|
June 15, 2013, 06:40:45 PM |
|
I think I'll post my experience with my reference HIS 7970. Overclocked to 1125 MHz core at 1.2 V, and 1750 MHz memory at 1.6 V, I was getting core temps of 85C and VRM1 temps of 91C while Bitcoin mining, with an ambient temperature of 25C and a fan speed of 100%. VRM2 temps were a nonissue. AMD rates their GPU cores at up to 85C. So if GPU-Z and HWInfo64 was wrong, and the temperature was actually higher, I'd be in hot water. So I decided to apply new thermal paste, Antec Formula 7, to my Radeon HD 7970. I ran into problems, two screws wouldn't come out of their screwholes. Due to that, I had to buy three small straight-tipped screwdrivers to try to take it out. One came out easily after that, the other took some work to get out. By this point, I scratched the PCB by accident, exposing a small amount of copper. I had removed the shroud from the card to expose the heatsink itself, and it had a lot of dust in it. A little compressed air later and most of the dust was gone. Anyway, it took some work to remove the cooler from the PCB. Once the PCB was separated from the cooler, I went to work. I used 91% isopropyl alcohol to remove the thermal pads from the memory chips and the VRMs, and the thermal paste from the GPU itself. Then, I applied a line of thermal paste over the VRMs, and dots of thermal paste over the GPU and memory chips. Then, I reinstalled the cooler to the PCB. At this point, I was hoping I didn't kill my Radeon HD 7970 with the PCB scratch. Luckily, the PC successfully went through the POST process, so, yay me. I went and quickly tested my GPU through Bitcoin mining. At the same clocks, 1125 MHz on the core at 1.2 V and 1750 MHz on the memory at 1.6 V, with the same ambient temperature of 25C and the same fan speed of 100%, my core temp went down to a maximum of 72C, and my VRM1 temp went down to a maximum of 76C, for an improvement of 13C and 15C respectively. Now I can run my card overclocked without fear of burning it up.
|