The fact that x86. Hasn’t changed its foundation much, isn’t that just a combination of hardware making up for original design shortcomings, while economy keeping better solutions at bay? (Not a chip guy, I’m likely wrong.)
The Convoluted Way Intel’s 386 Implemented Its Registers
Submitted 2 months ago by cm0002@lemmy.world to retrocomputing@lemmy.sdf.org
https://hackaday.com/2025/05/05/the-convoluted-way-intels-386-implemented-its-registers/
Comments
hbm@feddit.dk 2 months ago
hbm@feddit.dk 2 months ago
Never did SPARC assembly, but ISTR their registers were basically a ring of groups of registers allowing fast context switches as long as the call depth stayed shallow (fsvo shallow). When the ring was exhausted, you had to stash away in memory.
ThorrJo@lemmy.sdf.org 2 months ago
Ken Shirriff regularly posts cool hardware deep-dives and can be followed on fedi at @kenshirriff@oldbytes.space
PhilipTheBucket@ponder.cat 2 months ago
I knew it, man.
I learned MIPS assembly first, and it was like the registers were all a baseball team, with everyone trained and doing their part to try to get the job done.
When I learned about x86 assembly it was like the registers were a tiny band of wacky misfits who were going around in a van trying to solve mysteries.
paulh@lemmy.sdf.org 2 months ago
Fantastic description!
Z-80 was much the same: djnz uses C as the loop counter (like the 8086 LOOP used CX iirc), H and L combine to make a 16-bit pointer…