; SPDX-License-Identifier: GPL-3.0-or-later ; Authored differential fixture, adapted from VCC Plus's GPL reference loop: ; Public source discussion: https://vccplus-git.brianholman.com/VCCPlus/VCCPlusOSS/src/branch/main/docs/macos/CPU_CORE.md ; Explicit initialization avoids comparing undefined reset-register contents. ; Python run.py emits the exact bytes below; no assembler is required. ; ; Common ROM variant starts at $8000. The native variant prefixes LDMD #$01 ; (11 3D 01), moving the common code and stop address three bytes higher. ; CPU vectors all point to $8000. Unused ROM bytes are $FF. org $8000 start orcc #$50 ; 1A 50 IRQ/FIRQ remain masked lds #$7000 ; 10 CE 70 00 ldu #$0000 ; CE 00 00 ldy #$0000 ; 10 8E 00 00 clra ; 4F tfr a,dp ; 1F 8B ldb #16 ; C6 10 ldx #$4100 ; 8E 41 00 ; Comparison begins here, after every compared common register is initialized. loop inca ; 4C sta ,x+ ; A7 80 decb ; 5A bne loop ; 26 FA ldy #$4100 ; 10 8E 41 00 ldb #16 ; C6 10 clra ; 4F sum adda ,y+ ; AB A0 decb ; 5A bne sum ; 26 FB done bra done ; 20 FE, stopped BEFORE execution ; ; Expected after 115 compared instructions: ; PC=$8027 (native $802A); A=$88; B=$00; CC=$54; DP=$00; ; X=Y=$4110; U=$0000; S=$7000; RAM $4100-$410F contains bytes 1..16. ; The guest reads back all sixteen bytes through ADDA ,Y+; each trace accumulator ; delta identifies the corresponding byte modulo 256, not just a final checksum. ; E/F/V are intentionally neither initialized nor compared. Native MD is ; observed in VCC's snapshot; XRoar's text trace exposes the LDMD instruction ; but does not directly expose MD.