EE Bench PC, what OS? Suggestions?

[SiliconWizard]

Picoscope runs "fine" on Linux. Although its software eats up much more CPU than it does on Windows.
Oh, and Picoscope 7 is atrocious IMO.

[radiolistener]

I don't recommend Ubuntu and Linux Mint. I've installed them several times and every time I got broken video driver and apt after some next apt upgrade with no way to restore or update the system. Also ubuntu and mint don't support OpenCL for my Nvidia GPU.

Regarding to KiCad, I'm installed it from backports repository, just add this line to /etc/apt/sources.list.d/bookworm-backports.list

Code: [Select]

deb http://deb.debian.org/debian bookworm-backports main contrib non-free

and then install it with

Code: [Select]

sudo apt update
sudo apt install -t bookworm-backports kicad

currently it has KiCad v8.0.6+1.

[nctnico]

Where it comes to NVidia videocards: just install NVidia's own Linux drivers and be done with it.

[radiolistener]

Where it comes to NVidia videocards: just install NVidia's own Linux drivers and be done with it.

This is exactly what I did. But it appears that apt upgrade may fail due to some changes which conflicts with nvidia code. As result, executing apt upgrade sometimes leads to a broken system with non working video driver. And for some unknown reason it appears that this happens too often for linux mint and ubuntu and can lead to unrecoverable errors in package system. Sometimes it can be fixed manually with edit of the source code and recompile it, but sometimes its too complicated.  So, apt upgrade may damage your OS and will needs to reinstall it.

[nctnico]

That likely is a problem with Ubuntu if you ask me. From what I've seen Debian is far better at creating a stable distribution compared to Ubuntu. At the cost of not having the latest & greatest (and buggiest) versions of software. Then again, I have been using Debian since the mid 90's so it could be I'm avoiding the pitfalls by muscle memory.

[Nominal Animal]

The compatibility issues with Nvidia proprietary drivers stem from DKMS, Dynamic Kernel Module Support.

Essentially, the proprietary drivers are binaries that are compatible with a range of kernels, with a thin shim compiled for each kernel using DKMS.
These can be compiled beforehand and distributed as a specific kernel-package-version dependent packages, but not all Linux distributions have the manpower or volunteers to do so, and Nvidia definitely cannot afford to provide those for all Linux distributions.  On most systems, that shim is compiled on the target machine itself, automatically per DKMS configuration.

So, the solution is to ensure you have kernel headers and working DKMS auto-configured for all future kernels, before you install the Nvidia drivers.  That way, as long as you have a Nvidia driver compatible with your kernel version –– not distribution patch version; just the kernel version suffices –– any kernel update will cause DKMS to recompile the shim for the Nvidia drivers.

This is not specific to Nvidia, either: VirtualBox additions (like the video driver allowing X/Wayland video resizing) use the exact same mechanism.

A secondary problem with Nvidia drivers is that it may depend on innards (but not kernel-userspace API/ABI) of the DRI layer, which means that a specific Nvidia proprietary driver is compatible with only a range of Linux kernel drivers.  It is the DKMS that recognizes the incompatibility, too –– incompatibilities that result in runtime errors are rare, although glitches may occur on new kernel until Nvidia releases compatible drivers.

Thus, it is all about DKMS and its configuration, and monitoring it when updating/upgrading kernels; not so much about distributions per se.
Some distributions may have maintainers that do this for you (and all other Nvidia driver users), but it should not be a requirement.

(In case you have labeled me a Linux zealot, back in the AMD/ATI proprietary driver era, I compiled my own kernels and used the proprietary drivers for better 3D support on 680G/780G integrated-on-motherboard chipsets myself.  It was much more work back then, too.  DKMS was specifically designed to ease that work by Dell!)

This is unfortunate, I do agree.  I know the history, and it is a bit convoluted on both (Nvidia and Linux) sides.  Yet, it is not a Linux distribution issue per se –– as in, "X is compatible with Nvidia drivers, Y is not".  Some distros may do the DKMS config for you, but in the rest, you need to do it yourself.  Fortunately, after it is set up correctly, it should keep working even across kernel updates, only requiring you to keep up with Nvidia driver releases afterwards.

[paulca]

I recall the issues with the Nvidia driver on any distro was the proprietary driver requiring stubs compiled against the running kernel headers.

The driver is in two parts.  The X11 part and the kernel part.

Issues usually result when the "package system" in use upgrades the kernel breaking the NVidia kernel module, preventing it from loading and if not spotted an fixed will result in a failed boot to X11 and a constantly respawning display, not very nice or easy to get out of.

Other issues happens when it upgrades the X11 part of the driver, but not the kernel module, etc. etc.

I don't really run linux on any boxes with NVidia cards anymore, so this hasn't bothered me in a while.

[tatel]

I don't really run linux on any boxes with NVidia cards anymore, so this hasn't bothered me in a while.

+1

[SiliconWizard]

Nvidia was the better option like 20 years ago on Linux, but as many know, things have changed.
AMD GPUs these days are the path of least resistance on Linux at this point. Although, be careful to use preferably Linux lts kernels, if you don't want to have to occasionally deal with newer amdgpu versions introducing various bugs, which tends to happen more often than with Nvidia IME. So, if you're after cutting-edge, that may be a different story.

Frankly, for a "bench" PC, that's probably not going to require much GPU power (unless you use it as a workstation?), I'd stick with Intel integrated graphics. Or maybe an entry-level AMD GPU.

[paulca]

I use VMs for dev work.  So no GPU.

The only application I choose to reboot the Windows box for and use the 3080 is KiCAD.  The reason is the schematic editor and PCB layout editor are both OpenGL accelerated.  Using them on the VM works fine, but it's laggy, especially when dragging things, you really feel you are waiting on it all the time.  With the GTX3080 it's butter smooth and responsive