Atari ST
Sixteen-bit music machine
The Atari ST married a 68000 CPU with built-in MIDI ports, making it a darling of musicians, demosceners, and bedroom developers.
Released in 1985, the Atari ST aimed squarely at the new 16-bit market. Its Motorola 68000 CPU, GEM desktop, and dual MIDI ports made it a favourite for music production and creative coding across Europe.
Fast facts
- CPU: Motorola 68000 at 8 MHz.
- Graphics: 320×200 (16 colours), 640×200 (4 colours), and 640×400 monochrome.
- Sound: Yamaha YM2149 PSG; later STE models added DMA sound.
- Storage: 3.5″ floppy drive standard, with hard-drive options via ACSI.
The machine that delegates nothing
The Amiga arrived the same year with the same processor and the opposite philosophy. Its entry describes what it was reacting against: 16-bit contemporaries that “paired a fast CPU with a dumb framebuffer”. The ST is that contemporary, and it is worth reading as a deliberate position rather than a shortfall.
There is no copper and no blitter. The screen is 32K of ordinary RAM — Atari ST Internals records that “immediately after turning the ST on, an area of 32K bytes is initialized at the upper memory border as the video RAM”, with a pair of registers to move it elsewhere. The Shifter reads that memory and produces pixels. Nothing else touches it. Every sprite, every scroll, every filled polygon is the 68000 moving words.
To pay for that the CPU is clocked faster than the Amiga’s 7.16 MHz, and the figure was chosen rather than maximised: the 68000 “can operate at different speeds”, and “the ST’s 8 mHz was chosen as a ‘middle of the road’ frequency to avoid losing data at higher frequencies”. A blitter did arrive, but not until the Mega ST in 1987 — two years into the machine’s life, and never on the model most people owned.
The result is a machine that is simpler to reason about and harder to make sing. Amiga effects are scheduled onto co-processors; ST effects are budgeted out of one instruction stream, which is why the ST demoscene’s reputation rests on things like fullscreen overscan and cycle-exact raster splits — feats of timing by the CPU rather than of delegation to hardware built for it.
Why MIDI was the whole argument
The ST’s lasting claim is not graphical, and the hardware documentation shows
how deeply the decision went. MIDI is not a port bolted to the side: Atari ST
Internals lists “Keyboard and MIDI ACIAs” together at $FFFC00, two 6850
Asynchronous Communications Interface Adapters, one serving the keyboard and one
serving musical instruments. The same part, the same treatment, adjacent in the
I/O map. On this machine a synthesiser is a peripheral of exactly the standing
of the keyboard you type on.
Atari sold it as the plain differentiator — “MIDI expansion ports are fitted as standard” — and the claim was true at a time when adding MIDI to a PC or Macintosh meant buying an interface card. The consequence was that a musician’s entire rig, sequencer and all, cost what the computer cost.
The architecture and the MIDI decision are the same story from two directions. A machine with no co-processors has a 68000 that is otherwise idle, and sequencing is a task that wants exactly that: not throughput, but an uninterrupted processor able to deliver events on time. The design that made the ST unglamorous for games made it dependable for music.
Why it mattered
- MIDI powerhouse: built-in ports created an affordable sequencing workstation.
- Demoscene hub: 16-bit demoscene groups pioneered scrolling text, vector graphics, and sample mixing on the ST before migrating techniques to Amiga/PC.
- Publishing: hosts of 8-bit alumni shipped ST/Amiga versions of their hits, bridging our BASIC and Assembly course eras.
Modern legacy
- Emulation: Hatari and Steem SSE emulate ST/STE/TOS variants.
- Hardware mods: modern video adapters, storage replacements, and FPGA clones keep the machines gig-ready.
- Community: active demo parties (Silly Venture, STNICCC) still celebrate new releases.