Retro Consoles Used Fixed-Function Chips for Graphics Before Modern GPUs

Evolution of Console Graphics

For over two decades, console graphics relied on bespoke, fixed-function chips. This approach changed with the original Xbox in 2001, which incorporated a PC GPU. Early consoles like the NES exhibited quirks due to these fixed-function chips, such as sprite flickering when exceeding hardware limits.

Fixed-Function Chip Limitations

The NES's Ricoh 2C02 Picture Processing Unit (PPU) rendered graphics without executing code. It took tile artwork and lookup tables to generate a TV signal, one scanline at a time. This setup, where a dedicated picture generator was fed by the CPU, was common in consoles from the late 1970s to the early 2000s.

The Xbox's Introduction of Programmable GPUs

The first Microsoft Xbox, released in November 2001, departed from previous console designs. It featured an Intel Pentium III and a custom NVIDIA graphics chip based on the GeForce 3. The GeForce 3, announced in February 2001, was significant as it was the first GPU that enabled developers to program their own pixel and vertex shaders.

NES Graphics Processing Details

The NES's Ricoh 2C02 operated with approximately 10kB of memory, including 8kB for tile shapes from the cartridge and 2kB of RAM for background maps. Internal chip memories tracked up to 64 sprites. The output resolution was 256x240 pixels, constructed from 8x8 tiles. The 2C02 determined tile and sprite placement for each scanline in sync with the CRT's electron beam, preventing the CPU from altering the picture during drawing.