Graphics cards and chips for 80's arcade games?

Keatah · September 5, 2013

We all know PCs have graphics chips. So what did the early arcade games use? Games like Tempest, Missile command, Liberator, Pac-Man, Space Invaders.. What'd they use?

Osgeld · September 5, 2013

many used off the shelf chips like the TMS found later in coleco and msx, others used custom solutions made out of parts of the time

Retro Rogue · September 6, 2013

Arcade games in the beginning were custom "state machines" - no microprocessor, no graphics chip, etc. they generated pixels purely by manually generating pixels at specific timings (in a manner similar to the 2600's TIA) and used various discrete circuits to create those timings. The game was engineered, not programmed.

Then ROM chips for graphic element storage were introduced, upping the possibilities. Color was introduced early on as well (usually done again through special timing hoops they had to jump through).

The microprocessors came in and the game itself no longer had to be engineered, since it was separated as software. Hower the tradeoff was that every new game was essentially a new custom computer. One created just for that game and the needs of it graphically and sound wise, with custom graphics and sound circuitry (via discrete logic circuitry).

In the late 70s/erly 80s you had groups of games sometimes using a single specific custom board, giving these games similarities in graphics and sound much the same way you can tell a specific vintage console or computer immediately by it's graphics and sound. An example would be Williams games such as Defender, Robotron, Joust, and Sinistar which all run on the same Williams 6809 Rev 1 hardware. This is also the period you had sound sound co-processor chips appear, such as the early AY series or Atari's own custome POKEY chip. Sound chips were actually in use far before and graphics chips.

I'm not sure where Osgeld got the idea that they used off the shelf chips for graphics at this time, they did not. Coin monitor resolutions were far higher than the television resolutions those parts were designed for.

Likewise, Tempest shouldn't really be included in any of this (same with Asteroids, Asteroids Deluxe, etc.) That's a vector game. A vector display is completely different, and the graphics are done via the game hardware directly controlling the CRT's beam like an etch-a-sketch. The game coder literally maps out the vector points to tell the beam where to go, and the mathematical translations for scaling, rotations, etc. vector hardware, be ause of it's polygonal nature, is where we find the first graphics co-processors in use. Such as the custom math-boxes created for Battlezone and Tempest to aid in complex vector calculations, using 2901 bit-slice processors.

2901s are also used as the graphics co-processor for the very first filled polygon 3d game, Atari's I Robot. Around this time you also had raster based games using standard CPUs to aid in graphics calculations. Galaga, for instance, uses 3 z80s. One for the main CPU, one for a sound processor, and one on the custom graphics board.

As 3d games and hardware became popular in the 90s, you still saw a lot of custom 3d and 2d processors being developed, until the mid-90s when GPU manufacturers started appearing and getting popular, such as ATI, SGICKS, 3DFX, etc., coinciding with their rising popularity in personal computers.

If you're really interested in learning a bit more about coin-op gaming hardware, feel free to look through system16.com. It's a great reference source. Also try reading through the original manuals for various coin-op games. They usually cover their hardware's theory of operation, and specific parts.

SoulBlazer · September 6, 2013

Wasn't Sega and Capcom still using custom boards for their games in the mid to late 90's and Nintendo using them for their games now? (Or should I say Namco, I'm not really sure how much Nintendo is actually involved in arcade games anymore).