Understanding the 6502

The microprocessor that powered the Apple II, the Commodore 64, the Atari 2600, the BBC Micro, and the Nintendo Entertainment System was not the fastest chip of its day, nor the most capable. It won because it was cheap, predictable, and small enough to understand completely — three properties that, taken together, put computing into the hands of a generation.

A radical economy

When MOS Technology introduced the 6502 in 1975, it sold for around $25 at a time when comparable parts from Intel and Motorola ran four to six times that price. The story behind that number is itself a piece of computing history: a team led by Chuck Peddle, many of them refugees from Motorola's 6800 project, set out to build a processor that ordinary people and small companies could actually afford.

They got there by ruthless economy. The 6502 has fewer transistors than its rivals, a deliberately small register set, and an instruction decoder built from a compact programmable logic array rather than sprawling random logic. Where other designers added registers and addressing modes, the 6502 team subtracted — and then leaned on one clever idea to make up the difference.

The zero page: poor man's registers

That idea was the zero page. The first 256 bytes of memory ($0000–$00FF) get their own shorter, faster instructions. A normal absolute access needs a two-byte address and an extra cycle; a zero-page access needs only one byte and one fewer cycle. Programmers learned to treat the zero page as a bank of 256 fast pseudo-registers, and good 6502 code is, in large part, the art of deciding what lives there.

The visible programmer's model is famously spare: an 8-bit accumulator A, two index registers X and Y, a stack pointer, a program counter, and a status register of flags. That is nearly the whole machine. You can hold it in your head — which is precisely why so many people learned to program on it.

        LDX #$00        ; start at index 0
LOOP    LDA MESSAGE,X   ; load the next character
        BEQ DONE        ; zero byte means end of string
        JSR $FFD2       ; CHROUT — print the character
        INX             ; advance the index
        BNE LOOP        ; (always taken here) keep going
DONE    RTS             ; return to the caller

Ten lines, and a competent reader can account for every cycle. That transparency is the heart of the 6502's appeal.

Racing the beam

Because the 6502's timing was so regular — most instructions took a small, documented number of cycles — programmers could choreograph the processor against the rest of the machine with extraordinary precision. The most famous expression of this is "racing the beam": timing code to the electron beam sweeping across a CRT, changing colors, sprites, or registers in the narrow window between scan lines.

The 6502 didn't give you much. What it gave you, it gave you on a schedule you could count — and a programmer who can count cycles can perform magic.

On the Atari 2600, with almost no frame buffer to speak of, the entire picture had to be generated on the fly, line by line, by code synchronized to the display. On the Commodore 64 and Apple II, the same discipline unlocked effects the hardware was never specified to produce. Demos that "shouldn't" run still run today, and emulating them faithfully means emulating the 6502 down to the cycle.

The machines it made

The 6502 and its close relatives became the common engine of the home-computer boom:

• Apple II (1977) — the machine that made the personal computer a serious proposition, and the 6502's first great showcase.
• Commodore PET and VIC-20, then the Commodore 64 (1982) — the best-selling computer model of all time, a 6502 derivative (the 6510) at its core.
• Atari 8-bit and the Atari 2600 — the 6507, a cost-reduced 6502, anchoring the first console generation.
• BBC Micro — the machine that taught a nation to program, in British classrooms throughout the 1980s.
• Nintendo Entertainment System — a Ricoh chip built around a 6502 core, carrying the architecture into tens of millions of living rooms.

One small, cheap design, licensed and cloned and adapted, sat under an astonishing share of the era's software. To preserve that software is, in large part, to preserve the behavior of this one chip.

Why it still matters

The 6502 endures as a teaching instrument because it is complete and finite. A motivated person can read the whole instruction set in an afternoon, write a working emulator in a weekend, and understand exactly why a forty-year-old program behaves the way it does. Few modern processors offer that, and none with this much history attached.

For the Foundation, the 6502 is more than nostalgia. Faithful emulation — cycle-accurate, undocumented opcodes and all — is what lets a recovered disk image actually run the way it did in 1984. Get the chip right and the rest of the era comes back to life. Get it wrong and you have a museum of things that almost work.

That is why we treat the silicon with the same care as the software it ran: the 6502 is not a footnote to the story of personal computing. For a remarkable stretch of years, it was the story.