Research on Walnut Games

Excluded games re-evaluated (research, 2026-10-19)

Mon, 01 Jan 0001 00:00:00 +0000

The second pass over games first ruled out for reasons other than GL 3: RAM measured on the PC (amd64, Xvfb + llvmpipe, which adds about 100-200 MB), build and big-endian costs estimated. The first pass is build/tmp/newgames/RESEARCH.md, summarised in the games’ technical notes (gamedb/games//technical.md). Builds of DDNet, OpenTyrian, UQM and Hedgewars started the same night.

Where this lives now (2026-10-26). This survey moved from docs/research to gamedb/research. Every game it names has a bundle in gamedb/games/ (a draft when it is only a candidate, status and status_note saying where it stands) whose technical.md cites this file by section. Part B (OpenRCT2’s big-endian port estimate) moved into the game’s investigations/. The per-game tables and blocks here are kept by hand for now; they become generated blocks (tools/gamedb/blocks.py) in a later step, and until then a game’s current state is its bundle’s, not this file’s.

Game candidates for the suite (research, 2026-10-16)

Mon, 01 Jan 0001 00:00:00 +0000

This is the survey that picked the benchmark suite, done on 2026-10-16 before any of these games ran on the PS3. Step 0 had rested on OpenArena alone, which misled us about VRAM and sound, and the SPU offload and memory decisions needed 5-10 titles. Every pick has run since, and so have most reserves: Yamagi Quake II, Extreme Tux Racer and Armagetron joined the suite, Teeworlds, SuperTux and Red Eclipse were smoke-tested, TORCS waits for the user, and Nexuiz runs on our own older DarkPlaces build. Current results are in games/README.md; where this survey and the measurements differ, the measurements win.

GL 3 census: open-source games that need GL 3+, and what it takes to run them on the PS3 (2026-10-19)

Mon, 01 Jan 0001 00:00:00 +0000

The question: which open-source games need GL 3+ (or GLES 3), and what would it cost to get each one working on our RSX (Mesa nv30: GL 2.1, GLSL 1.20, GLES 2.0) with everything else that comes with it: memory, CPU, big-endian ppc64, data. Less bling is fine; not working is not. Games new on the PS3 come first. Vulkan-only engines are out.

The per-game detail is in the six parts next to this file, one section per game in a fixed template, with file:line citations: A, space, B, 2D and Java, C, RTS, D, FPS, E, id Tech, Doom, FreeSpace and F, engines and runtimes; the driver-level analysis is G-driver.md (it stays in docs/research: it is about the driver, not a game). The brief the agents worked from was not kept. This file is the cross-game view. Labels as in the parts: measured, read in source, estimate, UNVERIFIED.

Part A: space games (Endless Sky, Naev)

Mon, 01 Jan 0001 00:00:00 +0000

Endless Sky (own C++ engine, SDL2 + GLEW, upstream master 2026-10-18; Debian sid/ports 0.11.2-1)

Verdict: the only game of the pair worth porting, and upstream already ships an “experimental OpenGL 2” path (3D textures instead of texture arrays, no VAO, EXT FBO) that needs GLSL 1.20 shader files supplied as a plugin, so effort is M. The blocker is memory: 1.3 GB RSS on the PC with the lowest texture settings, 477 MB of it GPU textures; it fits 256 MB only after DXT compression plus more cuts, and that is unproven. ppc64 binary exists in debian-ports, novelty is 5 (no PS3 port found).

Part B: 2D games and Java (Taisei, OpenClonk, Hypersomnia, DDNet, Mindustry)

Mon, 01 Jan 0001 00:00:00 +0000

All measurements: Debian sid amd64 in a podman box, Xvfb + llvmpipe (Mesa 26.2.5, GL 4.6 core), no audio device. Taisei at 960x540, OpenClonk and Hypersomnia at 1280x720, Mindustry at 1280x720 (llvmpipe at 1080p ran at 7 fps for Taisei and starved the game logic, so the smaller windows were used; texture MB do not depend on window size, framebuffer MB do). llvmpipe baseline ~100 MB is subtracted for PS3 estimates. Reports (small) are under /mnt/data_linux/Temp/ps3/gl3census/traces/{taisei,clonk,hyper,mindustry}/c*/report.md; big traces and downloads were deleted.

Part C: RTS games (Warzone 2100, OpenRA, Beyond All Reason / Recoil, 0 A.D.)

Mon, 01 Jan 0001 00:00:00 +0000

Measured 2026-10-19 on the PC (Debian sid box wz, Xvfb 1920x1080, llvmpipe, Debian package 4.7.0-3). Census report kept at /mnt/data_linux/Temp/ps3/gl3census/traces/wz/c47/report.md, mem files in work/wz/.

Warzone 2100 (own engine, SDL2 + GLAD, QuickJS; master of 2026-10-17 = 4.7.x, Debian sid 4.7.0-3)

Verdict: the current line (4.4 to 4.7) needs GL 3.0/3.1+ and is not worth porting, but release 4.3.5 (2023-04-16) is the last one that runs on plain GL 2.1 or GLES 2.0 with GLSL 1.20 shaders, no instancing, no UBOs, no texture arrays. So the route is “build 4.3.x, change nothing in the renderer, maybe cut textures”. Effort S to M, RAM is the open question (estimate 300-450 MB, so it needs cuts to fit; not 256 MB friendly), novelty 5. ppc64: Debian ports ships 4.5.5deb-1+b1 on ppc64 (so the game compiles and links big-endian); 4.5.5 itself needs GL 3.0 though.

Part D: first-person shooters (Xonotic, Unvanquished, Red Eclipse 2, Warfork, OpenMoHAA, ET: Legacy)

Mon, 01 Jan 0001 00:00:00 +0000

Xonotic (DarkPlaces, release 0.8.6 of 2023-06-20 and engine git master of 2026-01-22)

Verdict: Xonotic 0.8.6 is NOT GL3-blocked: its own DarkPlaces still has the GL 1.1/1.3/2.0 renderpaths and runs on GL 2.1 + GLSL 1.20 as shipped (measured). Only engine master (and Debian’s darkplaces) dropped them for GL 3.2 core / GLES2. Route: build the 0.8.6 engine for ppc64 like darkplaces-gl2 (effort S-M), keep the GL20 path. The cost is memory (about 200 MB of textures, 850 MB PC RSS) and 450 draws/frame: a “needs cuts” game, not a port. FEATURES (0.8.6 GL20 path): glsl_min=120 core_only=n fallback=gl2 (also gl13, gl11) int_ops=no bitwise=no texelFetch=no textureSize=no ubo=no (guarded, 19 programs link at #version 120) instancing=no tex_arrays=no float_tex=opt float_rt=opt (HDR/bloom only) rg_tex=no mrt=1 vao=no sync=no geom=no tf=no compute=no derivatives=opt (offsetmapping, GLSL130 only) textureLod_fs=no vertexid=no flat=no arb_fbo=opt (EXT_fbo is enough; blit only for video capture) srgb_fb=no vtf=no dyn_loops=opt (blur/SSAO-style loops, bloom) big_shaders=opt (permutation uber-shader, 19 programs in the lowest preset, each of 74 KB source before preprocessing; the per-permutation size is UNVERIFIED on nv30; Nexuiz Classic’s older shader already needs the lean config) FEATURES (engine master, GL32 path): glsl_min=140 if available else 120 core_only=y (context request) fallback=none; GLES2 path exists, see section 1. int_ops=no ubo=opt (r_glsl_skeletal only, GLSL 140) vao=req (core profile) blit=opt SCORES: popularity=5 novelty=5 playability_256MB=2 effort=M ppc64=not-in-debian (no xonotic package in sid; engine builds on Debian ports as darkplaces 020180908+git20260122, which is GL32/GLES2 only) data=free size_MB=1180 (zip 1.24 GB; data pk3 317 + maps 626 + nexcompat 125 + music 111)

Part E: id Tech 4, the Doom source ports, FreeSpace 2 Open

Mon, 01 Jan 0001 00:00:00 +0000

The agent for this part stopped at the usage limit after its measurements; the coordinator wrote the part from its saved runs ($B/traces/dhewm3, $B/traces/gzdoom) and from the clones it left ($B/src/{dhewm3,gzdoom,tdm,fs2open.github.com,doomsday}). Clone dates: dhewm3 455b88e 2026-06-08, GZDoom c26ce2e 2026-08-10, fs2open 2026-10-18.

dhewm3 (Doom 3 GPL engine, 1.5.5 / git 455b88e)

Verdict: not GL3-blocked. It renders with GL 1.x fixed function plus ARB_vertex_program / ARB_fragment_program, which nv30 exposes; big-endian is a first-class build option. The blocker is memory: the free Doom 3 demo’s first map peaked at 1161 MB on the PC. Effort M (our ppc64 build, texture downsizing and compression, memory cuts), playable only if those cuts land. FEATURES: glsl_min=none (ARB assembly) core_only=n fallback=gl1 int_ops=no bitwise=no texelFetch=no textureSize=no ubo=no instancing=no tex_arrays=no float_tex=no float_rt=no rg_tex=no mrt=1 vao=no sync=no geom=no tf=no compute=no derivatives=no textureLod_fs=no vertexid=no flat=no arb_fbo=no srgb_fb=no vtf=no dyn_loops=no big_shaders=no (ARB programs written for NV30/R300-class cards) SCORES: popularity=4 novelty=3 playability_256MB=2 effort=M ppc64=not-in-debian (contrib, arch any: our build) data=non-free (free demo, 462 MB) size_MB=462 (demo) / ~1500 (retail)

Part F: engines and runtimes (OpenMW, Barotrauma, Godot 3 games, SDL3 GPU / raylib, skips, scan)

Mon, 01 Jan 0001 00:00:00 +0000

Measured on the PC in podman boxes (Xvfb 1920x1080, llvmpipe, apitrace). Everything not marked “measured” or “read in source (file:line)” is UNVERIFIED or an estimate.

OpenMW (OpenSceneGraph, 0.51.0, git 2026-10-18 master)

Verdict: Already GL 2.1 / GLSL 1.20 in its default configuration, so it is not GL3-blocked. Real work is big-endian (ESM and BSA readers do no byte swapping, so Morrowind.esm cannot be read on ppc64 today; M), then CPU and memory tuning for a 3.2 GHz in-order PPE with ~160 MB. Debian-ports ppc64 builds 0.51.0-1+b4. Data is non-free (Morrowind) unless one plays the CC0 Template and Example Suite. Novel on the PS3: yes.

Windows PC classics (1990-2010) with open engines

Mon, 01 Jan 0001 00:00:00 +0000

Popular PC games of 1990-2010 that now have an open-source engine, a decompilation, a reimplementation or an official source release, as candidates for the PS3. Researched 2026-10-20 from the projects' READMEs, osgameclones.com, PCGamingWiki, the awesome-unofficial-pc-ports list, GamingOnLinux and news articles. Nothing here has been built or measured on our console yet. Every alternative project per game, with GitHub stars, last activity and the pick for our port, is in Alternatives and popularity.