5DO — Implementation Plan and Missing-Feature Backlog¶

Snapshot: 2026-11-04 Current release: 1.2.1 Repository: https://github.com/purelogiccode/5DO
Branch: master
Status: 1,802 tests pass (last recorded); dotnet build reports 0 warnings / 0 errors; dotnet format --verify-no-changes is clean; jb inspectcode --severity=WARNING reports 0 results.
Goal: implement every Opera core feature, match or exceed Opera's accuracy, and ship the best standalone 3DO emulator on Windows, Linux, and macOS.

This is the single source of truth. It replaces the former 8,000-line dashboard, milestone log, feature spec, parity audit, and session handoff. Historical evidence stays in the linked docs; this file is the canonical backlog.

1. How to use this plan¶

1. Start with the urgent list. Pick the first unchecked item that is not blocked. Do not start optional or research work while a P0 accuracy/compatibility item is open.
2. One feature per change. Do not mix core accuracy, rendering, and UX work in one pull request.
3. Tests first. Every behavior change needs a unit, golden, differential, or scripted manual test.
4. One reference per change. State whether the behavior comes from Opera, 4DO/FreeDO, hardware documentation, or a 5DO-specific decision. Port behavior, not code.
5. Mark it done only when its “Done when” line is satisfied. Then move the entry to the completed log at the end of this file.
6. No duplicate tracking. Each feature has one canonical entry below. IDs are stable. BAK-* and MOON-* are retired aliases (see Appendix B); do not recreate them.
7. No default timing change without full-collection differential evidence.

Legend¶

• [ ] — remaining work. [x] is used only in the completed log.
• Pending — not started, or research/design only.
• Partial — a usable subset ships; the remaining scope is listed in the item.
• Blocked — depends on another item, legal review, or hardware.
• Priority — P0 correctness/compatibility, P1 expected best-in-class, P2 ecosystem multiplier, P3 long-term.
• Effort — S under 3 developer-days, M 3–10, L 10–30, XL 30–90+.

2. Current state¶

Scope decisions (unchanged): WebAssembly/browser, Android/iOS, Raspberry Pi, and Steam Deck are out of scope. Desktop Windows, Linux, and macOS (x64/arm64) remain the supported targets. 4K internal CEL rendering is research, not a release blocker; 4K output upscaling is supported. Distribution remains free, open-source donationware under the FreeDO modified GNU LGPL (non-commercial; voluntary donations only).

2.1 Shipped baseline¶

These are done and must not regress. They are listed here once; the backlog below contains only missing work.

Completed IDs (traceability only): FND-01, FND-02, FND-03, CORE-09, CORE-12, CORE-14, CORE-15, SYS-04, SYS-07, SYS-09, SYS-10, SYS-11, SYS-13, SYS-14, VID-05, VID-06, VID-14, VID-15, AUD-05, UX-01, UX-02, UX-04, UX-11, UX-12, UX-15, UX-19, UX-23, ENG-03, BAK-12, BAK-17.

Milestone aliases: M9R/M9P → P3/P6/TEST-04; M10 → PERF-02; M12 → PERF-08; M13.5/M13.6 → TEST-06; M14 → VID-01/02/03; M15 → CORE-11/16 + FND-05.

3. Missing features — canonical backlog¶

Each entry below is a single feature or a tightly-scoped bundle. The Tracks field preserves legacy IDs. No feature appears in more than one category. A few legacy aliases (BAK-04, BAK-05, BAK-13, BAK-18) intentionally have no separate checkbox; they are covered by the canonical items listed in Appendix B.

3.1 Priority — urgent¶

• CORE-01 — ARM60 cycle-class accuracy mode Status: Partial · Priority: P0 · Effort: M · Tracks: CORE-01 Remaining: run the full-collection differential on timing-sensitive titles; the 2026-11-04 40-title A/B found ARM60 17/40 matches versus FreeDO 15/40 with identical Opera captures, and the decision keeps the ARM60 mode opt-in until the 594-disc evidence is green. Per-class costs and the --cycle-classes ledger are complete and regression-tested. Done when: cycle counts match Opera within documented tolerances; golden hashes are unchanged or improved; no full-collection regression. Ref: opera_arm.c, opera_clock.c.

• CORE-02 — CLIO timer scheduler parity Status: Partial · Priority: P0 · Effort: M · Tracks: CORE-02 Remaining: validate Opera's per-slot scheduler, slack gap, and cascade carry across the full collection; add differential timer evidence; decide whether the Opera scheduler can become the default. The 2026-11-04 increment added 16-bit timer-register masking in Opera mode, per-timer telemetry (--clio-timer-stats), a bounded --clio-timer-trace, a reference-model conformance test, and a --clio-timers A/B override in the differential harness; the mode stays opt-in pending the full 594-disc run. Done when: timer traces match Opera within documented tolerances; no golden-corpus regression. Ref: opera_clio.c, opera_clock.c.

• CORE-03 — MADAM CEL busy timing parity Status: Partial · Priority: P0 · Effort: L · Tracks: CORE-03 Remaining: validate the deferred draw and pixel/texel busy deadline against Opera across CEL-status polling titles; tune the busy budget. The 2026-11-04 increment fixed line-rendered literal/packed pixel accounting and locked the 9M-pixel/s plus 64-cycle-per-CEL formula with tests. Done when: CEL busy traces match Opera within documented tolerances; timing-sensitive titles behave correctly; no frame-time regression. Ref: opera_madam.c.

• CORE-08 — Accuracy/performance profile coverage Status: Partial · Priority: P0 · Effort: S · Tracks: CORE-08 Remaining: profile validation across the full collection; decide when the Accuracy profile can become the default. The 2026-11-04 increment completed the remaining CLI surfaces: bench --profile reports the derived profile and timing modes, selftest validates determinism and state round-trip for every named profile, and the differential harness accepts --profile for a full-collection A/B. A 40-title, 300-frame A/B reports 17/40 matches for both compatible and accuracy with identical per-title statuses; the 594-title A/B remains. The Accuracy profile stays opt-in until that full A/B report shows no regression against the Compatible baseline. Done when: every profile round-trips through CLI/Settings/state and the Accuracy profile can be enabled by default after evidence is green.

• SYS-08 — CPU overclock/underclock validation Status: Complete · Priority: P0 · Effort: S · Tracks: SYS-08 Remaining: none. Non-stock clocks remain experimental and are not seeded into profiles. Done when: the clock matrix is documented and the stock 12.5 MHz default is unchanged. Evidence: Performance.md §8.1 records the 0.5x-4x synthetic matrix; §8.3 adds the 2026-11-05 five-title A/B (Road Rash, Lemmings, Alone in the Dark, Ballz, AutoBahn Tokio). Every non-stock clock changed the frame hash, and AutoBahn Tokio slowed from 9.1 s at 1x to 84.4 s at 2x and 271.7 s at 4x, confirming that non-stock clocks are a compatibility variable rather than a speed switch. The stock 12.5 MHz default is unchanged.

• CORE-04 — CD-ROM command and timing parity Status: Partial · Priority: P0 · Effort: XL · Tracks: CORE-04 Remaining: command latency table; per-command sector readiness; TOC/Sub-Q/raw-sector behavior; data-sector subcode and read-ahead; CD-side tagged status/chunks; retire the READ_ID media-access heuristic; raw-vs-Opera command/status/boot/reset transcripts. Done when: fixed command transcripts match Opera or have documented, tested divergences; the 594-disc sweep has no CD regression; CD-DA and ODE layer cleanly on the same model. Ref: opera_cdrom.c; docs/ReferenceStrategy.md §6.

• CORE-06 — SPORT/XBUS/DIPIR timing and DMA audit Status: Partial · Priority: P0 · Effort: M · Tracks: CORE-06 Remaining: full Opera trace diffs for SPORT/XBUS/DMA; DIPIR activation completion; XBUS poll/FIFO semantics; DMA start/completion traces; PBUS/SPORT edge cases; reset and state interactions. Done when: trace diffs are clean or documented; DMA/FIQ tests stay green; no regression on golden titles.

• CORE-07 — FIQ/IRQ latency and interrupt audit Status: Partial · Priority: P0 · Effort: M · Tracks: CORE-07 Remaining: Opera differential traces for masked/nested/return/re-entry interrupts; request-to-vector latency distributions; CLIO/DIPIR/XBUS ordering; timer and FIQ edge cases; document telemetry fields. Done when: latency metrics match Opera within tolerance; traces and run --json statistics are documented and regression-tested.

• SYS-01 — CD-DA and multi-track completion Status: Partial · Priority: P0 · Effort: L · Tracks: SYS-01 Remaining: CD-DA read-ahead and subcode; a known-title in-emulator CD-DA capture; a 10-second DSP+CD-DA WAV; multi-track edge cases. The 2026-11-05 increment lets ChdDiscImage open audio-only CHDs, gives them a stable TOC-derived game ID, reports them through info (audioCd/trackCount), and includes their metadata in the game database. The BIOS still needs a data track to boot, so the three local [CD-Audio] bonus discs are catalogued but not playable in-emulator yet; the CD device's PLAY_AUDIO_TRACK/mix path remains covered by synthetic tests. Done when: a known CD-DA title plays its audio; a 10-second mixed WAV contains both DSP and CD-DA; data-track titles keep identical hashes.

• SYS-02 — ODE and multi-disc completion Status: Partial · Priority: P0 · Effort: M · Tracks: SYS-02 Remaining: guest-driven ODE swap prompts; a real in-game multi-disc swap with save/state round-trip; playlist edge cases and recovery. Done when: a known multi-disc title swaps in-game and continues; state load after swap is correct.

• SYS-03 — PAL/NTSC completion Status: Partial · Priority: P0 · Effort: L · Tracks: SYS-03 Remaining: PAL2 title evidence; automatic region/profile application; full PAL library validation; 50 Hz audio/pacing evidence; region-aware screenshots, saves, and CLI output. Done when: PAL1/PAL2 titles render and pace correctly; NTSC behavior is unchanged; per-game region selection applies automatically.

• SYS-05 — Embedded open-source boot ROM Status: Pending · Priority: P0 · Effort: M · Tracks: SYS-05 Remaining: a legally cleared embedded open-source Opera Boot ROM; default/fallback selection; no-BIOS homebrew boot; CLI/Settings/state/tests. Done when: homebrew/demos boot without a user BIOS and legal sign-off is recorded. Ref: opera_boot.h, opera_bios.c.

• SYS-06 — RSA/unsigned-disc option Status: Pending · Priority: P0 · Effort: M · Tracks: SYS-06 Remaining: explicit unsigned-disc support (prefer a documented norsa BIOS or user-provided patch); legal review; CLI/Settings/state; tests. Done when: unsigned software runs with an explicit option and legal sign-off is recorded.

• SYS-12 — Portfolio OS SWI HLE Status: Partial · Priority: P0 · Effort: L · Tracks: SYS-12 Remaining: full-collection compatibility/performance validation; decide whether to charge the skipped guest handler cycles; document the titles where HLE changes timing. The 2026-11-04 increment implemented Opera's 14-SWI HLE set (0x50000-0x50012, excluding the object/Dot4 fallbacks), the --swi-hle/Settings/profile opt-in, save-state Options v9, --swi-comments diagnostics, 33 CPU/machine tests that match opera_fixedpoint_math.c, and a real-BIOS synthetic 4x4 benchmark (4,096 calls: 847,860 LLE steps / 4,054,999 cycles versus 24,578 HLE steps / 65,538 cycles). Default-off hashes are unchanged; HLE is intentionally not bit-identical to the guest matrix path because it skips the handler's cycles and register clobbers. Done when: HLE is opt-in, passes supported-title tests, and shows a measured performance gain without behavior changes when disabled. Ref: opera_swi_hle_0x5XXXX.h.

• SYS-15 — Input devices, calibration, and options Status: Partial · Priority: P0 · Effort: L · Tracks: SYS-15 Remaining: per-device calibration; scripted gameplay evidence for mouse, lightgun, arcade lightgun, flightstick, and Orbatak trackball; arcade start/coin/service mapping. The 2026-11-04 increment added the hide-lightgun-crosshair option, eight PBUS ports, an active-device cap of 0-8 (--active-devices/Settings), and a 256-byte PBUS chain buffer. Done when: a representative title works for each device; calibration and crosshair options are documented and tested. Ref: lr_input.c, opera_pbus.c, libretro_core_options.c (opera_active_devices, opera_hide_lightgun_crosshairs).

• SYS-16 — Six-player and multi-tap verification Status: Pending · Priority: P0 · Effort: S · Tracks: SYS-16 Remaining: six-player test ROM/title; hardware multi-tap trace. The 2026-11-04 increment added eight PBUS ports, the Opera chain for ports 7-8, a 256-byte MADAM queue, and an eight-player packet test through FiveDoMachine (EightPlayerPacket_ReachesTheMadamPbusQueue). Done when: six pads and multi-tap are verified on hardware or a test ROM.

• SYS-17 — Game database expansion and community DB Status: Complete · Priority: P0 · Effort: M · Tracks: SYS-17 Remaining: none for the local-collection gate. NVRAM/cheat seeding is not enabled; users supply their own cheat bundles. Done when: at least 95% of local images have metadata and the DB has a repeatable contribution path. Evidence: the embedded database grew from 518 to 653 records. gamedb stats --folder "J:\Panasonic 3DO" reports 1,366 images found, 1,366 readable, 653 unique IDs, 1,366 matched images, 100% coverage, and no errors; the three former exceptions were audio-only [CD-Audio] CHDs, which now open and get TOC-derived IDs. The schema now carries revision, discNumber/discTotal, bios/bios2, fixMode, inputDevices, nvram, and cheats; fixMode drives GameCompatibility, bios/bios2 drive BiosLibrary, and inputDevices drives GameControlDefaults. gamedb scan|export|merge provide the repeatable import/export workflow, and 12 new storage/host/CLI tests cover the parser, merge/save roundtrip, device spec, audio-only IDs, CHD audio-only open, and commands. See docs/GameDatabase.md.

• SYS-18 — Auto per-game profile completion Status: Complete · Priority: P0 · Effort: L · Tracks: SYS-18 Remaining: none for the auto-profile gate. nvram and cheats database fields are carried in profiles but no bundled cheat bundle ships; users supply their own bundle names. Done when: known problem titles boot with zero manual BIOS/region/fix/input steps and profile changes apply live. Evidence: AppSettings.ResolveInputDevices now falls back to the database inputDevices spec, and the CLI applies it when --input-device is absent. EnsureAutoInputDeviceOverride no longer persists a shadowing override for database titles. ShowSettingsAsync builds one effective profile snapshot, applies input/filter/high-resolution changes live, and restarts only when BIOS, NVRAM, fix, or timing fields change. GameProfile carries CheatBundle, and Clone makes profile edits independent. Five-title evidence with --bios-auto and no manual overrides: WorldCup Super Stadium (Japan) selected FZ-10J + Kanji and reached the "Story of the World Cup" screen; Powers Kingdom Europe selected the Anvil BIOS and PAL1 288-line output; Slayer USA selected FZ-10; Crash 'n Burn USA applied FIX_BIT_TIMING_1; Crime Patrol selected the lightgun default and reached its live-action intro. The suite includes 5 new profile/input tests.

• CORE-13 — Remaining compatibility failures Status: Partial · Priority: P0 · Effort: L · Tracks: CORE-13 Remaining: Crash 'n Burn (Japan); any 12k/65k-frame black/static titles; BIOS-aware sweep status merge; title-specific fixes; regression tests and compatibility notes. Done when: 594/594 discs render, or every remaining exception has a root cause and documented workaround; GameCompatibilityList.md is current.

• VID-04 — 480i/240p detection, deinterlacing, and field blending Status: Pending · Priority: P0 · Effort: M · Tracks: VID-04 Remaining: detect field/frame mode; bob/weave/blend options; per-game mode; screenshot and recording behavior; PAL/NTSC interaction; test patterns. Done when: 480i titles render with correct field handling; 240p output is unchanged; options are documented and tested.

• AUD-01 — Audio output quality: resampler, exclusive mode, and sync offset Status: Partial · Priority: P0 · Effort: M · Tracks: AUD-01 Remaining: resampler quality options; exclusive-mode output; audio sync offset; device-change and underrun behavior; latency measurement; documentation. Done when: audio quality/latency targets are met on Windows/Linux/macOS; device switching is safe; no crackle or drift in extended runs.

• ECO-04 — Linux/macOS platform validation and packaging Status: Partial · Priority: P0 · Effort: L · Tracks: ECO-04 Remaining: validate OpenGL/GLES on real Linux and macOS GPUs plus llvmpipe fallback; macOS runtime validation; macOS gamepad backend; Flatpak/AppImage packaging and desktop integration; platform smoke evidence. Done when: full suite + frontend smoke pass on Linux and macOS hardware; Flatpak/AppImage installs and runs; gamepad works on macOS.

3.2 Increase performance¶

• CORE-05 — Optional threaded DSP Status: Pending · Priority: P1 · Effort: M · Tracks: CORE-05 Remaining: bit-identical DSP worker thread; deterministic synchronization with the CPU/audio clock; save-state/rewind compatibility; performance measurement; disabled by default. Done when: threaded and single-threaded runs produce identical audio/state hashes; a measured speed gain exists; fallback is automatic. Ref: opera_lr_dsp.c, opera_lr_dsp_regular.ic, opera_lr_dsp_threaded.ic.

• ENG-02 — Interpreter hot-path optimization Status: Pending · Priority: P1 · Effort: L · Tracks: ENG-02 Remaining: interpreter dispatch/decode/memory fast paths; block cache and invalidation; allocation-free hot loops; flags/PC/cycle parity; benchmark evidence. Done when: hot-path improvements deliver a documented speedup with no accuracy or state change.

• PERF-08 — Managed ARM JIT Status: Partial · Priority: P2 · Effort: XL · Tracks: BAK-10 Remaining: end-to-end throughput is still below the interpreter on the local boot titles (1.17-1.85x slower) despite 0.8-3.3x per-class microbenchmarks; trace/superblock chaining across conditional branches, helper inlining, and a smaller JIT code footprint are the next steps. The interpreter remains the default and the accuracy oracle. Done when: JIT output is bit-identical to the interpreter on the test corpus, delivers at least a 1.5× speedup on hot titles, and falls back safely. Evidence: Arm60.Jit.cs compiles hot straight-line blocks into DynamicMethod IL behind --cpu-backend jit, verifies every instruction against the current code, stops at the machine's 16-cycle batch boundary, and falls back for unsupported instructions, stores (CEL check), FIQ entry, self-modifying code, and runtimes without dynamic code. Correctness: 100,000,000-instruction lockstep parity on Lemmings with identical PC/CPSR/registers/cycles, plus identical 3,000-frame Lemmings and 1,000-frame Road Rash/Powers Kingdom hashes. Performance: DP/load microbenchmarks 1.9-3.3x faster, but end-to-end boot is 1.17-1.85x slower and the 1.5x gate is not met. See Performance.md §8.4.

• PERF-07 — Tier B 1080p60 budgets and profiling Status: Pending · Priority: P1 · Effort: M · Tracks: BAK-06 Remaining: profile integrated GPUs and older CPUs; 1080p60 per-filter budgets; p99 and allocation targets; reference-machine matrix; documented tuning guidance. Done when: Tier B desktop sustains 1080p60 with p99 under budget; allocation-free behavior is tracked by the zero-allocation item.

• PERF-01 — Zero-allocation steady state on every real path Status: Partial · Priority: P1 · Effort: M Remaining: eliminate the ~2.3 KB/frame Avalonia/compositor residual; remove CHD decode allocations from the core stage; prove zero allocations for uploads, readback, audio, and gallery/OSD paths. Evidence: The 2026-11-05 rewind fix pools the capture stream, keyframe arrays, payload buffer, and Brotli scratch, cutting rewind capture allocations from ~75 MB per keyframe to ~1.6 MB with no gen2 collections in a 1,200-frame run. Done when: synthetic and representative real titles report 0 managed bytes/frame after warm-up; CI enforces the gate.

• PERF-02 + BAK-08 — NativeAOT/trimming startup, size, and audio validation Status: Partial · Priority: P2 · Effort: M · Tracks: BAK-08 Remaining: frontend NativeAOT audio validation; startup-time and size targets; trimming compatibility; optional publish profiles; document which artifacts are AOT and which are framework-dependent. Done when: CLI and frontend AOT builds start, render, play audio, and pass smoke tests; startup/size are recorded.

• VID-12 — Latency features: run-ahead, frame delay, VRR, audio sync Status: Pending · Priority: P1 · Effort: L · Tracks: VID-12 Remaining: 1–2 frame run-ahead; frame-delay slider; VRR/adaptive-sync presentation; audio sync adjustment; measured input-to-photon latency; per-game options. Done when: latency options work without desync or state corruption; measurements are documented; defaults remain safe.

• PERF-04 — Host disc streaming and CHD decompression Status: Partial · Priority: P2 · Effort: M Remaining: allocation-free CHD/ISO prefetch under continuous streaming; decompression buffers; cache sizing; read-ahead tuning; avoid frame spikes while keeping the emulated CD model unchanged. The 2026-11-05 increment pools prefetched sector/audio buffers, reuses the worker read buffers, adds an allocation gate for cached reads, and enables the prefetch cache for frontend disc loads. Cache hits do not allocate per read (one-time runtime allocation tolerated), but continuous miss/eviction streaming still needs an end-to-end allocation gate. Random CHD reads still allocate ~12.6 KB per read inside the container reader (5,000 reads allocated 63 MB). Done when: disc streaming does not allocate in steady state or stall presentation; emulated CD timing is unaffected. Note: emulated CD read-ahead/subcode is tracked in the CD items above; this item is host-side only.

• PERF-05 — High-resolution CEL performance Status: Partial · Priority: P2 · Effort: M Remaining: optimize the existing opt-in high-res mode; VRAM plane handling; rasterizer cost; per-title fallback; target 60 fps on Tier A/B where possible. Done when: high-res mode meets its documented frame-time budget and degrades gracefully on slower hardware.

• PERF-06 — GLES 3.1/ANGLE renderer performance Status: Partial · Priority: P2 · Effort: M Remaining: GLES 3.1 feature path; ANGLE parity/performance; state-change reduction; shader/pipeline cache; mobile-desktop GPU validation. Done when: GLES 3.1 and ANGLE match the desktop GL output and meet 1080p60 budgets.

3.3 Optional features¶

• UX-03 — Netplay Status: Pending · Priority: P2 · Effort: XL · Tracks: UX-03 Remaining: lobby/direct connect; lockstep or rollback; NAT/tunneling guidance; desync detection; state handoff; chat/ready flow; tests. Done when: two clients complete a 10-minute session with no unexplained desync; reconnect and state handoff work.

• UX-05 — Achievements Status: Pending · Priority: P2 · Effort: L · Tracks: UX-05 Remaining: provider integration (RetroAchievements or equivalent); offline cache; unlock UI; per-game progress; privacy controls; tests. Done when: achievements unlock and persist across sessions; offline behavior is documented.

• UX-06 — TAS tooling completion Status: Partial · Priority: P2 · Effort: L · Tracks: UX-06 Remaining: full TAS UI; input timeline; multiplayer display; lag detection; movie library; branch comparison; deterministic replay evidence. Done when: a complete TAS workflow (record, edit, branch, verify, share) works from the desktop and CLI.

• UX-24 — Lua scripting and automation Status: Pending · Priority: P3 · Effort: L · Tracks: UX-24 Remaining: sandboxed Lua host; memory/input/state/screenshot APIs; per-game scripts; CLI hooks; examples and documentation. Done when: example scripts automate a test run and a gameplay sequence without destabilizing the core.

• UX-07 — Controller profiles and input polish Status: Partial · Priority: P2 · Effort: M · Tracks: UX-07 Remaining: named profiles; deadzones/curves; hot-plug prompts; import/export; per-game profiles; conflict detection; tests. Done when: profiles round-trip, apply per game, and survive device changes.

• UX-20 + FND-06 — Per-game settings UI completion Status: Partial · Priority: P2 · Effort: M · Tracks: UX-20, FND-06 Remaining: cheat/shader-chain fields; inherit/reset; search/filter; live application; localized labels; profile import/export from the UI. Done when: every shipped option is editable per game, with inherit/reset and immediate effect.

• UX-21/22 — Debugger and profiler suite Status: Pending · Priority: P2 · Effort: L · Tracks: UX-21, UX-22 Remaining: CPU/memory/CEL debugger (DSP debugging is tracked separately in the audio section); breakpoints/watchpoints; step/run-to; trace viewer; frame/performance profiler; exportable traces. Done when: a developer can debug a stuck title and profile a frame without the CLI.

• ECO-09 — 3DO CD player / Video CD Status: Pending · Priority: P3 · Effort: M · Tracks: ECO-09 Remaining: audio-CD player UI; track navigation; VCD/MPEG expansion research and playback if feasible; tests. Done when: a compatible disc plays with correct controls and documented limitations.

• VID-07 — Shader preset system Status: Pending · Priority: P2 · Effort: L · Tracks: VID-07 Remaining: .slang/.glslp-style presets; hot reload; per-game presets; parameter UI; fallback on unsupported GPUs. Done when: popular presets load, render correctly, and fall back safely.

• VID-11 + BAK-19 — Vulkan surfaces on Linux/macOS Status: Pending · Priority: P2 · Effort: L · Tracks: VID-11, BAK-19 Remaining: X11/Wayland surface; macOS Metal/MoltenVK path or compositor embedding; no CPU readback; validation. Done when: the optional Vulkan backend presents natively on Linux and macOS with parity and no readback.

• VID-16 — VDLP output pixel format selection Status: Pending · Priority: P3 · Effort: S Remaining: select 0RGB1555/RGB565/XRGB8888 output; document behavior; parity tests. Done when: the format option changes only host conversion, not emulated output, and is covered by tests. Ref: libretro_core_options.c (opera_vdlp_pixel_format).

• VID-17 — Void-area patterns Status: Pending · Priority: P3 · Effort: S Remaining: 4DO-style void patterns in addition to the solid color; per-game presets; preview. Done when: patterns render identically across presenters and screenshots.

• AUD-03 — DSP debugger Status: Pending · Priority: P3 · Effort: M · Tracks: AUD-03 Remaining: DSP registers/memory; instruction trace; breakpoints; step/run-to; state inspection; integration with the debugger suite. Done when: a DSP program can be debugged interactively and traces export cleanly.

3.4 Future research features¶

• VID-01 + BAK-11 — True 640x480/480i internal rendering Status: Research · Priority: P1 · Effort: XL · Tracks: VID-01, BAK-11 Remaining: measure the current line-doubled mode; prototype true vertical 640x480; scale CEL/VDLP/clipping; independent 480-line output; per-title fallback; performance and visual comparison. Done when: true vertical detail is visible on a 640x480 title and a 3D title, 60 fps is met on Tier A, and the default remains 320x240 + filter. Ref: docs/HighResolutionCelResearch.md.

• VID-02 — 2x/4x supersampling Status: Research · Priority: P2 · Effort: L · Tracks: VID-02 Remaining: 2x/4x supersampling prototype; deterministic downsample; geometry/texture comparison; performance budget. Done when: supersampling improves edges without changing guest VRAM semantics, and a go/no-go decision is recorded.

• VID-03 — GPU CEL compositing Status: Research · Priority: P3 · Effort: XL · Tracks: VID-03 Remaining: parity prototype for packed/literal CELs; GPU CEL architecture; VRAM semantics; fallback; feasibility note. Done when: a prototype matches the software renderer on representative CELs, with a recorded go/no-go decision.

• VID-08 — NTSC composite/S-Video/RF, HDR, and gamma Status: Research · Priority: P2 · Effort: L · Tracks: VID-08 Remaining: composite/S-Video/RF signal models; CRT Royale-class presets; HDR output; gamma/color controls; visual validation. Done when: signal and HDR modes are validated against reference captures and meet performance budgets.

3.5 Testing¶

Testing items are cross-cutting suites and gates. Feature-specific test evidence stays in that feature's “Done when” line.

• CORE-10 — Hardware test ROM suite Status: Pending · Priority: P1 · Effort: L · Tracks: CORE-10 Remaining: test ROMs for CPU, CLIO, MADAM, DSP, SPORT, VDLP, CD, input; golden outputs; CI integration; documentation. Done when: the suite runs in CI and catches timing/behavior regressions without a copyrighted BIOS.

• CORE-11/16 + FND-05 — Full-collection differential and hash regression Status: Partial · Priority: P0 · Effort: L · Tracks: CORE-11, CORE-16, FND-05 Remaining: full 594-title differential run; PC-delta triage; state/CD transcript comparison; frame/audio hash coverage; nightly dashboard; allowlist and triage policy. Done when: the nightly run has no unexplained divergence, and every allowance is documented and dated.

• AUD-04 — Audio golden tests vs Opera Status: Pending · Priority: P1 · Effort: M · Tracks: AUD-04 Remaining: WAV corpus; spectral/RMS/envelope comparison; DSP mode changes; CD-DA mixes; hardware reference captures. Done when: the corpus runs in CI with documented tolerances and catches regressions.

• VID-13 — Video test suite and color accuracy Status: Pending · Priority: P1 · Effort: M · Tracks: VID-13 Remaining: test patterns; RGB555/CLUT/NTSC behavior; CPU/GPU parity; high-res comparison; color delta report. Done when: every presenter/filter passes the suite within documented color tolerances.

• TEST-01 — Save-state migration matrix Status: Pending · Priority: P1 · Effort: M Remaining: test every version bump; v1–v3 states; option/region/disc-hash changes; warn/reject policies; 4DO/Opera interop once save-state interoperability lands. Done when: every supported state version loads or fails with a typed, tested error.

• ENG-01 + ENG-09 — CI quality gates Status: Pending · Priority: P1 · Effort: M · Tracks: ENG-01, ENG-09 Remaining: full OS/arch matrix; coverage gate; static analysis; benchmark regression; SBOM; reproducible builds; six-bundle smoke; artifact checksums. Done when: every release candidate passes all gates and publishes signed, reproducible evidence.

• TEST-02 + BAK-09 — ViewModel and UI tests Status: Pending · Priority: P2 · Effort: M · Tracks: BAK-09 Remaining: settings, library, dialogs, hotkeys, gallery, playlist, and profile ViewModel tests; headless UI smoke; accessibility checks. Done when: UI regressions are caught without manual clicking.

• TEST-04 — GPU and platform CI matrix Status: Pending · Priority: P1 · Effort: L Remaining: GL/GLES/Vulkan/D3D11 coverage; Windows/Linux/macOS; x64/arm64; software fallback; GPU probe artifacts. Done when: each backend is exercised in CI or a documented hardware lab, with parity artifacts.

• TEST-05 — Performance regression matrix Status: Pending · Priority: P1 · Effort: M Remaining: Tier A/B and arm64 baselines; filters; high-res; threaded DSP/JIT; relative >10% gate; published trend. Done when: regressions fail the gate and the trend is visible to maintainers.

• TEST-06 — In-game/visual and high-resolution compatibility review Status: Pending · Priority: P1 · Effort: L Remaining: inspect final frames for every rendered disc; rerun black/static titles at 12,000+ frames; record Unplayable/Playable/Good; repeat with high-res; feed results into the compatibility and high-resolution items. Done when: every local disc has a reviewed status and the compatibility matrix distinguishes loadable, playable, and good.

--

4. Suggested execution order¶

This is an ID-only sequence; descriptions live in the canonical entries above.

1. Foundation: TEST-01, CORE-10.
2. Accuracy: CORE-01, CORE-02, CORE-03, CORE-08, SYS-08, CORE-04, CORE-06, CORE-07.
3. Disc and platform: SYS-01, SYS-02, SYS-03, SYS-05, SYS-06, SYS-07, ECO-04.
4. Opera parity and “just works”: SYS-12, SYS-15, SYS-16, SYS-17, SYS-18, CORE-13.
5. Visual/audio correctness: VID-04, AUD-01, CORE-11/16 + FND-05, AUD-04, VID-13.
6. Performance: CORE-05, ENG-02, PERF-08, PERF-07, PERF-01, TEST-05.
7. Product polish: UX-03..UX-25, ECO-06..ECO-11, VID-07..VID-17, AUD-02/03, SYS-19/21/22, ENG-05.
8. Research: VID-01/02/03, VID-08, ECO-08, RES-01/02.

Rule: do not start step 7 or 8 while a step 2–5 P0 item is open, unless the work is independent and explicitly marked as such.

5. “Best 3DO emulator” gates¶

These are the release/quality gates for the goal, not a second backlog. They reference canonical IDs only.

6. Reference strategy¶

6.1 Primary and secondary references¶

6.2 Rules¶

1. Port behavior, not code. 5DO is a clean C# implementation; do not copy C/C++ source.
2. One reference per change. Record the source in the commit/PR and in docs/PortingNotes.md if the behavior diverges.
3. Tests first. Add the failing test or differential fixture before the fix.
4. Preserve determinism. No wall clock, unordered iteration, or environment-dependent seeds in emulated state.
5. Keep the core single-threaded by default. Threaded DSP/JIT must be opt-in and bit-identical.
6. Do not ship copyrighted BIOS/game content. Embedded boot ROM work needs legal review.
7. Document intentional divergence in docs/PortingNotes.md and docs/Compatibility.md.

7. Verification commands¶

Run these before marking work done.

# Build and test
dotnet build 5DO.sln -c Debug
dotnet test 5DO.sln -c Debug --no-build --no-restore
dotnet format 5DO.sln --verify-no-changes --no-restore

# Static analysis
jb inspectcode 5DO.sln -o=inspectcode.sarif -f=Sarif --severity=WARNING --no-build

# Compatibility sweep
pwsh -NoProfile -File tools/compat-sweep.ps1 -Roots "J:\Panasonic 3DO","G:\MAME\MAME Software List CHDs\3do" -Frames 3000 -Resume

# Differential smoke and showcase
dotnet run --project tools/FiveDo.DiffHarness -c Release -- --manifest benchmarks/diff/smoke.json
dotnet run --project tools/FiveDo.DiffHarness -c Release -- --manifest benchmarks/diff/showcase.json

# Benchmark gate
dotnet run --project benchmarks/FiveDo.Benchmarks -c Release -- --frames 120 --label ci --json --gate

# GPU probes (Windows)
dotnet run --project src/FiveDo.Frontend.Avalonia -c Release -- --gl-probe --gl-probe-backend wgl --gl-probe-capture
dotnet run --project src/FiveDo.Frontend.Avalonia -c Release -- --d3d-probe artifacts/d3d-probe/parity.json --d3d-probe-size 320x240 --d3d-probe-frames 20 --d3d-probe-warmup 5 --d3d-probe-filter accurate --d3d-probe-capture artifacts/d3d-probe/parity.png
dotnet run --project src/FiveDo.Frontend.Avalonia -c Release -- --vk-probe

# Frontend smoke
dotnet run --project src/FiveDo.Frontend.Avalonia -c Release -- --smoke-test

Linux validation: use the local Hyper-V VM LinuxMint (D:\Hyper-V\VMs\LinuxMint). Read AGENTS.local.md for credentials. Publish linux-x64 on the host and copy the output into the guest, or install the .NET SDK in the guest. Record evidence in docs/Compatibility.md §10.3 and docs/KnownIssues.md.

Release: run the commands in ReleaseChecklist.md; build bundles with packaging/package-bundle.ps1; keep a local copy in src/FiveDo.Frontend.Avalonia/bin/Release; remove PDB/XML files; publish SHA256SUMS.txt.

8. Maintenance rules¶

1. This file is the single source of truth. Do not recreate separate plan, backlog, handoff, or roadmap files.
2. One canonical entry per feature. Do not add a second checklist for the same work. Reference IDs instead.
3. Retire duplicates. BAK-* and MOON-* are aliases only; see Appendix B. Do not reintroduce them.
4. Check the box only when “Done when” is satisfied. Include behavior, tests, docs, and no regressions.
5. Move completed items to the completed log. Do not leave them in the missing-feature category.
6. One feature per pull request. No mixed accuracy/rendering/UX changes.
7. No default timing change without full-collection evidence.
8. Keep evidence current. Update TestCorpus.md, Compatibility.md, PortingNotes.md, and KnownIssues.md when behavior changes.
9. Refresh the snapshot monthly or after every release: version, test count, compatibility counts, performance numbers, and platform validation status.

9. Completed log¶

Append new entries here; do not delete history. Each entry is a completed canonical item or a release milestone.

• 2026-11-05 - Rewind capture stall and state-size increment. Fixed the sub-second gameplay freezes caused by the default rewind history capturing a keyframe on the emulation thread every 120 frames. Rewind keyframes now use CompressionLevel.Fastest and unchecked transient CRCs, the memory section (version 4) stores only the live high-resolution RAM range, StateCrc32 is slicing-by-8, and RewindBuffer/StateWriter pool their capture stream, state arrays, payload buffer, and compression scratch. The BIOS MD5 in the header is cached, and the frontend opens discs through the prefetch cache. On Lemmings the worst keyframe capture fell from 139.3 ms to 7.4-9.2 ms (one 14.1 ms outlier), allocations fell from ~1.5 GB to 16 MB per 1,200 frames, and the uncompressed stock-layout state fell from ~26 MB to ~10 MB. See Performance.md §8.5 and SaveStates.md.
• 2026-11-04 - SYS-07 memory capacity options. Added the eight Opera DRAM/VRAM layouts (21, 22, 41, 42, 81, 82, E2, F1) through MemoryConfiguration/MemoryLayout, dynamic RAM/VRAM decode, VRAM base and masks for VDLP/SPORT/CEL, capacity-derived MADAM RED SYSBITS, --memory/Settings/profile/state plumbing, memory state section v3, Options section v8, a homebrew probe ROM that uses 4 MB DRAM + 2 MB VRAM, and 34 memory tests. Stock hashes are unchanged; expanded memory is documented as homebrew-only.
• 2026-11-04 - SYS-08 clock matrix increment. bench --cpu-clock/--memory measure the supported clock range; the 0.5x-4x matrix and unchanged stock default are recorded in Performance.md §8.1.
• 2026-11-05 - Library duplicate visibility. The library identifies discs by the MD5 of their first two data sectors (the 4DO game ID), so renamed copies and GoodTools/Redump/No-Intro variants collapse to one entry. The header now reports 653 games · 1,366 images, and Settings > Paths > Show duplicate disc images lists every file. GameLibraryScanner.BeginScan/InspectPending/Scan accept deduplicate, and library verify checks every image. Added scanner tests for duplicate listing and ImageCount.
• 2026-11-05 - Settings game folders and fast library listing. Added a Game folders editor to Settings > Paths: type or paste one or more folder paths, press Enter or click Add, Browse, or Remove. The list persists in LibraryFolders and refreshes the library when it changes. GameLibraryScanner now builds a per-directory cover index instead of calling File.Exists for four extensions per disc, sibling folder, and cover folder, and LibraryItem decodes covers on bounded background tasks at 256 px so the non-virtualizing grid never blocks the list. On the local 2,732-image collection the cached listing dropped from 2.8 s to 0.33 s (86% of the old time was cover probes), and a cold listing now returns 2,732 placeholders in 17 ms before background metadata hashing and cover decoding. Added cover-index tests for cached entries and PNG-over-JPG precedence.
• 2026-11-05 - PERF-08 managed IL JIT (partial). Added Arm60.Jit.cs, a DynamicMethod JIT behind --cpu-backend jit that compiles hot data-processing, load/store, and branch blocks. Blocks verify every instruction against current memory, stop at the machine's 16-cycle batch boundary, return after stores/branches, and fall back for unsupported instructions, SWIs, mode changes, register shifts, self-modifying code, and runtimes without dynamic code. Correctness: 100M-instruction lockstep parity on Lemmings and identical 3,000-frame Lemmings plus 1,000-frame Road Rash/Powers Kingdom hashes. Per-class microbenchmarks are 1.9-3.3x faster, but end-to-end boot is 1.17-1.85x slower, so PERF-08 remains Partial and the interpreter remains the default. See Performance.md §8.4.
• 2026-11-05 - PERF-04 prefetch buffer increment. PrefetchingDiscSource now reuses evicted cache arrays through a bounded pool, keeps one worker buffer for 2048-byte data reads and one for 2352-byte audio reads, and clears the pool on dispose. Cache hits are allocation-free on the caller thread and covered by a 1,000-read test. Continuous miss/eviction streaming still needs an end-to-end allocation gate.
• 2026-11-05 - SYS-18 auto per-game profile completion. Database inputDevices now drives AppSettings.ResolveInputDevices and CLI runs without --input-device; auto-input seeding no longer persists a shadowing override. The settings dialog applies one effective profile snapshot, updates input/filter/high-resolution live, and restarts only for BIOS/NVRAM/fix/timing changes. GameProfile gained a CheatBundle field and Clone. Evidence: WorldCup Super Stadium (JP) auto-selected FZ-10J + Kanji and rendered "Story of the World Cup"; Powers Kingdom Europe auto-selected Anvil/PAL1; Slayer USA auto-selected FZ-10; Crash 'n Burn USA applied fix bit 1; Crime Patrol auto-selected the lightgun and rendered its intro.
• 2026-11-05 - SYS-17 game database expansion. Grew GameDatabase.xml from 518 to 653 records, added revision, discNumber/discTotal, bios/bios2, fixMode, inputDevices, nvram, and cheats, and wired fixMode into GameCompatibility, BIOS fields into BiosLibrary, and input fields into GameControlDefaults. Added gamedb stats|scan|export|merge, the filename parser, serial retry for transient CHD read failures, and docs/GameDatabase.md. gamedb stats --folder "J:\Panasonic 3DO" reports 1,366 readable images, 653 unique IDs, and 100% metadata coverage with no errors; the three former exceptions were audio-only [CD-Audio] CHDs, which now open and get TOC-derived IDs.
• 2026-11-05 - SYS-01 audio-only CHD increment. ChdDiscImage no longer requires a data track, audio-only discs get a stable TOC-derived GameId, and info reports audioCd/trackCount. This let the game database catalogue the three local [CD-Audio] bonus discs. In-emulator playback still needs a disc with a data track or a guest CD player, so SYS-01 remains partial.
• 2026-11-05 - SYS-08 CPU clock validation. Recorded the five-title 0.5x/1x/2x/4x A/B in Performance.md §8.3. Every non-stock clock changed the frame hash, and AutoBahn Tokio's wall time rose from 9.1 s at 1x to 84.4 s at 2x and 271.7 s at 4x. The stock 12.5 MHz default and the no-per-title-clock decision are unchanged.
• 2026-11-04 - SYS-12 Portfolio OS SWI HLE. Ported Opera's 14 fixed-point math SWIs (0x50000-0x50012, with the object/Dot4 cases left to the guest), added --swi-hle/Settings/profile/state v9 plumbing, --swi-comments histograms, 33 CPU/machine tests that match opera_fixedpoint_math.c, and a real-BIOS synthetic 4x4 benchmark. Default-off hashes are unchanged; HLE is an opt-in performance mode and intentionally skips the guest handler's cycles and register clobbers, so some titles (AutoBahn Tokio, Ballz) change frame hashes.
• 2026-11-04 - FND-01 machine option registry. Added MachineOptionRegistry (13 validated options with CLI/Settings/profile/state mappings), typed MachineOptionValidationException/UnknownMachineOptionException, Opera DIPIR and soft-reset controls, --opera-dipir/--opera-soft-reset/--cpu-backend, settings options generated documentation, docs/MachineOptions.md, save-state Options v7 coverage for every option, mismatch warnings, and CLI/Settings/profile parity tests.
• 2026-11-04 - Plan rewrite. Consolidated the former dashboard, parts I-IV, feature spec, parity audit, and handoff into this single backlog; retired duplicate BAK-*/MOON-* trackers.
• 2026-11-03 — v1.2.1. Six x64/arm64 bundles plus SHA256SUMS.txt; CI verified every apphost machine type and every checksum.
• 2026-11-03 — v1.2.0. Direct3D 11 presenter, startup release checks, deterministic fuzz harness, per-edge overscan/void border, screenshot gallery filters/compare/export, display-refresh control, flat CHD/raw ISO detection, FIQ/DSP telemetry, CLIO/DIPIR/FIFO/MCTL/XBUS audit fixes.
• 2026-11-02 — Hardening pass. Fuzz harness, overscan profiles, gallery features, FIQ request-to-vector latency, DSP per-tick statistics, 1,622 tests.
• 2026-11-01 — Disc/UX/accuracy pass. CD-DA and multi-track, M3U/live swap, PAL/NTSC, BIOS manager, per-game NVRAM, rewind, fast-forward, cheats, TAS, disc browser, input devices, OSD/quick menu, library metadata, portable mode, instance lock, save-state format v3.
• Earlier: v1.0.0 documented-compatibility release; 594-disc sweep; legal/attribution sign-off; six-RID packaging; OpenGL/GLES, Vulkan, and Direct3D 11 presenters; 4K output; differential harness.

Appendix A — Opera option coverage¶

This is a crosswalk, not a second backlog. “Canonical item” is the only place the remaining work is tracked.

Additional 4DO/Opera UX gaps already covered elsewhere: physical CD drive (SYS-21), disc conversion (SYS-22), void patterns (VID-17), auto-load/auto-open (shipped), localization (UX-09), installers/file associations (UX-16), session focus behavior (UX-25).

Appendix B — Retired and duplicate IDs¶

Dropped scope (do not add checkboxes): WebAssembly/browser (ECO-02), Android/iOS (ECO-03), Raspberry Pi/arm64 GLES optimization (ECO-05), Steam Deck. The libretro core wrapper is not planned; 5DO stays standalone.

Appendix C — Source-of-truth links¶

• Status — short current answer.
• Compatibility — supported set and platform evidence.
• Known issues — current limitations.
• Performance — budgets and measurements.
• Test strategy and Test corpus — test evidence.
• Differential harness — Opera comparison.
• High-resolution CEL research — VID-01/02/03.
• Release checklist — release gate.
• Porting notes — intentional divergences.
• Reference strategy — Opera/4DO mapping.