[bymaxmarc]

To be taken after the host PRs it depends on (listed in the note's "Depends on" and below); the branch itself is off main and touches only the profile and docs.

What and why

Nintendo Switch Sports Resort, update v65536 (the whole game, packed from a NUP with no base game), SDK 22.2.10 built for the Switch 2 (Ounce), Nintendo EPD's sead / agl / gsys with Havok, NVN2.

• Profile dm/games/04008E001F098000/: the NSP script's provisional game.tsv (its heap and address-region checks: sys_resource, mem_reserve, mem_alias, mem_stack, app_pool) plus depth_mode zero and render_cond on; the shader tables the script generates (shader_map.tsv, shader_facts.tsv, shader_pairs.tsv, pso_pairs.tsv, bf_want.tsv, shader_map.tsv.conflicts.tsv: the same kinds of file as PRAGMATA's profile, no extracted game data); and pso_manifest.tsv with the 31 task / mesh pipelines the offline archive cannot hold (22 from the device list over boot, menus and sports, 9 more from a 37-minute session that built them at their first draw for up to 0.8 s each). No patch, no rewriter cut.
• Note dm/games/04008E001F098000/README.md: the tested build and device, what each key is for, the pipelines, frame rate and scene loads, motion controls, the Mac route, what is open, and the PRs it depends on.
• Its row in the games table and its line in README's Switch 2 list. The Home button protection entry of services.md, which already named the game, links the note.

Depends on: #2376, #2377, #2380 (boot and saves); #2384, #2385, #2386 (controllers and motion); #2373, #2375, #2372 (pictures); #2371, #2374, #2378, #2379 (pipelines and frame time); #2381 (diagnostics). Their reference entries name the game without a link while it has no note on main; once both are in, the links and the note's "Found on this game" lines can be added (I can send that as a follow-up).

Remaining limitations, as the note records them: the model heaps the game sizes from its resource size table are overrun by the host's command records, and the sessions still ran with the DM_ROMFS_OVERLAY stopgap (patched resource size tables in the player's env.txt, outside the repository); the system data 0400000000000849 the game mounts as MiiResource is not present on the host (a red line; nothing visibly missing noticed); six AnnaSea mesh shaders are refused (task payloads over Metal's 16 KB); the court's first frame takes about 225 ms (scene loads); iPhone and iPad not built.

Checked

• Tests on this branch (main ccd834e + 1 commit), Apple-silicon Mac (M2 Max, 32 GB), macOS 27.2:
  • dm/tools/tests/test_game_profile.py: 7 tests OK; test_docs.py: 6 tests OK.
  • python -m unittest discover -s dm/tools/tests: 1,469 tests OK (27 skipped).
  • make -k -C dm/gfx/tests GAME=04008E001F098000 -j8 run: every test passes (the CPU and Metal tests, with this profile's header).
• Games run: Nintendo Switch Sports Resort 04008E001F098000 v65536 on the user's Mac (M2 Max, macOS 27.2), as a Mac app with its full offline pipeline archive (710.7 MB), built from the fork's work branch (an earlier main plus the PRs above): boot, menus, the motion calibration, and matches with motion from a right Joy-Con 2 (Bluetooth LE, DM_SW2=1) and from a DualSense, confirmed by the user. The latest, a 37-minute session on 2026-11-11 with the Joy-Con 2: no error, no compute pipeline built at run time, frame time p50 33.3 ms (the game presented at 30 fps there); earlier, the tennis match at 60 fps (p50 16.67 ms, p95 17.2 ms). Red lines in the latest session log of 2026-11-11: the MiiResource mount, two texture type mismatch lines (noted as open) and one audio device stall of 213 ms (the silent sink stood in).
• Not run: the game from this branch (it needs the PRs above; the profile was taken unchanged from the work branch, so no new build was made from it), scripts/regress.py (no host change here), iPhone / iPad (no IPA built), sports other than tennis in a scripted route.

Docs

• The finding is in its game note; the games table and README list are updated; python3 dm/tools/docs_symptoms.py was run (no change)
• No game files, keys or data extracted from a game in the change or the attachments (the shader tables hold hashes, sizes and resource paths, as every profile's)

[FifteenPen]

How does this differ from #2107? Looks like you went and did a bunch of stuff I already did.

[seedds]

Landed on main as f4b7e77 (profile, note, games table row and README line), with you credited as co-author: game.tsv keeps your comments and render_cond, and the note cites your 37-minute session with real controllers. Taken from your set: #2376, #2377, #2380, #2384, #2385 and #2375. The shading-rate fix landed from FifteenPen's #2103 instead of #2373 (the same value in EmitSReg, which both instruction sets reach). DM_ROMFS_OVERLAY was not needed on current main: a doubles match loaded with its NPCs. pso_manifest.tsv is a list recorded on the current host, since the earlier lists' pipeline keys no longer match it. #2386 is answered on its own page; #2371, #2372, #2374, #2378, #2379 and #2381 are still to be judged on their own. Thank you!