Hey, when idle the power draw is about 55w which seems very high. I’m using a dual 1080p 144hz monitor setup, with one of the monitors reporting as 143.61hz which also has g-sync. The “PowerMizer” in nvidia-settings seems to be stuck at level 4 which is probably what is causing this.
However, when I set both monitors to 60hz the power draw gets reduced to 37w and the “PowerMizer” goes back down to level 0, but 37w still seems high when idling.
The high power draw is causing the fans to constantly spin up to high speeds and then shut off again, which is annoying the hell out of me.
nvidia-smi with both monitors at 144hz
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.57.02 Driver Version: 470.57.02 CUDA Version: 11.4 |
|-------------------------------+----------------------+----------------------+
| 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 NVIDIA GeForce ... Off | 00000000:08:00.0 On | N/A |
| 0% 57C P0 55W / 215W | 311MiB / 7979MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 962 G /usr/lib/Xorg 133MiB |
| 0 N/A N/A 1009 G /usr/bin/gnome-shell 175MiB |
+-----------------------------------------------------------------------------+
nvidia-smi with both monitors at 60hz
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.57.02 Driver Version: 470.57.02 CUDA Version: 11.4 |
|-------------------------------+----------------------+----------------------+
| 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 NVIDIA GeForce ... Off | 00000000:08:00.0 On | N/A |
| 0% 54C P8 37W / 215W | 353MiB / 7979MiB | 9% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 962 G /usr/lib/Xorg 146MiB |
| 0 N/A N/A 1009 G /usr/bin/gnome-shell 203MiB |
| 0 N/A N/A 30877 G nvidia-settings 0MiB |
+-----------------------------------------------------------------------------+
If I set any of my monitors (I have 2) to 144hz the power level will not drop below 4. If I disable the secondary monitor the power level will drop normally with the primary on 144hz but will remain on level 4 if that second monitor is connected even at 60hz.
If I turn my monitors to both 60hz refresh then my power level also drops normally.
In effect I can completely verify the OP’s issue with the added test of turning off one monitor.
This seems to be a regression as I remember a similar issue a while back (maybe drivers 395?) where if you had both monitors set at 144hz the power level would stay at 4 but if you turned the secondary to 60hz the power levels would drop as normal under light loads.
I am using a more recent driver that the OP so the issue has not been resolved with the newer driver.
I am using Arch and an using the drivers from the Arch repos. Linux kernel 5.13.13-arch1-1
this is nvidia-smi under no load, just the desktop and a few app running but nothing that would not normally have the power level dropping down to 1
The power level should be at one and the temp should then drop to about 30-40C. With the power level stuck at 4 the temp stays at around 50C under the same conditions. The fans do work as they should when the temp rises a bit but the card should not be maintaining this temp unless I have a game or other graphically demanding program running
Are the devs currently working on a fix so that the next update may include a fix for this issue and if so do we have an eta and driver version of the potential fix?
Others have mentioned this is intentional; probably because there are too many pixels to shove through the pipeline
But that makes no sense because this wasn’t an issue with his 3080 card, and it’s no issue on Windows either.
I just wish if it were intentional Nvidia would at least comment on one of these threads and tell us, or if it’s a bug that they acknowledge it exists and are working on it.
Upload nvidia-bug-report and hope this thread gets noticed.
Communicating with NVIDIA in regard to bugs has always been fucked up. They have a special linux email address as well, but I’ve sent a number of emails to it and haven’t received a single confirmation or reply.
Same problem here.
My GTX 1080 sits almost all the time at P0 consuming 48-50W when doing nothing.
It sometimes drops to lower power modes but it lasts only a few seconds before ramping up to P0 again.
Would love to have a fix to save some energy and have my card working in passive mode (fan 0%).