No standard platform
The single most important thing to understand about golden age arcade hardware is that there was no standard platform. Each manufacturer — Atari, Namco, Williams, Midway, Konami, Capcom, Sega, Irem — designed their own circuit boards, often for individual games. There were no shared development tools, no common processor architectures, no equivalent of what the NES or the PC would eventually provide: a stable target that developers could learn once and apply repeatedly.
This was both a constraint and an accelerant. It was a constraint because each game required engineering work from scratch — you couldn't take code from one Atari arcade game and run it on the board for the next Atari arcade game without significant rework. It was an accelerant because it meant hardware designers were continuously inventing new solutions to new problems, and the pace of advancement was determined by engineering creativity rather than by any platform holder's release schedule.
The CPU at the centre of most golden age arcade boards was the Zilog Z80 or the MOS 6502 — cheap, well-understood 8-bit processors that ran at 2-4 MHz. These were the same chips inside home computers of the period. The difference between a home computer game and an arcade game wasn't the CPU. It was everything around it.
The display: raster versus vector
Two distinct display technologies coexisted in the golden age arcade, and they produced entirely different visual aesthetics. Raster displays — used by Pac-Man, Donkey Kong, Space Invaders, Galaga, and the vast majority of golden age games — worked by scanning an electron beam across the screen in horizontal lines, top to bottom, 60 times per second. Each position on the grid could be set to a specific colour by the hardware. The sprites — the moving objects on screen — were defined as rectangular arrays of coloured pixels stored in ROM chips and drawn at specific positions on each scan line by dedicated sprite hardware.
Vector displays worked differently. Rather than scanning the entire screen in a fixed pattern, the electron beam was directed precisely along mathematical lines — vectors — at any angle. This produced perfectly sharp lines at any orientation, which raster screens of the period could not replicate cleanly. Asteroids, Tempest, Battlezone, and Star Wars all used vector displays. The geometric, wireframe aesthetic of these games was not a stylistic choice — it was the direct visual consequence of the display technology. Curves and filled areas were difficult or impossible; straight lines at any angle were the natural output.
The tradeoff was colour and complexity. Early vector monitors were monochrome. Colour overlays — physical transparent films placed over the monitor — provided the appearance of colour but not real per-pixel colour control. Raster screens could display dozens or hundreds of distinct colours simultaneously. Vector hardware excelled at crispness and geometric precision; raster hardware excelled at colour, detail, and the kind of visually complex sprite animation that Pac-Man and Donkey Kong required.
Custom chips: the source of the magic
The visual difference between an arcade game and a home computer game of the same era was not primarily software. It was hardware — specifically, the custom chips that arcade manufacturers designed to handle display tasks that general-purpose processors couldn't do fast enough.
Namco's hardware for Pac-Man included a sprite rendering chip capable of drawing eight 16×16 pixel sprites per scanline, each independently positioned and coloured, at 60 frames per second. A general-purpose Z80 running software rendering could not do this — the processor would have spent all its time drawing sprites and had none left for game logic. By offloading the sprite drawing to dedicated silicon, the Z80 was free to run AI, handle input, manage game state, and produce sound simultaneously.
Williams's hardware for Defender included a custom video chip designed by engineers at the company specifically for that game. The chip handled the horizontally scrolling playfield — a 360-degree wrap-around world — and the sprite positioning within it. The radar display, which showed the full extent of the world in miniature, required its own display logic. None of this was off-the-shelf. Williams designed it, had it manufactured, and used it in one game. That was the economic model: design your own hardware, use it for a year or two until the game has run its commercial life, then design new hardware for the next game.
Sound
Sound in golden age arcade games was produced by two distinct approaches. Analog sound generation — used in many early games through approximately 1980 — produced tones and noise by directly manipulating oscillator circuits. The characteristic sounds of Space Invaders (descending pulse tones) and Asteroids (thrust rumble, explosion noise) were analog circuits. They had no digital sound stored anywhere; the sounds were generated electrically in real time.
Digital sound, which became common from around 1980, stored waveforms in ROM chips and played them back through digital-to-analog converters. This allowed voices — the taunts in Berzerk, the speech in Gorf, the sample voice of Pac-Man — and more complex musical structures. Namco's Pac-Man hardware used three voice channels generated by the Namco custom chip; Williams's games from Defender onward used dedicated audio processors running their own code independently of the main CPU.
The FM synthesis chips that appeared in arcade hardware from around 1983 — particularly Yamaha's OPL and OPN series — produced the characteristic warm, bell-like tonal quality of games from that period. Gradius, 1942, Ghosts 'n Goblins, and dozens of other Capcom and Konami titles of the mid-1980s used Yamaha FM chips. The sound is immediately recognisable to anyone who grew up with those games: a particular brightness and warmth that neither the earlier analog circuits nor the sampled audio of the 1990s replicated.
The Capcom CPS-1 and the end of the era
By 1987–1988, the era of fully custom one-game hardware was giving way to programmable platform boards that multiple games could share. Capcom's CPS-1 board (1988) was one of the first commercially successful examples: a hardware platform designed to run multiple games through software differences rather than hardware redesigns. Street Fighter II, Final Fight, Ghouls 'n Ghosts, and dozens of other Capcom titles ran on CPS-1 hardware with different ROM chips containing different games.
The CPS-1 represented a maturation of the industry — or, depending on your perspective, a loss. The era when every game demanded entirely new hardware solutions, when the constraint of the blank board was also the freedom to invent anything at all, had produced some of the most creative engineering in computing history. The move to shared platforms traded that creativity for efficiency and reliability. Games still advanced rapidly, but the advances were in software and design rather than in the fundamental electronics of the display. The golden age of arcade hardware was, by about 1990, over.