Bit width multiple of 16

[wpace918]

Why are bus widths are multiple of 16 bits

 

[gamerk316]

Meaks sense. In X86 assembly [the primary platform for most devices], the size of a single program word is 16 bits. Which makes sense, as the size of standard X86 registers is 16 bits. Granted, they've been expanded over the years [the extended registers can now hold 256-bit AVX instructions], but the "base" registers [AX, BX, CX, etc] are still 16 bit registers. So unless you wanted to break a lot of old X86 code that operates with the base registers, you have to have bus widths that increase by 16 bits at a time.

 

[hunter315]

Most of it is because a byte is 8 bits so everything has to be in multiples of that, you cannot pull half a byte from memory that would go horribly wrong. Most things come in multiples of two, so either you pull two single words from memory or you pull one double word from memory or you have a 32 bit bus that can pull a four word from memory, thats about as large as it usually gets though for words at a time, but a wider bus lets you get more at a time, so 16 comes out to be a fairly convenient size to increase by.