Out of curiosity: can anyone w/ the relevant experience explain why older consoles like the NES and the GB have "mirrors" in their memory map? I haven't program

Out of curiosity: can anyone w/ the relevant experience explain why older consoles like the NES and the GB have "mirrors" in their memory map? I haven't programmed embedded systems so I really don't understand what the advantage might be. I thought the Gameboy was odd enough with it's 'echo' of WRAM. (Which isn't even a complete mirror; it stops partway through WRAM#1.) Reading the NES memory map though it gets quite silly! There's a mirror ($2008-$3FFF) that repeats a set of 8 registers ($2000-2007) _every 8 bytes._ Was this simply a matter of having more address space than physical memory? Or were these mirrors actually useful when programming these consoles? --- I thought it might have to do w/ different addressing modes; perhaps you could save some bytes by just storing offsets or something. But that makes no sense in the context of the Gameboy. Many of the z80's extended instructions are gone; the only "interesting" addressing mode is a handful of instructions which treat an 8-bit address as an offset from $FF00.