I would be totally on your side if it was Intel, AMD, MSI, Asus or meanwhile RaspberryPi Foundation. They have built up a large development team, established codebase and all that. Pi 5 e.g. is sold as “The everything computer. Optimised.”. Not as a development board. These two things are not comparable! The VF2 is for people to experiment with RISC-V, the corresponding codebase and software infrastructure, not primarily as a SBC for drop into home automation or other daily use scenarios. If all goes well, that will be the case in 1-2 years. And to reach that goal, the board is released in its current state.
To perhaps better understand my point, a short clarification on why I personally bought a VF2 and what my goal is:
I’m sick of Intel/AMD and meanwhile partly also ARM rolling out beta releases of their hardware and board manufacturers doing the same on the mainboard and/or gpu side of things. Im tired of getting to know there is something like an ME (intel) or PSP (AMD) doing some magical stuff which I cannot see or influence.
I want in lets say late 2024 or early 2025, when my next deskop PC will be a topic, buy a RISC-V based computer. Something like the Milk-V Pioneer. Just a bit cheaper and with faster CPU cores. And I want to run Archlinux on that one. So all I’m doing right now for the VF2 helps that goal. That is my motivation.
Besides that, after playing around with the VF2 for some weeks now, I’ll buy a second one with a large SSD as my new home server, replacing my old Intel Atom based box.
That are my thoughts regarding that topic and for me, I feel I spent the money well on the VF2.
I also rate VF2 higher than that. Even its existence is a real asset.
My VF2 was the first usable home server machine and I use as such for 1 year now. Is is just not used to view movies and I don’t understand why we are loosing so much time on that (it is probably the job of the GPU vendor).
This release gives NVME booting and for me, this is much more important than Mesa/Firefox because it impacts all systems. And Big thank you to the team for this OS-independent feature.
I still expect some OS-independent features like boot order, not using my physical hands to switch the boot device, etc
Most Debian architectures (ppc64, ) don’t provide GPU acceleration enough to make them usable as a home pc.
Does anyone know the exact packages that are custom that apt upgrade will break? Couldnt we use “apt-mark hold” to prevent those packages from being modified and update everything else?
I guess I can’t talk about the GPU or the other three letter competitor and how they are working on drivers. Anyhow, someone seems to be working on something. Hopefully the community will receive the data it needs soon.
Is there a specific reason to not have all controllers enabled in the kernel’s cgroup support? not having cpusets as an enabled controller makes using nomad hard.
the log showed your partition may have problems. please check your partition.
l tried to flashing the latest spl、uboot and nvme image on 1.2A 4G board，all of them works well.
Can anyone run the following on this release:
sudo apt-get install mesa-utils
glxinfo | grep OpenGL
And reply if the OpenGL renderer is still softpipe?
My initial thoughts is that its kernel related with the driver handling although I need to do much more testing to isolate the individual calls causing it to create the artificial load. I have been able to recreate the condition and the fix on both the VF2 and Pine64 JH7110 SBCs however.
With an empty micro SD card formatted exfat finally cpu load is zero. I don’t know the reason but I want to report this issue to the riscv team developer @Michael.Zhu
sorry @Tenkawa , if I forgot to mention you in the post, I wanted to try what you wrote in the post and I confirm that now the load is zero as it should be.
I wanted to write here on the official post of release 23010 to submit the problem of the untrue artificial CPU load to the RISCV team.
I hope that this issue will be resolved in future releases.
