speeds using HWinfo or a similar tool.
theres just bad quality paste that can pump out and cause heat spots if you download hw monitor it will show you the heat spot which is what i go by if your heat spot is close to 90-100c then it will need repasting
Usually this would be the step to disable, but Nvidia has disabled the hot spot sensor for the 50 series cards, making it no longer feasible. With the problems Gigabyte experienced, I wouldn't be surprised if cooling materials were failing. It could extend beyond just the TIM on the GPU—possibly pads or putty in the VRAM/VRM as well. Determining the exact cause would be difficult without disassembling the card.
That's quite unusual how people think about identifying if a GPU is properly connecting. I'd recommend using Putty because those pads are extremely problematic when the cooler isn't level, and most coolers aren't model-specific. Sometimes they reuse a cooler from another card, which saves effort if you just place a rubber dimple underneath—something some ASUs have implemented.