Peephole Optimizer

Post-compilation optimizer that applies target-specific sliding-window pattern matching to reduce instruction count. Operates on the generated assembly before it reaches the assembler.

Location: hcbcomp/peep.c / hcbcomp/peep.h

Z80 Optimizations

Pattern	Before	After	Saved
Local variable load	ld hl, -N ex de, hl push ix pop hl add hl, de ld a, [hl] inc hl ld h, [hl] ld l, a	ld l, [ix+N] ld h, [ix+N+1]	7
Local store (immediate)	ld hl, -N / ex de, hl / push ix pop hl / add hl, de / push hl ld hl, IMM / pop de / ex de, hl ld [hl], e / inc hl / ld [hl], d	ld hl, IMM ld [ix+N], l ld [ix+N+1], h	9

Pattern

Before

After

Saved

Local variable load

ld hl, -N
ex de, hl
push ix
pop hl
add hl, de
ld a, [hl]
inc hl
ld h, [hl]
ld l, a

ld l, [ix+N]
ld h, [ix+N+1]

Local store (immediate)

ld hl, -N / ex de, hl / push ix
pop hl / add hl, de / push hl
ld hl, IMM / pop de / ex de, hl
ld [hl], e / inc hl / ld [hl], d

ld hl, IMM
ld [ix+N], l
ld [ix+N+1], h

8086 Optimizations

Pattern	Before	After	Saved
LEA + deref chain	`lea ax,[bp-N]; mov bx,ax; mov ax,[bx]`	`mov ax, [bp-N]`	2
Global dereference	`mov ax,label; mov bx,ax; mov ax,[bx]`	`mov ax, [label]`	2
Store immediate to local	`lea ax,[bp-N]; push ax; mov ax,IMM; pop bx; mov [bx],ax`	`mov word [bp-N], IMM`	4
Push/Pop copy	`push ax; pop bx`	`mov bx, ax`	1

Typical Results

Target	Reduction
Z80	~32% fewer instruction lines
8086	~28% fewer instruction lines