In testing the electrical power consumption of my machines using a watt meter, I noticed that there is a consistent increase of 0.7 watts at idle. This may seem small, but when the 22.04 draw is 0.9 watts to begin with and goes up to 1.6 watts simply by running 24.04, that to me is a significant increase.
I have more testing to do but was wondering if anyone has any idea why this is happening. What are we gaining with this increased power draw? Are there settings that can be changed?
I'm thinking about going back to 22.04 for my Nextcloud server as that is on 24/7.
Have you tried using tools like powertop to determine what's eating the power?
Also, might be worth running the 22.04 kernel on your 24.04 system (or if you downgrade, run the 24.04 kernel on 22.04) to see if it's as simple as the kernel which is consuming more power, or might be something in userspace.
These were fresh installs as well as upgraded installs on laptops tested with the screen blanked. It was the upgrade to 24.04 on my server that I noticed it.
Others should be able to provide confirmation of the phenomena. Unless physics is a little different where I'm at (always a possibility).
I hate to be little johnny-one-note here (because I have a whole thread elsewhere griping about this issue) but I think I might have an insight that could be relevant. Modern hard drives have sophisticated power management, and even SSD will attempt to 'spin down' into a lower power state when not actually in use. Due to this bug, this can't happen because the system is CONSTANTLY WRITING TO THE HARD DRIVE.
In the thread I linked above I have a simple test to check for this .xsession-errors log spamming issue:
Open the caja file manager on your home folder.
Select "Show Hidden Files" from caja's "View" dropdown menu.
Right-click on the .xsession-errors file and select the bottom "Properties"
selection from the resulting caja right-click menu.
Note how quickly the file "Size" parameter is counting up in bytes/sec .
More than likely, if this is happening, all this HD activity is keeping your hard drive so busy that it can never enter it's low power state, and this could easily account for the small but significant extra power draw . . .
Thanks! I can confirm that the .xsession-error file is growing on one of the systems that I'm testing. You may be right that that is the source of the increased power usage.
Yesterday I created 3 separate disks, each a SATA SSD, for 22.04, 24.04 and Rocky Linux MATE. I then tested them on a desk top system and 24.04 drew less power than 22.04 suggesting that it's NVME specific. BTW Rocky drew the least power of the three.
All the laptops that I've tested so far are ASUS in case that is a factor.
Is there a bug report for this behavior? I'd like to report that it's also affecting me.
Both the Laptops I am seeing the problem on are also high performance laptop NVME SSDs. So you may well be onto something, because a few folks in the thread I linked above were unable to duplicate the issue. Perhaps they also have older SATA type hard drives (although why that would make a difference, I can't even hazard a guess)
I noticed an increase in power usage when I installed 24.04, too. I did have a 22.04 partition for a while; and I tried to check (with powertop) that settings were the same across both, but I never got to the bottom of it. The battery is quite old on my laptop so I don't rely on it much - hence I moved on to other things.
I would like to know what's causing the increase. I'd start with different kernels (if I could still be bothered).
I don't think log spam is the culprit because I noticed the increase in power usage regardless of whether or not I was using Caja.
I fresh installed 24.04 on my MSI gaming laptop. I get about 8 hours battery when using on board graphics, when graphic intensive things arise, it uses the Nvidia card and i get 5 or 5.5 hours. It seems fine to me. To be truthful, I am not a gamer, just wanted a really good laptop. My old laptop only had built in video and ran about 6 hours on a charge, with 22.04
update: I was mistaken. It lasts just 5 hours on onboard 3 hours using video card. Sorry about that
My ASUS Zephyrus G15 (2021 model) lasts about 2.5 hours on battery, and way less when gaming, but the battery pack had a fault (out of warranty) so its capacity is about half what it should be. Even then it's not the best battery life.
Power draw is about 11-14W when doing ordinary tasks (web browsing, code, etc).
I did see a noticeable hit on battery when I was testing and the really massive spamming was going on, but Stephen has raised a very good point. because as I noted in the other thread, further testing has shown that only certain things trigger the mass spamming. For example, one smoking gun thing that definitely causes the mass spamming to start up is to enable the display of hidden files in caja. So, ironically, trying to examine the problem CAUSES IT.
There are also some other less obvious things that an unsuspecting caja user can stumble across that will also cause the mass-spamming to start up. For example, even with the display of those troublesome hidden-files disabled in caja, something as simple as opening your home folder in caja and single click highlighting ANY folder will ALSO cause the .xsession-errors log mass-spamming to start up.
But the important thing I wanted to note here is that further testing has shown that ALL the caja log spamming issues relate to caja trying to do attribute checks - so as Stephen has correctly pointed out, it's fairly easy to eliminate this as a culprit while power testing by simply closing all your caja windows.
There are still onesie-twosie errors being reported, because caja is managing the desktop, even with all file manager windows closed, so something as simple as just clicking an icon on the Mate desktop will cause a GLib-GIO-CRITICAL error to be reported in the .xsession-errors log -- but these single event error reports are not significant enough to cause a hit on power.
I too found that the excess logging only occurred when I had Caja open. When I would SSH into the machine, it didn't show this behavior.
22.04 has 2 kernels 5.15 and 6.8 depending on whether HWE was installed. In both cases the power draw was the same. 24.04 has 6.8 and exhibits the increased draw. I don't suspect the kernel.
That would leave MATE, Ubuntu and or Debian as the culprit. Lots of testing ahead.
