If you can read both files and the the checksums are the same, that would imply that there is no storage corruption. I would still be inclined to try a different storage device.
Is there a possibilty of RAM corruption? Could a broken control chip result in things like these? I did use this thing a lot.
I noticed the board doesn’t clear its RAM fully on reboots. Might be related, if it’s not normal.
DRAM is like a very very large array of capacitors, unless power is fully removed for long enough they keep their contents. And the first thing the memory controller does is continuously refresh their contents to keep the data that is in them in them. It has no idea what is there. In theory the boot process could be made slower by deliberately zeroing out all RAM (except for the location of the program that is zeroing the memory). But nobody wants slower reboot times.
There is even an attack on PC’s where the RAM is sprayed with coolant to delay the discharging process, giving an evil person enough time to remove all the memory modules, transfer them to a custom built piece of hardware to dump the contents of RAM which typically includes unencrypted security keys for disk encryption.
So on a warm boot, it is normal and expected behavior due to how memory works at the lowest level.
1 Like
Thanks for the ansver, it is very informative.
On subject of storage devices, it eventually happened on every type. I used to have my system on nvme, now it’s on emmc+nvme, it happened there, and it happened on my sd cards, as well. Always had to do something with apt and reconfiguring kernel as I now recall.
See what I meant by “gaslighting error”? Here I am questioning my sanity and setup, while others on the web have the same problem 
I am not a coder of any sort, I just like full control over my working setups and am willing to go the extra mile – but I did do a fair share of reverse-engineering back in the day, and that’s the reason I call it a typographical error: it seems as if somewhere in the driver code, the name is spelt wrong and looking for the “p” variant is indeed the ansver to that. >_>
Namely: the driver is reported found once the ‘p’ copy is there. The original driver. NOT the ‘p’ copy version. Which means its internal search (for the non p version) is satisfied…
if you found the system was very slow，please use this command to try
dpkg-reconfigure linux-image-6.12.5-starfive
2 Likes
Thanks. I tried it, but it doesn’t feel as if it has a big impact.
Perhaps it’s just that I haven’t used the VF2 for quite some time and it is the slowest SBC that I currently have in use.
1 Like
Have you tried to change the screen resolution from 4k to 1920x1080, that usually give me the most speedup.
Thanks, mine was crawling, like 80 seconds to open a Terminal. I ended up hitting Control+Alt+F4 to open a text login terminal, typed your command and sudo reboot and everything is back to normal speed (In my case for Ubuntu 24.04.3 LTS).
1 Like
I use 1600*900, so there is no need to enlarge the font. presbyopia
1 Like
@LivingLinux : on GPU acceleration in firefox – note there are two versions available from the repo: firefox and firefox-esr. From what I gather, firefox-esr is supposed to have GPU support; it reports not supporting WebGL when visiting aquarium, so I’ve no idea and have not tested it further.
Note that the system supports two different GL implementations: OpenGL and GL-ES. Glxinfo will always report using softpipe. Glxinfo-es2-wayland or whatever the alternative is, should report using the PVR GPU, however.
This is interesting and confusing. Regular firefox will be using the GPU when visiting the 3D aquarium demo; it’s impossible to get to 500 fish with 5+ fps framerate on a slow 4-core PC under softpipe. You can check with “cat /sys/kernel/debug/pvr/status” command while the demo is working.
By the way, you’re doing the board a disservice if you’re testing acceleration with glxgears, try glxgears-es2-wayland instead.