Temps de 6950xt
Temps de 6950xt
Hello, I’m dealing with some issues regarding the temperatures on my GPU (PowerColor Red Devil 6950XT). It’s been getting quite warm since I purchased it last year. I’ve read that newer GPUs tend to run hotter now, and even reaching 90°C is considered normal. Despite monitoring the temps, I’ve recently seen readings above 100°C, especially during intense scenes in Baldur's Gate 3 and Starfield, where the hotspot hit 107°C and 105°C respectively. I haven’t overclocked it; in fact, it’s been undervolted. I’ve tried adjusting the BIOS settings and manually lowering the voltage, but the results haven’t changed much. I’m playing on a 4K display and have experimented with limiting FPS to 60 and allowing unlimited on Starfield, yet still experience these spikes. On Baldur’s Gate 3, locking FPS at 60 gave me 107°C with 60 FPS, but the game would drop below that. My setup has good airflow, using a Corsair 680x with three front fans, two bottom fans blowing directly into the card, one back fan, and two top fans for the AIO cooler. All fans are running at full speed, adjusting automatically based on GPU temperature. The card is large and heavy, so I use a stand to prevent sagging. My power supply is a Corsair 1000W unit. If you need more details, just let me know.
This refers to a general topic or model, not a specific variant.
These cards operate efficiently with a hotspot up to 110°C. Adjustments can be made by raising mounting pressure when using washers under the retention bracket—just watch out for potential GPU die cracking if you apply too much force. I've managed to lower temperatures on my 6800XT by nearly 20°C using the Thermal Grizzly Carbonaut pad; next time I'd consider Kryonaut. Edit: Your card isn't undervolted even with the voltage slider in AMD Adrenaline Software—it only alters the frequency/voltage curve, but the card still hits its maximum voltage at peak frequencies.
I completed the task within the warranty period. I didn’t test the auto UV utility, unsure about its functionality. To truly limit your maximum voltage, you’ll need to retrieve the BIOS file using GPU-Z and import it into More Power Tools. There you can adjust core, SOC, VRAM voltages, modify core, SOC, VRAM, Infinity Fabric clocks, and more—including power limits. It doesn’t actually update the BIOS on the card; any changes will disappear after a driver update. Be careful with what you do, as overdoing it could brick the card. I managed to reduce the card’s usage by 100W while achieving the same results as before MPT.
Various countries have different rules, but in the US companies aren't legally bound to stop you from taking off your GPU cooler and swapping it out, even with warning labels. If you're monitoring hotspot readings instead of core temperatures, 100-105°C during gaming is typical. My Radeon VII handled it well, supporting up to 115°C and occasionally reaching that point before I submerged it. It never caused any problems with the card—though I eventually removed it because of driver issues, not hardware faults.
I'm checking the hotspot readings. The typical GPU temperature is generally 10-15 degrees less than this.