Next Emulator Implementation Recommendation

Date: January 2026
Status: Proposal
Recommended System: Sega Master System / Game Gear

Executive Summary

After comprehensive analysis of well-documented retro gaming systems and evaluation of Hemulator's existing architecture, Sega Master System (SMS) and Game Gear are recommended as the next emulator implementations. These systems offer:

  1. Optimal difficulty level: Simpler than modern systems (GBA, SNES) but more feature-complete than already-implemented 8-bit systems
  2. Excellent documentation: Extensive technical references available from Sega, SMS Power!, and the emulation community
  3. Code reuse: Can leverage existing Z80 CPU core and audio components
  4. Dual implementation: Master System and Game Gear share ~95% of their architecture, providing two systems for minimal extra effort

Current Implementation Status

Completed Systems (Production Ready)

  • ✅ NES: ~90% game coverage, 14 mappers, complete CPU/PPU/APU
  • ✅ Atari 2600: Complete TIA, RIOT, cartridge banking
  • ✅ Game Boy: ~95% game coverage, MBC0/1/2/3/5 support

In-Development Systems

  • 🚧 SNES: Basic infrastructure, minimal PPU, no APU/input
  • 🚧 N64: 3D rendering functional, limited game support
  • 🧪 PC: Experimental, CGA/EGA/VGA modes

Why Sega Master System?

1. Well-Documented Hardware

The SMS has exceptional technical documentation:

2. Appropriate Complexity

SMS strikes the perfect balance:

Simpler than:

  • GBA (ARM CPU, complex PPU modes, DMA)
  • SNES (65C816, Mode 7, DSP)
  • PlayStation (MIPS + GPU + complex 3D pipeline)

More feature-rich than:

  • Already-completed NES and Atari 2600
  • Offers new challenges without overwhelming complexity

3. Excellent Code Reuse Opportunities

Hemulator's existing architecture is well-suited for SMS:

CPU: Z80 (Already Implemented)

  • Location: crates/core/src/cpu_z80.rs
  • Current State: Stub implementation with registers and state management
  • Required Work: Complete instruction set implementation (~200-250 opcodes)
  • Reference: Comprehensive Z80 documentation already exists in docs/references/cpu_z80.md

Audio: SN76489 PSG

  • Channels: 3 square wave + 1 noise (similar to NES APU structure)
  • Reusable Components:
    • PulseChannel (for square waves)
    • NoiseChannel (for noise)
    • PolynomialCounter (for LFSR-based noise)
  • Location: crates/core/src/apu/ has all needed building blocks
  • New Work: System-specific wrapper for SN76489/SN76496 variant

Graphics: VDP

  • Based on: TMS9918A (well-documented, simpler than NES PPU)
  • Features: Tilemap-based, sprite system, 16KB VRAM
  • Resolution: 256×192 (similar complexity to NES 256×240)
  • Reusable: Existing tile decoder utilities from crates/core/src/ppu/

4. Two Systems for One Effort

Master System and Game Gear share ~95% of their code:

Component Master System Game Gear Notes
CPU Z80 @ 3.58 MHz Z80 @ 3.58 MHz Identical
VDP Chip 315-5124/5246 315-5378 Minor differences
PSG SN76496 (mono) SN76496 (stereo) Same chip, different output
Resolution 256×192 160×144 Different display window
Colors 64 palette (32 on-screen) 4096 palette (32 on-screen) Extended color space
Input Gamepad Gamepad + Start Nearly identical

Implementation Strategy:

  1. Build Master System first
  2. Add Game Gear as variant with:
    • Different resolution/viewport
    • Extended color palette (12-bit vs 6-bit)
    • Stereo audio panning

5. Mature Ecosystem for Testing

Test Resources:

  • Homebrew test ROMs available
  • Well-known game library for compatibility testing
  • Existing emulators for validation (Gearsystem is open-source)

ROM Detection:

  • Simple header detection (similar to existing NES/GB loaders)
  • Common formats: .sms, .gg (raw binary with optional headers)

Technical Implementation Roadmap

Phase 1: Z80 CPU Completion (1-2 weeks)

  • [ ] Complete Z80 instruction set (~200-250 opcodes)
  • [ ] Implement all addressing modes
  • [ ] Add interrupt support (IM 1 primarily, IM 2 for completeness)
  • [ ] Shadow registers and index registers (IX/IY)
  • [ ] Comprehensive unit tests
  • [ ] Cycle-accurate timing

Estimated LOC: ~1,500-2,000 (similar to existing cpu_6502.rs)

Phase 2: SN76489 PSG Audio (1 week)

  • [ ] Create Sn76489Psg struct using existing components
  • [ ] 3 square wave channels (reuse PulseChannel)
  • [ ] 1 noise channel (reuse NoiseChannel + PolynomialCounter)
  • [ ] Register interface (frequency, volume, control)
  • [ ] Implement Sega variant differences (16-bit LFSR)
  • [ ] Unit tests for each channel

Estimated LOC: ~400-600

Phase 3: VDP Graphics (2-3 weeks)

  • [ ] VDP state machine and registers
  • [ ] 16KB VRAM management
  • [ ] Tilemap rendering (background)
  • [ ] Sprite engine (64 sprites, size modes)
  • [ ] Color palette (64 colors, 32 simultaneous)
  • [ ] Scrolling support
  • [ ] Line interrupt timing
  • [ ] Implement as Renderer trait

Estimated LOC: ~1,000-1,500

Phase 4: Master System Integration (1 week)

  • [ ] Memory map implementation (ROM, RAM, I/O ports)
  • [ ] Input handling (controller ports)
  • [ ] System struct implementing System trait
  • [ ] ROM loading and cartridge banking
  • [ ] Save state support
  • [ ] Integration tests

Estimated LOC: ~800-1,000

Phase 5: Game Gear Extension (3-5 days)

  • [ ] Extended color palette (4096 colors)
  • [ ] Resolution/viewport adjustment (160×144)
  • [ ] Stereo audio panning
  • [ ] Game Gear-specific input (Start button)
  • [ ] ROM format detection

Estimated LOC: ~200-300 (mostly configuration differences)

Phase 6: Testing and Refinement (1 week)

  • [ ] Test ROM implementation for smoke tests
  • [ ] Game compatibility testing
  • [ ] Timing refinement
  • [ ] Bug fixes and edge cases

Total Estimated Time: 6-8 weeks
Total Estimated LOC: ~4,000-5,500 (comparable to existing system implementations)

Comparison with Alternatives

Game Boy Advance

Pros:

  • Popular system with large game library
  • Well-documented

Cons:

  • ARM7TDMI CPU (new architecture, no existing implementation)
  • Complex PPU with multiple modes, rotoscale, affine transforms
  • DMA and advanced hardware features
  • Estimated 8-12 weeks implementation time

Sega Genesis/Mega Drive

Pros:

  • Z80 + 68000 CPUs (could reuse Z80 for sound)
  • Popular system

Cons:

  • Requires new 68000 CPU implementation
  • Complex VDP with multiple modes
  • FM synthesis audio (YM2612) is more complex
  • Estimated 10-14 weeks implementation time

Neo Geo Pocket (Color)

Pros:

  • Well-documented
  • Simple architecture

Cons:

  • Less popular/smaller library
  • TLCS-900H CPU (new architecture)
  • Less code reuse opportunity
  • Estimated 6-8 weeks but with less benefit

Alignment with Project Goals

Hemulator's Implementation Philosophy

From AGENTS.md:

"Always prefer full, tested implementations of each module/component, even if all parts aren't immediately used"

SMS/Game Gear fit this perfectly:

  • Full Z80 implementation will be reusable for future systems (ZX Spectrum, MSX, CP/M)
  • SN76489 PSG is used in many systems (ColecoVision, TI-99/4A, BBC Micro)
  • VDP techniques applicable to similar tilemap-based systems

Code Reuse Benefits

  • Z80 can later be used for: ZX Spectrum, MSX, ColecoVision, Amstrad CPC
  • SN76489 PSG for: ColecoVision, TI-99/4A, Tandy 1000, BBC Micro
  • Tilemap rendering patterns for: Various 8/16-bit systems

Success Criteria

Minimum Viable Implementation:

  • [ ] Z80 CPU with full instruction set
  • [ ] SN76489 PSG with all 4 channels
  • [ ] VDP with background and sprite rendering
  • [ ] Master System ROM loading and execution
  • [ ] 80%+ compatibility with commercial games

Stretch Goals:

  • [ ] Game Gear support
  • [ ] FM sound unit (optional, Master System only)
  • [ ] Light gun support
  • [ ] 90%+ game compatibility
  • [ ] Comprehensive test suite

Risk Assessment

Low Risk:

  • ✅ Well-documented hardware
  • ✅ Existing CPU stub to build upon
  • ✅ Proven audio component architecture
  • ✅ Similar complexity to completed systems

Medium Risk:

  • ⚠️ VDP timing accuracy (requires careful testing)
  • ⚠️ Cartridge mapper variety (though less than NES)

Mitigation:

  • Use existing emulators (Gearsystem) for validation
  • Implement test ROM suite early
  • Leverage SMS Power! community resources

Conclusion

Sega Master System and Game Gear are the optimal next implementation for Hemulator due to:

  1. Perfect difficulty balance: Challenging enough to be rewarding, simple enough to be achievable
  2. Excellent documentation: Multiple high-quality technical references
  3. Maximum code reuse: Leverages existing Z80 CPU and audio components
  4. Two systems for one: Game Gear comes almost free with Master System
  5. Future-proof: Z80 and SN76489 implementations will benefit future systems
  6. Active community: Resources and testing support readily available

Estimated implementation time: 6-8 weeks for complete Master System + Game Gear support

Recommendation: Proceed with Sega Master System implementation as the next major system addition to Hemulator.

References

  1. Sega Master System Architecture - Rodrigo Copetti
  2. SEGA Mk3 Hardware Reference Manual - Archive.org
  3. SMS Power! Development Documents
  4. Charles MacDonald's VDP Documentation
  5. SN76489 PSG Documentation - Wikipedia
  6. VGMPF SN76489 Technical Reference
  7. Z80 CPU Documentation - Hemulator