“K” versus “k”

“K” versus “k”

A perennial source of ambiguity in literature and documentation regarding DSP hardware and software has been the use of the abbreviation “k” (for kilo) to mean both multiples of 1000 and 1024 (210 ). In this guide, we will strive to adhere to the SI convention of using the lower case “k” to denote only multiples of 1000. We will use an upper case “K” when we are talking about multiples of 1024. For example, you should always be able to read 48k as 48000 and 8K as meaning 8192 (8 x 1024).

Properly speaking, 210 probably should be abbreviated “Ki” (short for “Kilobinary”), to disambiguate it from “K” for Kelvin or Karat, however there isn’t much danger of anyone thinking we measure FFT sizes by weight or temperature, and “Ki” has yet to come into very common usage.