Hi,
For some weeks, I had a fan doing loud rattle on my gfxcard, a 3060ti, so I just bought new fans direct from Shenzen, branded from the same company. They spin well under laboratory power supply and on PC’s start.
Man can not know everything and I was totally ignoring things have considerably evolved in terms of thermal conductivity’s usage for the recent GPU, with the use of thermal pads instead of thermal paste, which I used after a cleaning of the GPU. My mistake. (Hopefully I used non conductive thermal paste, the noctua’s one.)
PC was booting ok, in VGA mode, but as soon as I launched Xorg - I use Linux - things went dark, moreover: some sort of pink rasters.
I did a shutdown and restart, the ASUS logo of my motherboard was well displayed, but instead of the normal black background, it was made of the same pink sprites pattern in tilling mode…
I bought then good thermal pads, but the pink things remained, and unable to launch Xorg, only VGA was going well.
Trying to get infos from nvidia-smi gives nothing, except a shot ms “flash”’ on the screen, and Xorg.log complains having no gpu found.
I requested the GFXcard company’s support to get exact thickness to use => they answer they aren’t allowed to communicate about “industrial process”… Thermal pads the boredom.
Here is the long story.
My question is simple, if by chance any of you reading this have long experience in GFXcards troubles, can you give your point of view, is the card dead or just overheating because of wrong pads thickness ?
Or is it the sign of, by instance, one memory chip, even worse, the GPU itself having fried ?
I’m no GPU “repair guru,” but I know that a pink screen can be caused by factors other than the GPU. For instance, a poor or bad connection to the monitor can lead to this issue. Although not very common, faulty RAM module(s) can also be a culprit. I recommend ensuring that the cable you are using to connect your GPU and your screen is in good condition and securely inserted into the appropriate ports. If you have another compatible GPU, try using it instead of the “faulty one” to see if the pink screen issue persists.
Furthermore, if you have a CPU with a built-in GPU, you can test your other components by removing the GPU module and plugging your screen cable into the appropriate port on your motherboard. Once you’ve confirmed that it’s not your other PC components causing the problem, you can focus on the GPU, knowing that the issue is likely related to it.
For a 3060 Ti, some people recommend using 80x40x1.5mm thermal pads for VRM, VRAM, and the sides. However, others suggest using 2mm pads specifically for the VRM. It’s important to note that neither Nvidia nor GPU manufacturers provide these recommendations; they come from people who have successfully re-padded their GPU cards.
One crucial point to emphasize is: DO NOT USE THERMAL PADS INSTEAD OF THERMAL PASTE WHERE THERMAL PASTE IS NEEDED. I do not recommend this. You can use a thermal paste designed for CPUs on a GPU. I’d recommend Thermal Grizzly Kryonaut or, at the very least, Arctic MX-4.
Lastly, regarding the fans, since you mentioned that you obtained the same model and brand, and that they are new, from the original manufacturer, I assume they should be fine.