I recently noticed that my GPU is stuck at P0. I’ve gone through a bunch of the posts on this subject and have spent time on this before. I made an important discovery.
With 2 monitors, it’s fine. It quickly ramps down to P8 within 15 seconds. As soon as I plug a 3rd monitor (I have 4 monitors), it immediately goes into P0 and never exits P0. It will fortunately go into P8 once the monitors turn off to save power via DPMS. I ssh’d into my workstation and verified this!
The power draw varies between 35-50W which is really excessive
OS: Debian GNU/Linux unstable
Kernel: Linux 5.10.9 (stock Debian kernel)
Nvidia: 460.39
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.39 Driver Version: 460.39 CUDA Version: 11.2 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 GeForce GTX 1080 On | 00000000:0D:00.0 On | N/A |
| 7% 52C P0 37W / 215W | 1207MiB / 8110MiB | 3% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
I’ve verified the power draw + multi monitor behavior without any processes running in nvidia-smi. The above snippet from nvidia-smi indicates stuff running, but I have definitely verified this behavior without anything running on it. Furthermore, running xset dpms force suspend
allows the GPU to go into P8. Wake the displays back up, and it’s in P0.