zaval wrote:
I don't know why these vendors don't accept PC form factors, like mini-ITX for example.
I would think that the reason would be obvious: because they aren't intended as stand-alone computers, but for use in a larger embedded system. The small form factor is the entire point of these devices.
The real issue is that the SBCs we're talking about aren't meant to be desktop PCs. They are made for use as training tools in schools (which, as I keep pointing out, and is the real intended purpose of the RPi), embedded mobile systems in things like drones, IoT controllers, drop-in systems for things like Media Centers (Kodi, etc.), retrogaming consoles (e.g., Retropie, Lakka, etc.), and other uses where the small form factor is a necessary part of how they are being used.
I would agree that it would be nice to see some mini-ITX form factor ARM and MIPS boards (actually, at least one ARM board, the Asus Jetson, does exist, but it is insanely expensive and even the newest model is now long out of date). There are some PowerPC ATX motherboards, or were at one time, but PowerPC itself is sort of a dead tech today (for better or worse).
However, you wouldn't really want to use a SoC for something like that, and the majority of ARM and MIPS CPUs made today are SoCs, and the ones which aren't are generally designed for server farm and HPC uses (and thus would not be really suitable for a personal workstation - server CPUs generally run at a lower clock rate than PCs, because cycles per watts is the primary metric used in both server farms and HPC supercomputers, with performance being achieved through the distributed use of dozens or even hundreds of CPUs). There is no real reason why a desktop-grade CPU couldn't be designed for those ISAs (there certainly have been in the past - the ARM was designed for one, after all), but that market segment is currently locked up by mass-market Windows systems, and while there is an ARM version of Windows 10, it is specifically for certain mobile uses, and is only available on systems it is specifically tailored for (and which are locked down into using
only that version of Windows).
I should add that part of the issue is that there is no equivalent to 'standard PC' hardware which Windows could target; as with the various versions of Linux used on these SBCs (based on generic kernel ports), a version would need to be made for each different board, even if the CPU used in them were the same. Fans may be willing to work on such projects, and the manufacturers of the boards can do so given the existing OS kernels, but Microsoft is only going to do this if they benefit from it (either from profit-making sales, or from enhancing mind-share on their profitable markets).
(In fact, the real relationship is reversed: these consumer-grade SBCs really only exist because there is a generic FOSS OS kernel for those processors that they can be readily adapted to the specific SBCs. As wretched as Linux may be in some ways, no one would be making SBCs if they had to write their own OSes from the ground up. There would be maker-grade ones such as Arduino and Beagleboard, made with the expectation that the purchaser would be coding to the bare metal, but the larger market for SBCs with ready-use OSes wouldn't be feasible.)
As I've said before, unless you can convince the CPU manufacturers to build PC-class CPUs in volume, and convince several major mobo developers such as Asus, Gigabyte, and EVGA to develop and publish a standard hardware suite for a class of non-x86 motherboards (including the memory, video, and disk/SSD interfaces),
and get major software vendors such as Microsoft, Oracle, etc. to support this hardware, desktop systems based on ARM, MIPS, RISC-V, or anything else are unlikely to emerge. The alternative is for the SBC manufacturers to make their own hardware standard (and agree to use it), then scale up over time - which would remain meaningless unless they can can get software vendors behind them.
Personally? I think that a third option is more likely - that the rise of mobile systems, on the one hand, and computer-like functions in Smart TVs (possibly with pluggable upgrades, but maybe not), on the other, will marginalize the desktop and laptop markets to writers, students, and programmers; end the viability of game consoles (since the Smart TVs will be more than capable of running games directly from Steam et. al.), and push SBCs back to being a fringe interest.
Oh, and make new hobby OS development next to impossible. I rather hope this won't be the case, but...