News Linux developers want to remove i486 and i586 Pentium CPU support to unburden kernel developers

Embedded systems.
Only issue my tiny PoV can think of, would be the ancient infrastructure still in place (esp. outside of US).

AFAIK, there are still a lot of industrial control systems using various forms of i486 and Pentium.
That said, they might already be 'frozen in time' on the software-side.

I'll be truly concerned when talks of dropping i686 support, start becoming more-common.
 
I still have a few of those in one of my "silicon jewelry boxes". But that's only because they were small enough for quite a few to fit into such a small box and to compensate that I game my 80286 and 80386 chips away with the computers they ran in.

Today I wish I still had one of each variant I ran throughout my four decades of PC ownership and IT career.

After going 32-bit with a few 80386, my first i486-33 ran Unix System V Release 3 from Interactive Systems and got replaced by an Intel DX2 and then an AMD DX4 variant, which had me stick with AMD for many iterations, basically until Intel got its game straight with the Q6600.

Somehow all those AMD 586 to K6-III+ 500 chips still blur into "80486" for me, I still have some of those CPUs, nothing to run them on.

For the OS I went from Unix to BSD and Solaris and don't really remember why or when I finally converted to Linux, most likely it was because of Xen: secretly I've always adored the VM/370 abstraction, while I hated the IBM mainframe almost as much as Microsoft, commerce over engineering was the underlying issue.

I drew the line at 64-bit and VM support a long time ago, the oldest hardware in storage still theoretically capable of running would an ancient Merom notebook and a Core2 board currently holding a modified Harpertown Xeon running cool at 3.4 GHz. But my original Kentsfield Q6600 is still in that CPU box and should run Windows 11 up to 23H2.

I've relived some of those "golden old times" via PCem a few years ago and that experience reminded me of all those endless hours spent on making those ancient PCs work with their different busses (XT/AT/MC/EISA/VESA-LB/PCI/PCI-X/PCIe) and how you had to configure all those add-on cards and BIOS options... some things actually did get better with the years!

I don't know if some aircraft carrier is kept alive only by a 80486 based system on Linux, I seem to remember that Space Shuttles needed 8086 CPUs at one point. But for software archaeology, emulation seems good enough for me, I don't spend much time with my Apple ][ emulator these days, which allowed me to let go of that hardware.

I also got Hercules to boot a z/Arch image of SuSE once, but finally being able to run an IBM mainframe on my PC didn't give me the kick I had come to expect ever since I nabbed one of those IBM metal stickers from one of them and put it on my first PC-AT clone in 1986.
 
Ahem, my dx4 ran 133 MHz so the 20-100mhz span is off
(And i newer heard of a 486 slower than 25, but maybe embedded)
... And it clocked at 166, and ran circles around the first pentium
 
Ahem, my dx4 ran 133 MHz so the 20-100mhz span is off
(And i newer heard of a 486 slower than 25, but maybe embedded)
... And it clocked at 166, and ran circles around the first pentium
I've got a 486sx 16/20mhz machine in my collection here. It always was used at 20mhz, unless you had some "old" software (games) that ran too fast, then you'd hit the Turbo button and it dropped to 16.

I *think* I have a 20/25mhz box around here somewheres too...
 
Ahem, my dx4 ran 133 MHz so the 20-100mhz span is off
(And i newer heard of a 486 slower than 25, but maybe embedded)
... And it clocked at 166, and ran circles around the first pentium
The [Intel] DX4 was actually a x3 clock multiplier from the external 33 MHz, which means 100MHz.
Only AMD had DX4 which were actually x4 clock multiples of the 33 MHz input clock and thus ran at 133 MHz.

Not so sure a 166 MHz 486 would have always beat even an early Pentium: a 64-bit data bus, a much better floating point implementation and most importantly, a non-native x86 ISA implementation and RISCy translation layer went a long architectural way towards beating faster clocks.

I went with a 83 MHz Pentium Overdrive for a short while, which offered the worst of both worlds, but a smoother upward transition if you had already bought an Overdrive ready mainboard.

I remember that first overdrive mostly for the fact that it included a tiny little fan, a moving part, that could obviously fail! Up to that point all CPUs were passive cooled.

But Intel offered life-time warranty replacement for those fans, which I'd like to claim today.... but that CPU is long gone.
 
I would be surprised if there ARE i486 embedded systems being updated right now in 2025
Not every country is as fast-moving (technologically) as the US, EU, etc.
Outside the 'western world', I'd imagine a lot of industrial control equipment, etc. is still utilizing 80s and 90s technology.