+mov ax, cx ; ax = idx
+mov bl, COLS ; bl = COLS
+div bl ; al = idx / COLS ; ah = idx % COLS
+
+mov [bp - 1], al ; row = idx / COLS
+mov [bp - 2], ah ; row = idx % COLS
+mov byte [bp - 3], 0 ; neighbours = 0
+
+mov byte [bp - 4], -1 ; i = -1
+.i: ; do {
+
+ mov byte [bp - 5], -1 ; j = -1
+.j: ; do {
+
+ mov al, [bp - 4]
+ or al, [bp - 5]
+ jz .continue ; if (!i && !j) continue
+
+ mov al, [bp - 1]
+ add al, [bp - 4] ; al = row + i
+
+ cmp al, 0
+ jl .continue ; if (row + i < 0) continue
+
+ cmp al, ROWS
+ jge .continue ; if (row + i >= ROWS) continue
+
+ mov ah, [bp - 2]
+ add ah, [bp - 5] ; ah = col + j
+
+ cmp ah, 0
+ jl .continue ; if (col + j < 0) continue
+
+ cmp ah, COLS
+ jge .continue ; if (col + j >= COLS) continue
+
+ movzx bx, ah ; bx = col + j
+
+ mov ah, COLS
+ mul ah ; ax = (row + i) * COLS
+
+ add bx, ax ; bx = cell index
+ shl bx, 1 ; bx = grid offset
+
+ mov dx, [es:si + bx] ; dl = cell state
+
+ test dl, ALIVE ; if (!ALIVE) continue
+ jne .continue
+
+ inc byte [bp - 3] ; neighbours++
+
+.continue:
+ inc byte [bp - 5]
+ cmp byte [bp - 5], 1
+ jle .j ; } while (++j <= 1)
+
+ inc byte [bp - 4]
+ cmp byte [bp - 4], 1
+ jle .i ; } while (++i <= 1)
+
+movzx ax, byte [bp - 3] ; ax = # alive neighbours
+
+leave