The Nintendo 64 used RAMBUS memory, width 9 bits including parity; the latter would seem unnecessary for a console, but the machine actually uses it to store an extra bit of data.

This is an unusual design choice (relative to just omitting parity), and I'm curious about the reasoning. Is it that they particularly wanted exactly 18 bits per word, and this was the way to get it? Did it happen to be the case that the best deal they could get on memory modules included a parity chip for less than 1/8 extra money? Would the same idea also work on the more commonly used EDO RAM?

  • 6
    As an aside: I've used 9-bit wide memory for storing data on an FPGA before; it can be quite convenient at times. Particularly, with 18-bit words you can get a vector of 3x6-bit components, or 12-bit components in a 36-bit word. The particular job I was working on was a digital signage system that used RGB LEDs driven by shift registers, and 6 bits per channel was ideal for that purpose...
    – Jules
    Apr 16, 2018 at 7:58

1 Answer 1


RAMBUS was choosen for speed. They always came in 9 bit, which in turn was used for more precice storage by the graphics hardware by using 18 bit words for RGB/Coverage and Z/Delta-Z.

(And no, it's not 'repurposing' a parity bit, as the RAMs are simply organized as 9 bit words. There is no special purpose of either bit - thats something external logic may apply - or not)


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .