PowerColor 7900XTX Red Devil: issues with stock TIM application
I purchased PowerColor 7900XTX in January 2023. Card is a performance beast, however PowerColor might have messed up the paste application on at least early batches.
Issue description
Since early 2023 there was a lot of reports of Red Devil cards having very high delta between GPU temperature and Hot Spot temperature (around 30-35 K) 1 2. Unfortunately, my GPU was also affected by this. I must admit I did not even realize that this is the case in the beginning. When I got the card I tested it quickly with Furmark and the delta between GPU and Hot Spot was perfectly acceptable 20 K. However I did not realize that, when Furmark is running, GPU goes into some kind of throttle mode (GPU Clock is under 2 GHz!), so it does not show full picture. Few weeks later I saw Hot Spot hitting 100 °C when playing DOOM Eternel and I knew something is not OK.
I have contacted PowerColor, but they claimed that such big difference between temps is normal. Well, I did not agree and was considering repasting the card myself (that would mean voiding the warranty on over 1k EUR GPU, which I was not keen on). Luckily around that time Igor’s Lab did a great job describing the issue in detail 3. This actually forced PowerColor to react, after couple more mails they accepted the fact that temps on my card are not OK and I RMAed the card (I must add that the RMA process was pretty quick at least!).
Measurements
Before sending card for RMA among I logged all GPU parameters during Fire Strike Ultra and Time Spy Extreme Stress Tests. Stress Tests were set to default of 20 loops each. Ambient temperature in both cases was almost identical (if I can trust my super cheap kitchen thermometer). Card was warmed up before the runs and “OC Bios” was selected in both cases.
Test setup
- Ryzen 9 3900X
- ASRock X570 Taichi
- GSkill F4-3200C14D-32GTZKW (2x16GB B-Die)
- Samsung 980Pro 1TB (system)
- PowerColor 7900XTX Red Devil
FireStrike Ultra Stress Test
| Parameter (average value) | before RMA | after RMA |
|---|---|---|
| GPU temp | 67.3 °C | 74.6 °C |
| Hot Spot | 99.3 °C | 95.8 °C |
| GPU clock | 2381 MHz | 2426 MHz |
| Fan speed | 1337 rpm | 1286 rpm |
Time Spy Extreme Stress Test
| Parameter (average value) | before RMA | after RMA |
|---|---|---|
| GPU temp | 67.5 °C | 74.9 °C |
| Hot Spot | 97.7 °C | 95.2 °C |
| GPU clock | 2466 MHz | 2495 MHz |
| Fan speed | 1313 rpm | 1276 rpm |
Conclusion
From the numbers above we can see that RMA/repasting the card definitely did something. While the Hot Spot temps did not drop significantly, card is boosting higher, so that is something. I am still not 100% happy, I did not expect such expensive GPU to have a TIM application issue in 2023, this is unacceptable in my opinion. I will think twice before purchasing something from PowerColor again.
The thing that surprises me most in this story is that, there was almost zero reaction on this obvious issue from most of the HW news sites. Only smaller sites linked Igor’s article on main pages, in a world where every slip up from manufacturer is immediately top news everywhere, this is really unexpected.