Hi everyone. I don’t think I was able to find this exact issue on the EndeavourOS forum anywhere else, so here goes.
I have had worse battery life ever since I reinstalled EndeavourOS with GRUB+BTRFS, whereas before I was using systemd-boot+ext4. I had EndeavourOS installed alongside Linux Mint (and Windows, sadly) and decided to take Mint out of the equation, so I deleted my Endeavour and Mint installs and created a new EndeavourOS install in the place of both.
Powertop says that I’m discharging at ~15-18W on idle with my new install, and my old install (the one with systemd-boot and ext4, that is) gave me something more like ~5-8W. I get even higher consumption when Baloo is indexing files (I use KDE), and at first I thought that was the issue because my previous install didn’t have all my documents on the system partition, but the discharge remains well over 10 watts even with the file indexing complete.
To make matters more perplexing, it looks like my Nvidia card is active even when I’m not trying to use it. While nvidia-smi doesn’t show any active processes, I’m getting an “active” status when I run cat /sys/bus/pci/devices/0000:01:00.0/power/runtime_statusper the Arch Wiki’s NVIDIA Optimus page. Powertop’s output is missing processes and currently lists “-nan” for various key stats, but it is quite interesting that it shows my Nvidia GPU at 100%.
I could try go through all the tips from the Arch Wiki, of course, but I had a good setup on EndeavourOS before, so I figured I would ask the nice people here for advice first.
Oh, if it’s relevant, I made the somewhat questionable decision of copying various KDE related configs (i.e. ~/.config/{k*,KDE}) from the backup of my old install, but I’m not sure if that would make any difference, honestly. Also, I did my old install with the Titan installer and my new install with the Titan Neo installer.
Installing EndeavourOS using Btrfs enables zstd compression by default. It’s largely imperceptible performance wise, but it would require additional CPU overhead, which in turn requires more power.
You can disable Btrfs compression from now on (existing data remains compressed until re-written), if you so choose, by removing the compress=zstd flag from your Btrfs mounts in /etc/fstab.
That won’t fully resolve the compression issue, as it’ll still require CPU overhead to de-compress the existing data on access, but it should improve over time as packages are updated and data re-written.
The compression idea sounds plausible. My first thought was that it had something to do with the filesystem itself, as that and the bootloader are the only things I changed deliberately. I guess I could try installing without zstd compression, but that sounds like a lot of work.
I wonder if I’m misreading the Nvidia stuff also, somehow… It would be good to see if I can somehow get more data, but again Powertop is acting funky.
With respect to your Nvidia considerations, unfortunately I’ll need to defer to others who have more experience with managing discreet Nvidia GPU’s on Intel mobile platforms, with Optimus.
That makes sense. Thank you. Am I right in thinking that I posted in the right place for Nvidia stuff? Perhaps I ought to have added a tag or something?
My thoughts on this are still more peripheral, as I’m not able to be Optimus specific. But I wanted to check with you, might you have been using X11 before, and Wayland now?
Wayland and Wayland enabled applications will lean much more on the GPU for acceleration, perhaps loading your Nvidia more.
I thought I’d give it a bit of a test. I copied a development directory which is large (92GB), but full of various file types, many of them compressible, between my two Kingston KC3000 M.2 SSD’s.
Both drives are using Btrfs with default zstd compression. Just visually monitoring with btop, the CPU usage jumped from idling at about 1%, up to around 15%. The CPU’s power consumption jumped from around 45W to over 100W. This is using a desktop AMD Ryzen 9 5900X CPU.
Part of that load is possibly the speed of copying between two high-end M.2 drives, which entails both decompression and compression under high load. I’m not sure how that translates to regular usage, I dare say this is a more extreme example. I guess it would largely depend on the the tasks being performed. Seems at least noteworthy though.
Copying data uses CPU. For that test to be meaningful, you would need to compare it to copying data from one of the btrfs partition to a non-btrfs partition or vice-versa.
Of course, that is way different than seeing a jump of 10w at idle.
Great question. I just unplugged the laptop from the AC adapter and this is what I see under the Tunables tab after pressing “r” to refresh. I have the Balanced power profile enabled, by the way.
PowerTOP 2.15 Overview Idle stats Frequency stats Device stats Tunables WakeUp
>> Bad VM writeback timeout
Bad Runtime PM for PCI Device Intel Corporation Raptor Lake Crashlog and Telemetry
Bad Runtime PM for PCI Device Samsung Electronics Co Ltd NVMe SSD Controller PM9A1/PM9A3/980PRO
Bad Runtime PM for PCI Device Intel Corporation RST Volume Management Device Controller
Bad Runtime PM for PCI Device Intel Corporation RST VMD Managed Controller
Bad Runtime PM for PCI Device Intel Corporation Alder Lake-P Integrated Sensor Hub
Bad Runtime PM for PCI Device Intel Corporation Raptor Lake-P 6p+8e cores Host Bridge/DRAM Controller
Bad Runtime PM for PCI Device Intel Corporation Raptor Lake PCH CNVi WiFi
Bad Runtime PM for PCI Device Intel Corporation GNA Scoring Accelerator module
Bad Runtime PM for PCI Device Intel Corporation Raptor Lake Dynamic Platform and Thermal Framework Processor Parti
Bad Runtime PM for PCI Device Intel Corporation Raptor Lake LPC/eSPI Controller
Good Wireless Power Saving for interface wlan0
Good NMI watchdog should be turned off
Good Bluetooth device interface status
Good Enable Audio codec power management
Good Autosuspend for unknown USB device 3-10 (8087:0033)
Good Autosuspend for USB device xHCI Host Controller [usb3]
Good Autosuspend for USB device Integrated_Webcam_HD [CN0679GY8LG0031AC0WTA00]
Good Autosuspend for USB device xHCI Host Controller [usb1]
Good Autosuspend for USB device Goodix Fingerprint USB Device [Goodix Technology Co., Ltd.]
Good Autosuspend for USB device xHCI Host Controller [usb4]
Good Autosuspend for USB device xHCI Host Controller [usb2]
Good Runtime PM for PCI Device Intel Corporation Device a73f
Good Runtime PM for PCI Device Intel Corporation Raptor Lake PCI Express 4.0 Graphics Port
Good Runtime PM for PCI Device Intel Corporation Raptor Lake PCI Express 5.0 Graphics Port (PEG010)
Good Runtime PM for PCI Device Intel Corporation Device 51fb
Good Runtime PM for PCI Device Intel Corporation Raptor Lake-P Thunderbolt 4 NHI #0
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH-P SMBus Host Controller
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH USB 3.2 xHCI Host Controller
Good Runtime PM for PCI Device Intel Corporation Raptor Lake-P Thunderbolt 4 USB Controller
Good Runtime PM for PCI Device Intel Corporation Raptor Lake-P [Iris Xe Graphics]
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH Shared SRAM
Good Runtime PM for PCI Device NVIDIA Corporation AD107M [GeForce RTX 4050 Max-Q / Mobile]
Good Runtime PM for PCI Device Intel Corporation Alder Lake-P PCH PCIe Root Port #4
Good Runtime PM for PCI Device Realtek Semiconductor Co., Ltd. RTS5260 PCI Express Card Reader
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH Serial IO I2C Controller #0
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH Serial IO I2C Controller #1
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH HECI Controller
Good Runtime PM for PCI Device Intel Corporation Raptor Lake-P/U/H cAVS
Good Runtime PM for PCI Device Intel Corporation Alder Lake-P PCH SPI Controller
Good Runtime PM for PCI Device Intel Corporation Raptor Lake-P Thunderbolt 4 PCI Express Root Port #0
<ESC> Exit | <Enter> Toggle tunable | <r> Window refresh
I was tired of not getting any process info from powertop, so I switched from the extra repo’s version (2.15) to AUR’s powertop-git (currently 2.16-rc3). Now I have actual process info, and nvidia-smi processes appear close to the top. I just restarted my system, and the wattage is even more crazy than usual: 26.8W.
I’m sharing the overview tab. Since @cactux wanted to see the Tunables tab, I have included that also, and I’m seeing more “Bad” items listed than in Powertop 2.15.
PowerTOP 2.16-rc3 Overview Idle stats Frequency stats Device stats Tunables WakeUp
>> Bad Autogroup scheduling
Bad Enable Audio codec power management
Bad Intel energy performance bias
Bad Intel energy performance preference
Bad Intel energy performance preference
Bad Intel P-state minimum performance
Bad Tuning for default qdisc
Bad Tuning for scan sleep millisecs
Bad Tuning for sleep millisecs
Bad VM dirty ratio
Bad VM writeback timeout
Bad Runtime PM for PCI Device Intel Corporation Alder Lake-P Integrated Sensor Hub
Bad Runtime PM for PCI Device Intel Corporation GNA Scoring Accelerator module
Bad Runtime PM for PCI Device Intel Corporation Raptor Lake Crashlog and Telemetry
Bad Runtime PM for PCI Device Intel Corporation Raptor Lake Dynamic Platform and Thermal Framework Processor Participant
Bad Runtime PM for PCI Device Intel Corporation Raptor Lake LPC/eSPI Controller
Bad Runtime PM for PCI Device Intel Corporation Raptor Lake PCH CNVi WiFi
Bad Runtime PM for PCI Device Intel Corporation Raptor Lake-P 6p+8e cores Host Bridge/DRAM Controller
Bad Runtime PM for PCI Device Intel Corporation RST VMD Managed Controller
Bad Runtime PM for PCI Device Intel Corporation RST Volume Management Device Controller
Bad Runtime PM for PCI Device Samsung Electronics Co Ltd NVMe SSD Controller PM9A1/PM9A3/980PRO
Good Wireless Power Saving for interface wlan0
Good Batch size for KSM
Good Bluetooth device interface status (hci0)
Good NMI watchdog should be turned off
Good Tuning for cursor blink
Good Autosuspend for unknown USB device 3-10 (8087:0033)
Good Autosuspend for USB device Goodix Fingerprint USB Device [Goodix Technology Co., Ltd.]
Good Autosuspend for USB device Integrated_Webcam_HD [CN0679GY8LG0031AC0WTA00]
Good Autosuspend for USB device xHCI Host Controller [usb1]
Good Autosuspend for USB device xHCI Host Controller [usb2]
Good Autosuspend for USB device xHCI Host Controller [usb3]
Good Autosuspend for USB device xHCI Host Controller [usb4]
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH HECI Controller
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH Serial IO I2C Controller #0
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH Serial IO I2C Controller #1
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH Shared SRAM
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH USB 3.2 xHCI Host Controller
Good Runtime PM for PCI Device Intel Corporation Alder Lake PCH-P SMBus Host Controller
Good Runtime PM for PCI Device Intel Corporation Alder Lake-P PCH PCIe Root Port #4
Good Runtime PM for PCI Device Intel Corporation Alder Lake-P PCH SPI Controller
Good Runtime PM for PCI Device Intel Corporation Device 51fb
Good Runtime PM for PCI Device Intel Corporation Device a73f
Good Runtime PM for PCI Device Intel Corporation Raptor Lake PCI Express 4.0 Graphics Port
Good Runtime PM for PCI Device Intel Corporation Raptor Lake PCI Express 5.0 Graphics Port (PEG010)
Good Runtime PM for PCI Device Intel Corporation Raptor Lake-P [Iris Xe Graphics]
Good Runtime PM for PCI Device Intel Corporation Raptor Lake-P Thunderbolt 4 NHI #0
Good Runtime PM for PCI Device Intel Corporation Raptor Lake-P Thunderbolt 4 PCI Express Root Port #0
Good Runtime PM for PCI Device Intel Corporation Raptor Lake-P Thunderbolt 4 USB Controller
Good Runtime PM for PCI Device Intel Corporation Raptor Lake-P/U/H cAVS
Good Runtime PM for PCI Device NVIDIA Corporation AD107M [GeForce RTX 4050 Max-Q / Mobile]
Good Runtime PM for PCI Device Realtek Semiconductor Co., Ltd. RTS5260 PCI Express Card Reader
<ESC> Exit | <Enter> Toggle tunable | <r> Window refresh
I have never owned a hybrid laptop so I am not the “sharpest pen in the drawer” about this.
However about the results from powertop, you could turn all those “bad” values to “good” ones by running sudo powertop --auto-tune.
Now I don’t believe that this would dramatically change the power drain issue but worth a try.
Also, in order to get more accurate estimation from powertop , you may need to calibrate it first:
Lately, I have become a bit “obsessed” about power consumption of my laptops. They are what one may call in their middle age and the batteries are showing signs of fatigue. So I am trying to slowdown the degradation. In my case it is the combined WiFi and Bluetooth card that won’t suspend when idle or even when turned off.
Here are some command line that may be useful to you as well:
More, I don’t know even this is feasible/advisable but I thought you could unload your graphics card module and see if it would affect the power consumption. As said above, not having any hands-on experience on these type of hardware, I would let other “sharper pens” to give their opinion about this.
Thank you, @cactux. I will try those approaches soon. I forgot to calibrate powertop after installing version 2.16.
I have, however, confirmed that all of my current processes are running on the Intel iGPU after all. I did this by running nvidia-smi and by manually launching processes on specific GPUs and comparing those processes to others in the KDE System Monitor. The processes running on the iGPU say “GPU 2” in the System Monitor, just like all the other processes. Regardless, the output of nvidia-smi suggests that it is still using 3W.
ue Jun 2 14:56:51 2026
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 610.43.02 KMD Version: 610.43.02 CUDA UMD Version: 13.3 |
+-----------------------------------------+------------------------+----------------------+
| 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 RTX 4050 ... Off | 00000000:01:00.0 Off | N/A |
| N/A 46C P8 3W / 35W | 0MiB / 6141MiB | 0% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| No running processes found |
+-----------------------------------------------------------------------------------------+
@Bink@dalto Given my most recent findings, I was wondering if the switch to BTRFS, along with the transfer of a few hundred gigabytes of documents several days ago, is still the key variable here. I did see something about battery usage of BRFS with NVMe drives,
