I'm not confident that adjusting the voltage on my GPU fixed the issue.
I'm not confident that adjusting the voltage on my GPU fixed the issue.
I had some issues with my GPU temperatures. My Gigabyte RTX 3070 Ti was getting up to 85ºC during intense games, such as BF6, Hogwarts Legacy, and World of Warcraft in open areas (like Suramar), when all graphics were set to maximum on the Max settings. My aim was to maintain around 60FPS minimum, with a bonus at 1440p. The room temperature inside my case stayed cool, not hot at all, even near the GPU.
Problem 1:
I observed it reaching 85ºC only, never exceeding that level, which was unusual and raised concerns. As a result, the fans ran at full speed to try to cool it down, but it didn’t help much. My PC became extremely loud during gameplay, which was quite bothersome...
Problem 2:
Due to this, my PC would shut down automatically. Not because it failed completely, but because I consistently saw a black screen, all fans at maximum speed (even noisier than before), and no response whatsoever. The only solution was to physically power it off by pressing the top button.
The "solution" I discovered was a video from two years ago on YouTube, by a user named ImWateringPSUs, titled:
Undervolt your RTX 3070 for more FPS! - Tutorial
I had never tried overclocking or undervolting before, as I’m new and worried about damaging my hardware. The costs were too high. But I didn’t find any comments saying "Sorry, but that didn’t work for me." So, I decided to give it a shot. The video was straightforward and easy to follow. It worked! I tested it on all the games I mentioned, and the highest temperature reached was 67ºC during heavy play. That’s pretty good. But now I’m not sure if this truly fixed my issue. And if it did, what exactly happened? How can I explain this? What was going on and why was it happening?
After reading numerous discussions across various forums, many people mentioned the PSU. My unit is a Corsair 750W 80 Plus Bronze from 2019, which everyone said was decent but not the best. It’s normal for it to start failing over time. Before I started undervolting, I had replaced it with a new one—a Husky 850W 80 Plus Gold, which has received many positive reviews. I expect it to arrive soon, so I can’t test it yet.
Did I waste my money? Did I make a good decision by replacing the PSU? I’m really confused now.
It's also likely that the paste and/or pads on the card are worn out. Since there are no temperature sensors on the video card VRM, it's difficult to confirm the issue without disassembling it. From my experience, a black screen with all fans running at maximum speed usually points to a power delivery problem. To fix this, you'd need to be able to open the card and possibly replace the paste or pads yourself.
Replacing the power supply won't solve the core problem. The 85°C GPU temperature indicates it's overheating, likely requiring new thermal paste. The card probably throttles under heat, and turning the fan up to compensate is why it gets loud. The abrupt shutdown suggests the GPU is trying to prevent damage. Undervolting offers only temporary relief and risks long-term harm if not addressed properly.
Other sensors not discussed include the hot spot (hottest sensor report) and memory junction (Vram). Use tools such as hwinfo or GPU-Z to check them. The main report reflects an average value. TPU Vbios Collection: 3070Ti. Operating at 85C isn't sufficient to halt the GPU, especially since the core limit is 90C. Hot spot temperatures can reach up to 110C*, while Vram stays within safe limits around 95C. [Information gathered from temperature discussions and forums] This situation might also connect to: Spoiler View: https://www.youtube.com/watch?v=wnRyyCsuHFQ Power supplies are degrading, reaching their peak in Year 1, with worsening performance each subsequent year. TL;DR: The issue seems linked to both temperature and power, possibly requiring a replacement PSU (which is already in progress).
I don't know if you guys are gonna read this, here's an update:
So, my new PSU has arrived, a friend of mine helped me installing it and... nothing changed! The "problem" was still there after getting to 85C temperature. So everytime I boot my PC, I have to open MSI Afterburner so I can undervolt my GPU to stop the problem. I have also done the same thing to my CPU inside the BIOS config and managed to lower its temperature as well (not that this was a problem).
Idk whats gonna be my next step, cause the prices of GPUs nowadays are way too expensive. I only play games in 1440p! I have the 3070ti paired with a R7 5700x3D - 32gb RAM 3200mhz and I can run every game that I have at 60fps MINIMUM. The prices of a 9070xt and the 5070ti are too pricy here, so I think I'll have to wait for the next gen while I try and save some extra money
🙁
Here are the key points you want to remember:
You can adjust MSI Afterburner settings in several ways: save them, enable startup launch, and ensure they load automatically when Windows starts. I’ve used these methods for years with MSI Afterburner.
You noted that swapping your 3070 Ti for a 5070 Ti is costly; that’s reasonable, but you don’t need the higher-end model if you’re satisfied with your current performance at 1440p. If you just want a GPU without the issues of yours, consider a 5060 Ti 16 GB. It’s nearly half the price of a 5070 Ti and offers more VRAM than your 3070 Ti while being more efficient—about halfway between the 3070 Ti and 3080 in performance, but with more VRAM.
Consider this: a 5060 Ti would be a) stronger than your current GPU, b) provide TWICE the VRAM (16 GB versus 8 GB), and c) operate much cooler thanks to its 180 watt TGP (your 3070 Ti uses 290 watts).
You're already experiencing signs that your GPU might be overheating and could benefit from a refurbishment. It's a straightforward process if you handle it carefully, especially when removing the heatsink from the board. There are video guides demonstrating this for Gigabyte cards, and the steps are largely consistent for the 3070, 3080 models, or other brands as well.
The 3070 should maintain temperatures between 75-80°C on the GPU, 85-90°C at the hotspot, and stay under 15°C above the GPU temperature, with a maximum of 105°C for VRAM.
The RTX 5060 TI 8 or 16GB versions are only about 5-7% faster than the 3070 TI. The main advantages of purchasing them would be increased VRAM capacity, improved DLSS 4 performance (though it may run slower on RTX 30 series), and reduced power consumption. However, these benefits likely don't justify the expense of refurbishing your existing GPU.
The 3070 Ti remains a solid choice, but I wouldn't recommend upgrading unless you're looking for a significant jump—such as an RTX 5070 or RX 9070. For now, it's best to stick with what you have unless it's close to a major upgrade.