Recent measurements show elevated hot spot temperatures around Vega at a distance of 56 parsecs.
Recent measurements show elevated hot spot temperatures around Vega at a distance of 56 parsecs.
Hi all, I've noticed that the hot spot temperatures on my Sapphire PULSE Vega 56 card are higher than they probably should be. I'm not super familiar with the "safe" temperatures for silicon, but as far as I can tell it shouldn't exceed about 100 degrees Celsius or more than 10-20 degrees Celsius over the rest of the card. I've noticed that pretty much immediately after opening a game, my hot spot temps jump to 105 degrees while the rest of my card stays at around 50 degrees. I bought this card used and the VBIOS has been flashed to a Vega 64, but I've disabled this by flipping the vbios switch and have confirmed it runs the Vega 56 VBIOS full time now. I've been having hard crash issues with this PC for a long time now and am hoping that fixing this issue will finally solve them. I've replaced the entire system outside of the GPU and PSU, so it must be one of the two and everything is pointing me towards the former. What can I do to bring hotspot temps down? I really, really do not believe I have the technical expertise to completely disassemble the card so I'd prefer to use that as a last ditch effort. I have tried to undervolt it using guides online, but every time it hard crashes it resets the settings I've applied in Radeon Software. This happens 5-10 times a day so unless there is a way to get those settings to stick around, I'm not sure that's a solution either. Additionally, I tried to undervolt and monitor the temps but they remained the same. The only thing that appeared to help was manually adjusting the fan curve to kick the fans up much higher earlier in the temperature range. Unfortunately, this was also reset on a hard crash. I have noticed that the fans do NOT spin up faster before a crash, but after the screen turns black, the fans begin to spin at 100% and the entire case gets really hot. It's like the GPU is continuing to work after the crash. I'm pretty scared of what would happen if I wasn't at the system and it continued to run like that. Thank you all so much! This forum is a really valuable resource and I'm hoping I can find some answers (finally). PC Specs: https://pcpartpicker.com/list/MHPVqp Other info: GPU drivers up to date (23.4.3 at the time of posting) Windows 11 Pro version 22H2
It seems like the project will require taking it apart. The process isn’t as complicated as it looks. I’m still new to this (completed 4 disassemblies and reassemblies) and with the proper tools—a suitable screwdriver, a clean cloth, and some high-quality IPA and thermal paste—it’s just about being patient. For minor spots, a few cotton swabs worked fine. The biggest challenge was remembering where each part ended up, so I used a cookie sheet and tape to organize everything. The first attempt took roughly 40 minutes, including the time it took to remove a stubborn screw that lingered for about five minutes on its own.
What are the long-term dangers of skipping thermal paste replacement? I have someone more experienced who might be able to assist, though it could take weeks to months. Also, can I safely reuse leftover Noctua CPU cooler paste on the GPU die since it lacks an IHS like a CPU does?
The hotspot throttle temperature appears to be 105°C. No references for the maximum hotspot temperature were located.
Avoiding thermal paste replacement could harm your graphics card over time. You can use any available paste. Noctua, Arctic Mx-4 and similar brands work well. It depends on your GPU and how easy it is to remove. For example, the Radeon RX 580 is straightforward, while the Radeon 6000 series is more challenging. I own a Radeon 6700 XT and tried to take it apart but couldn’t. I planned to change the paste, but others advised against it. I watched YouTube tutorials, though they were difficult to follow.