0 Fork 1

mirror of https://github.com/jkdansereau/goldeneye-pc-port.git synced 2026-10-23 15:52:42 +02:00

Files

T

James Dansereau bf195d3eb5 port(determ): experimental GE_DETERM=1 fixed-tick mode, not yet achieved (D117)

Adds an env-gated (GE_DETERM=1, default OFF), test-only deterministic-clock
mode to port/src/libultra.c per the design in docs/dev/findings.md section D117:
osGetCount() returns a virtual clock advanced synchronously in osRecvMesg()
when __scMain (the sole consumer of the scheduler's retrace queue) asks for
the next message, instead of tracking real elapsed time. Goal: two runs of
the same build take the same simulation path, for exact frame-diff
regression testing and reproducible crash repro.

Four passes this session, each is a real, verified result, not a guess:
1. Fixed the predicted pre-first-frame deadlock (a poll-driven fallback
   caused it in the first cut).
2. Replaced the poll-driven fallback with synchronous per-ask generation
   in osRecvMesg -- fixed the deadlock cleanly, but two-run divergence was
   still 33-38%.
3. Capped the pre-first-task tick backlog to a fixed constant -- divergence
   dropped to 17-20%, still not the ~0% target.
4. Added GE_DETERM_TRACE=1 instrumented tracing and diffed two runs' clock
   event logs directly. Proved the tick mechanism is now fully
   deterministic (byte-identical clock values through the first real task).
   Root-caused the actual remaining divergence: every synthesized retrace
   increments the scheduler's frameCount, which gates a half-rate client
   (the audio client) by parity -- since the pre-task spin count is still
   real-time-scheduling-dependent, that parity isn't guaranteed to match
   between runs.

GE_DETERM is NOT achieving its determinism goal yet and is not meant to be
used for anything today -- it ships here purely as a well-documented,
zero-risk-when-off checkpoint (verify.sh bunker1 PASS, worst_cell unchanged,
confirmed after every change this session) so the next attempt starts from
proof instead of re-deriving four sessions of dead ends. Full trace numbers,
exact line citations, and the mechanism: docs/dev/findings.md section D117
(four M-52 addenda).

Claude-Session: https://claude.ai/code/session_01P3WrUHAFQZUuhrHhiJgTUr

2026-09-04 22:10:58 -04:00

fast3d

Merge branch 'main' into feat/f10-options-overlay

2026-09-03 10:03:23 -04:00

include

port: D184 F10 overlay v2 — mouse control + resolution/fullscreen rows

2026-09-03 09:54:21 -04:00

shim

linux-runtime: declare getenv()/realloc() for the strict-host build

2026-09-02 23:03:46 -04:00

src

port(determ): experimental GE_DETERM=1 fixed-tick mode, not yet achieved (D117)

2026-09-04 22:10:58 -04:00

README.md

Prepare for public release

2026-09-01 19:28:59 -04:00

README.md

GoldenEye 007 — PC port layer

This directory contains the PC port layer: the code that lets the (unmodified) GoldenEye game sources run on a modern host instead of N64 hardware.

It mirrors the layout of the Perfect Dark PC port, which ports the same Rare engine family. See docs/internals.md (repo root) for the full research write-up and phased plan.

Layout

port/
├── fast3d/            # Software RSP: interprets GBI display lists -> OpenGL
│   ├── gfx_pc.cpp     #   RSP interpreter (matrices, lights, tris, fog)
│   ├── gfx_opengl.cpp #   OpenGL rendering backend (shaders, textures, zbuf)
│   ├── gfx_sdl2.cpp   #   SDL2 window-manager backend
│   ├── gfx_cc.cpp     #   color-combiner -> GLSL shader compiler
│   └── ...
├── include/           # port-facing headers
│   ├── audio.h  config.h  fs.h  input.h  mixer.h
│   ├── platform.h  romdata.h  system.h  video.h  utils.h
│   └── versioninfo.h.in
└── src/
    ├── main.c         # PC entry point (replaces boot.s / init())
    ├── libultra.c     # shims for the libultra OS API
    ├── video.c        # SDL2 window + GL context + frame pacing
    ├── audio.c        # SDL audio device, buffer queueing
    ├── mixer.c        # audio mixing
    ├── input.c        # SDL2 keyboard/mouse/gamepad -> N64 controller structs
    ├── fs.c           # filesystem abstraction over the ROM + data dir
    ├── romdata.c      # loads the .z64 ROM from disk, sets up segments
    ├── config.c       # INI config (ge007.ini)
    ├── system.c       # platform primitives (time, logging, paths)
    ├── crash.c        # crash handler / stack traces
    ├── gesched.c      # replacement RCP scheduler that drives the software RSP
    └── utils.c

How it works (short version)

The R4300 game code (src/game, src) is compiled unmodified.
The RSP is emulated in software (fast3d/): it interprets the Gfx display list the game builds and emits OpenGL calls, bypassing the RDP.
The RCP scheduler (gesched.c) replaces src/sched.c's hardware task submission: instead of writing RSP registers it calls gfx_run().
libultra OS functions are shimmed in libultra.c (single-threaded, host clock, VI/PI/SI/AI mapped onto the video/audio/fs/input layers).
Saves (Memory Pak / EEPROM) are backed by files in the data dir.

GE-specific RSP work

The PD fast3d already handles G_TRI4 (GE's 4-triangle command). The new work for GE is:

G_SETTEX — GE's custom texture-bank command (include/gbi_extension.h, gsSPUseTexture). Must be added to the RSP interpreter + GL backend.
Verify GE's custom color-combiner modes (G_CC_*FADE*) and render mode (G_RM_CUSTOM_AA_ZB_XLU_SURF) produce correct GLSL / GL state.

Use rsp/graphics/gmain.s (the real GE RSP ucode) as the ground truth for the exact command encodings.

Status

Scaffolding only. The files in src/ and fast3d/ are stubs that define the interfaces and mark the work to be done. The actual implementation is brought in phase by phase (see docs/internals.md §8), largely by adapting the PD port's port/ layer.