Replacement GPU heat

Andrew Buck

Honorable
So one of my cards started crashing my PC a few months ago. I ruled out the card and used only one for a while. I suspected a capacitor being dead, and it was true. I sent it in to MSI after a month of laziness. A different cars came back. No crashing, except for it runs a whole 30 C or even more hotter than my current one. That's right, basically an R9 290x. My good one runs at 54 C 100% fans, this one gets up to 89. Should I send it in again? The only deal is I still have a warranty. I have some Arctic Silver thermal paste laying around, should I just go and paste it up and see what happens?

Also a few things: I am away on vacation right now and won't be back home to my PC until probably Saturday or Sunday. Today (unlikely) my new GTX 970 is going to arrive. I bought it today with next day shipping, but it was Christmas, so I can't expect that. Probably Monday.
Lastly, I need to sell this card along with the other soon and will not sell it with a thermal problem.

No rush right now. I have time. Going to sleep now anyways. Merry Christmas.
 
i personally have never seen a good thermal paste application on a gpu or a laptop. that said, i wouldn't be removing the stock air cooler until i know for sure that it wont void a warranty. but as far as a "botched" tim application, i would consider that fairly uncommon. though msi is known for having bad tim jobs, but check your warranty, some use stickers on screws and they will know if you have tried to take apart the card which could be an issue for warranty.
 
If you have 2 running, and they have no PCIE spacing between them, then the top card will be almost choked out. Check that both fans (if applicable) are spinning. I have a problem with my MSI 7950 and one of the fans doesn't spin until I crank up my fan speed to 85% or higher. It got up to 94 C once, and was still playing Batman Akham City just fine O.O
 
The fans were at 100% in all of my testing to rule out the fan. Nothing wierd. As for 17seconds' answer, that is an interesting idea. I will have to check and see. I don't recall seeing voltages much higher and I definitely think that a little more voyage would add such an immense amount of heat. For reseating the heatsink, wouldn't that void my warranty? I still have a little bit left and I am planning to sell it. Voided warranty doesn't seem like a good selling point.
 
Wait a moment - I just noticed something. Not sure if this is a false reading or not, but GPU-Z shows that the good card (@100% fan speed) is at 3000ish RPM. The new card seems to be at 100% as well, but at 1000 RPM. It goes up a little bit when I lower fan speed for some reason. Seems like a fan problem. I will look into it. The Voltage seems to be perfectly fine. My cards used to stay around 1.206 V, this one jumps all around in the 1.1 range. Not sure if this is some sort of custom BIOS doing this stuff, but it is very strange.
 
I noticed that the BIOS was different for each card, so I flashed the new card's to the newest BIOS (my old card's BIOS). No help. Can't wait for my 970 to come tomorrow. My worry is selling it. Not interested in sending it back AGAIN.
 


Found out - fan stuck, and the back of the GPU looks burnt orange. Sending it back. Now I am having issues with my new card. On the latest drivers even at stock, on Battlefield 4, I get a device hung error. This happens after a while of playing and then the game will keep saying that on startup after until I restart. Watch Dogs keeps crashing as well. I cannot seem to get stable in benchmarks and stress tests, either. Right now, I am on second to newest driver (seems to be more stable) with only +50 MHz/+225 MHz. Anything past that gets worse. Any help? It is a Gigabyte GTX 970 G1 Gaming, and I would expect much better than that.