You all need mandatory supervised access to the Internet from now on.  - for me it's like and end signal to stop reading

It is important to note that consideration should be given when using exFAT or NTFS file systems on Linux systems. These file systems (M$) are not natively supported by Linux, and their usage carries risks due to potential compatibility issues.

In other words, micro$ FS should remain on your desk next to the window — like a broken vase. Though I find exFAT extremely useful for file exchange.

Hardware: Yep, right now I'm fighting with an AMD Ryzen AI 9 HX 370. All firmware is up to date. Turbo boost is disabled, the governor is set to conservative while on battery, and energy_performance_preference is set to power. The kernel is new and allows AMD control.

However, the system constantly ramps up from 600 MHz to 2 GHz even when in idle state (0-2% load on some but not all cores). At least turbo boost is disabled; otherwise, while running tasks on battery, it would reach up to 5 GHz.

This machine is compact and lightweight, so I can take it with me. This is the reason to have it.  On Intel hardware, I don't have such issues except for sometimes very long wake-up times (~1 minute) from sleep mode. I suspect this is because ME (Management Engine) is disabled. Starting from the 11th generation, disabling ME can make machines unbootable at times. It might take 1 or 3 minutes to power on; if it doesn't want to power on, I have to switch it off and then back on again. So, I prefer not to shut down too often.

Unfortunately, the corporations have won. Now these hardware backdoors like Intel ME and AMD PSP (I don't know the names of hardware backdoors for Mac and Huawei) are becoming mandatory. Unfortunately, there weren't enough people who cared or small businesses that cared about this issue. Pity.

Now, if you have old hardware like Intel 10th generation or earlier, keep it carefully as all new hardware comes with permanent backdoors.

I'm not sure whether eliminating software cancers like systemd will win the war when there are hardware backdoors and proprietary BIOSes. 

I have another machine running devuan. No problems there. After the second attempt on the first machine, I mounted the filesystem using 'mount -o remount,rw exfat' and rebooted it. It worked fine. I've realized for a while now that computer science isn't an exact science. :lol

  - noted for the future, thanks.

Solved itself.

fsmithred, thank you. I'll check what can be done. You are right; it prefers debian. I've updated the first install (grub, linux-image etc) and it did not destroy the second one named devuan. However, it changed the EFI boot order, making debian the first one and refind the last one. It also completely eliminated "the second partition on the SD card" (which is Ventoy with .iso files). This strange behavior might be linked to the motherboard.

This is why I want to keep two similar systems for a while to test different configurations and approaches. There are too many things to resolve: find the right values for nbfc, make ryzenadjdJ understand ryzen_smu (it doesn't currently), etc.

I'll mark this topic as solved.

What I'm trying to say is that many applications are impossible to install because they depend on something else. Almost all these packages are present in Ceres or Daedalus. In other words, Excalibur has the fewest number of packages. What's unclear to me is why a package can exist both in Daedalus and Ceres but not be available in Excalibur?

rbit, if I remember correctly, I've downloaded the .deb file as the driver, unpacked it, and placed the firmware folder on a USB flash drive.

I see that there is a new . iso to download.

What about this Debian bug https://github.com/mattermost/desktop/issues/2853?
Will Devuan follow Debian's package policy?

Some packages that are present in Daedalus (Bookworm) are missing in Excalibur (Trixie). Some of them, like gconf2 and gconf-service, are only available in Daedalus/Bookworm. Others are in Ceres/Sid. This produces a bit of a 'Frankendevuan' situation."

What is wrong with Wayland? I use Wayland because only Wayland is able to scale all apps correctly even those that cannot be scaled with qt or gtk scale.

fsmithred, unfortunately I could not see your post before.

I've made install with devuan_excalibur_6.0-20250605_amd64_netinstall.iso to AMD Ryzen AI 9 HX 370, AMD Radeon 890M. I've flashed it to usb flash drive (no ventoy, rufus etc).

Expert install.
First error: The missing firmware files are : iwlwifi-ty********

Second error: No network interfaces detected

Then "The system can be set up to symlink /bin and /sbin to /usr/bin and /usr/sbin, respectively. ....If unsure, the safe choice is "No".
However, the default choice was Yes.
So,Yes.

Third error: Bad archive  mirror #of cause, there is no internet
Then "standard_system_utilities" - selected
"Update NVRAM variables to automatically boot into Devuan?" - Yes

Forth error: wpa_passphrase not found  #got it from https://packages.debian.org/trixie/wpasupplicant  unfortunately there is no such site for Devuan. Probably, wpasupplicant is stored somewhere in devuan_excalibur_6.0-20250605_amd64_netinstall.iso...
Then mounted Devuan flash drive to /media/cdroom to satisfy dependencies.
The same story with dhclient
Then tasksel, laptop....
Finally, apt install DE_of_choice (KDE) and reboot

I was able to log in successfully. To my surprise AMD graphics drivers and vulkan were installed.

BUT... shit happens. Fifth error: /EFI/debian is overridden. It was "debian" because devuan1 has been installed through migration. Why it's called "debian" again ?

Conclusion. Too complicated. May be it'll be simpler to install Devuan through Vmware on physical partitions and then boot normally?

Impossible. The installer does not detect network interfaces. I've only wireless. Guess how does it know which driver to request