How is array access translated to assembly addressing?

array[i] becomes scaled-indexed addressing: (base, index, element_size), computing base + index × element_size.

Array indexing in C is really just address arithmetic, and the x86 scaled-indexed mode expresses it in a single operand. The scale equals sizeof(element), which is exactly why the legal scales are 1, 2, 4, and 8.

int arr[100];
int x = arr[i];

# arr address in %rax, i in %rcx
movl (%rax,%rcx,4), %edx   # %edx = arr[i], because int is 4 bytes

The address %rax + %rcx×4 is precisely &arr[i]. Change the type, change the scale:

movb (%rax,%rcx,1), %dl    # char  (scale 1)
movw (%rax,%rcx,2), %dx    # short (scale 2)
movl (%rax,%rcx,4), %edx   # int   (scale 4)
movq (%rax,%rcx,8), %rdx   # long  (scale 8)

Reverse-engineering tip: spotting a ×2, ×4, or ×8 scale in an addressing mode immediately tells you the element size, and therefore the likely C type of the array.

Go deeper:

• Addressing mode (Wikipedia) — how scaled-indexed mode maps onto array[i] with scale = sizeof(element).