Sidewinder said:
As a matter of clarification, The Atari is a 4 bit system, Commodore Vic-20, Commodore 64, NES and Sega Master system are 8 bit, And Sega Genesis and SNES are 16 bit.
The Atari VCS is a fairly basic 8 but system based on the MOS Technologies 6507 processor, which was a stopped-down version of the 6502 (as used in the Apple ][, asking others) packaged into a low-cost DIP-28. What I learned on was the 6502-based Atari 800, which, like all 8-bit machines, can access 65536 bytes (64k) of RAM before having to resort to technical wizardry to switch memory in and out of the system. The commodore 64 was based on the MOS 6510, which was essentially a modified 6502 with some additional I/O, and it overmapped the upper memory with the ROM to get 64k useable if you bank-switched the ROM out.
The Atari 400/800 (the latter had expandable RAM to 48k, mechanical keyboard, and dual cartridge slots) was based on a custom chipset designed by Jay Miner, who did the groundbreaking (for 1977) TIA chip that was the heart of the VCS (later called the 2600 to distinguish it from the ill-fated 5200). The 400/800 had a trio of custom chips; CTIA (and the later updated GTIA that enabled three more graphics modes), PIA, and POKEY. These offloaded some of the work from the CPU, as well as delivering a very robust set of system interrupts such as DLIs that allowed 18 clock cycles as the election beam in the TV monitor reached the end of the scan line and returned to scan another line; this was enough time to do a PHA to save the accumulator on the stack, alter one memory location with LDA/STA, and PLA to restore the accumulator so as not to disturb the program running before the interrupt triggered. One could change color registers, character set maps (including a full remap of all 128 characters that could be placed in RAM) and all sorts of cool stuff, provided one could use assembly. Sadly, a nasty little bug in Atari BASIC caused the tokenized program to have to search through the entire set of tokens each time an instruction was decoded, rather than prioritizing the search of the token table by instruction frequency and optimizing the search. This made the BASIC rather slow and hobbled what was in many ways a remarkable machine.
Jay Miner went on to design the Amiga chipset, which was another rather remarkable machine that got caught up in the infighting and politics of Commodore's implosion. By 1991, I was coding on PCs instead of anything Amiga or Atari, and traded my Amiga 2000 for a VW Rabbit that needed a transmission swap. Should have kept the Amiga - it was nicer than the Volkswagen, and much cleaner.
You may just have to trust me on this stuff - I lived and breathed this from the time I was 11 and got access to my school's TRS-80 Model 1 all the way until I graduated college, having paid a lot of my way by developing standalone applications on the now-ancient Access 2.0.
Sometimes I miss the down-and-dirty coding style of the old 8-bit home computers. No real OS, at least not in the proper sense, so you just poked directly at the hardware. But I can still do that with microcontrollers - I've done a lot of fun stuff with AVRs, both in assembly and C. PLCs pay better, though, and there aren't that many people around here who can work with them, so I've got pretty ironclad job security.
