Bunch of libertarians argue about CPU stuff from 40 years ago

pealing off from
viewtopic.php?t=25720&start=15

Yeah and I think even learning very different processors like ARM and MIPS was quite trivial compared to my first assembly language.
The first time I think it took at least a year before I started to feel comfortable with it because there were so many small things I just didn't understand and had to learn little by little, slowly solving the puzzle. When I learned other architectures I didn't have to go through that process as I already had all the basics and could basically start coding right away (though with a lot of referencing of the MPU data sheets).

The trick with programming Assembly is not writing the commands, CPU design is basically CPU design the core components of a computer were worked out with gears. Basically every computer can be made with subtraction and compare with 0, everything on top of that is gravy. But the nominal set is every CPU can +,-,*2,÷2 compare and jump. The hard part is learning how to transform every task into these steps by learning how to break things down into small enough steps to be converted. Once you have that skill, the specific command names you type became a slight translation issue.

Well and something that helps too is nowadays RISC is generally the umbrella new stuff falls under, but the further back you go, the more diverse and esoteric instruction sets get. The VAX is a classic example of a bulky CISC implementation, AMD64 has so much cruft in it and somehow Intel and/or AMD cannot seem to convince the computing world to let go of any of it, god forbid someone might be able to write x86-ish code and *gasp* it won't work flawlessly on an 8086 and a Ryzen whatever. For the record it didn't start out that way, I've got a prelim 8086 manual from the late 70s and it's thinner than both the comparable MOS document on the 6502 and the earliest 68k manual I have. Compare with the last print run of AMD64 manuals from like the mid 2000s which is in 5 volumes, each thicker than both the 6502 and 68k manuals combined.

RISC is dead, has been for a while(apart from tiny microcontroller stuff). But RISC-V has come back as the true believe to deliver us the RISC promised land that either the other CPUs have left, aka ARM or died a quick and painless death MIPS, SPARC, SUPER-H, PA-RISC, Alpha, PRISM, i960 or had a long drawn out painful death PPC. To misquote Colin Chapman "All CPUs put on complexity".
Why does X86 win, why does Windows win, why has everything else fallen to dust.. because you can still run 8086 code on your modern computer ( mostly kinda not really, you need an emulator to run real mode etc ) the others didn't do it, and the others all died because of it. Support and compatibility is the key to it winning. Even the MS-DOS compatible machines died a fast and hard death. Seen as Osborne or a Apricot for sale of late? dead. IBM compatible was the only path forward. Growth and development needs a stable ground to stand on, you can't build tall on shifting sand.

Ken Shirriff pointed out that the Intel printed book documentation for last set of CPUs they did it for, has more pages than the 6502 has transistors.

As a career programmer who often doesn't have an option, I don't really have a lot of sympathy for shops with the opinion of "but *working* on our software is hard and expensive we don't wanna, money please". Should be tough cookies, but instead the entire computing landscape is held back on multiple fronts on the regular by those exact attitudes. Used to be you took risks to achieve rewards, now it's like risk is a dirty word in business but they still want to act like they're playing free enterprise. It's kinda insulting tbh. But surprise surprise the western world embracing all the worst parts of capitalism and trying to brutally murder whatever redeeming qualities might potentially exist down in there somewhere. And yes I do try and speak with my wallet, I don't buy enshittified crap, but most people have for whatever reason just rolled belly up and accepted that as the norm? It's so tiring being told this sort of thing over and over and over again, that effectively making changes like this has a cost. Yes, it does, I'm not ignorant, but business *has costs*.

Given IBM have only just recently lost the battle to have Trigraphs removed from the C standard, there is a lot of big systems still running very old code. They only stopped making punch card readers in 84, there were people using them still in the 90s. Just as there are still shops using Commodore 64s as their Till or to control something in their shop. There is a district of schools in America were all the heaters/AC is controlled by a single A2000. The Navy still use MS-DOS on their ships. If a system works nobody wants to change it, if it ain't broke don't fix. But then machines break and you can't get the parts. To solve this companies like Vortex86 exist that make for the industrial controller systems that still use 486s. Things still have VGA ports and PS/2 ports, RS232 still lives. You can buy modern PC motherboards with Parallel ports on them or get a breakout cable that connects to a header for them. The computer is just one part of the massive system and changing it means changing the whole massive system and the rules and checks and safety and legal compliance etc It is just not worth it, fix the 360 mainframe and make an emulator to fake being a terminal.

They are looking at bringing Three Mile island power plant back online, they are going to probably still use the 60s computer tech that it has build in to it, maybe make some adapters to control some buttons here and there but you don't pull down and rebuild the whole power plant and reactors because it should use USB and not Teletype.

Apple IIgs had a separate Apple II chip with a separate 6502 because I dunno

no it doesn't, it has the Mega-II chip to give all binary compatibility with the apple II systems,namely the odd floppy disk controllers, but it doesn't have a 6502 in it, all Apple II code runs on the 65816.

Statistics: Posted by Oziphantom — Thu Jan 16, 2025 12:15 am — Replies 3 — Views 117

---