build(release): contributed pipeline lists from #2034 (3 lists) #2445

Merged

hckrman101 merged 1 commit from build/contrib-lists-2034 into main

2026-11-11 19:39:20 +00:00

hckrman101 commented

2026-11-11 19:27:30 +00:00

(Migrated from github.com)

Adds the three pipeline lists posted in #2034 that aren't in main or in an open pull request yet. I added each one with nx2/pipeline_states.py add, so each list keeps only the pipelines that no other contributed list already has. Each one has a note beside it.

What's added

List	From	GPU, OS	Build and settings	Route	Pipelines sent	Kept in the list	New to the shipped list
`contrib/yeetbot90-amd-rx-9070-xt-fsr3-4k`	@yeetbot90 (second list, "MORE PIPELINES")	AMD RX 9070 XT, driver 32.0.32015.7029, Windows 11	v2.1.0, 3840x2160, FSR 3.1 Auto, frame generation 3x, HDR on	Kokiri Forest through Jabu-Jabu's Belly	2,047	1,013	992
`contrib/jabbaxipil-rtx-4070-linux-dlaa`	@jabbaxipil	NVIDIA RTX 4070, driver 615.71.09, Ubuntu 26.04.1 (Wayland)	nightly v2.1.0-rc.3-46, 2560x1440, DLAA (model K), frame generation 3x	child, by day: Kokiri Forest out to Gerudo Valley	960	778	710
`contrib/disgustingusting-web-intel-uhd-intro`	@disgustingusting-web	Intel UHD (i3-1115G4, integrated), driver 32.0.101.7088, Windows 11	v2.1.0, 1920x1080, no frame generation	the intro only	9	6	0 (the tour's list already has all 6)

All three are version 2 lists (HKSTATE2). The merged list grows from 25,793 to 27,441 pipelines (1,648 new; the two big lists share 54). I kept the Intel list even though it adds nothing today: the tour's list is replaced whole each time, and this one holds pipelines no other contributed list has.

What's left out

@pmwork-personal's first list, @yeetbot90's first list and @Imasvpp's list: already in main (#2075 and #2082).
Lists from @Brett-D (#2143), @pmwork-personal (#2133) and @jarynclouatre (#2064): each is in its contributor's own open pull request.
@disgustingusting-web's hookshot_log.txt: it's a log, not a list.

How it was checked

I downloaded each zip into its own empty folder outside the repository and listed it without extracting. Each one held a single hookshot_pipeline_states.bin: no paths, no symlinks, nothing encrypted, and sane sizes and compression ratios. Only that one file was extracted, with a size cap.
Each file starts with HKSTATE2, uses 552-byte records, has a whole number of records, and has no duplicates.
I also checked every field of every record: mesh, sample, depth and flag fields in range, no attributes or answers past their counts, padding zero, and each variant matching its answers' hash. The only exceptions were records written with no answers, which tour.bin.xz has in the same form.
python -I nx2/pipeline_states.py check: passes. The merge gives the same bytes twice.
Tests: tests/test_pipeline_state_sources.py, test_ship_state_list.py, test_pipeline_states.py, test_pipeline_notes_open.py, test_pipeline_state_counts.py and test_pipeline_state_portable.py all pass.

Not tested

I didn't run the game: no first-start shader preparation with the new shipped list, no game build, and no check that these pipelines remove stutter on these GPUs.

Adds the three pipeline lists posted in #2034 that aren't in `main` or in an open pull request yet. I added each one with `nx2/pipeline_states.py add`, so each list keeps only the pipelines that no other contributed list already has. Each one has a note beside it. ## What's added | List | From | GPU, OS | Build and settings | Route | Pipelines sent | Kept in the list | New to the shipped list | |---|---|---|---|---|---|---|---| | `contrib/yeetbot90-amd-rx-9070-xt-fsr3-4k` | @yeetbot90 (second list, "MORE PIPELINES") | AMD RX 9070 XT, driver 32.0.32015.7029, Windows 11 | v2.1.0, 3840x2160, FSR 3.1 Auto, frame generation 3x, HDR on | Kokiri Forest through Jabu-Jabu's Belly | 2,047 | 1,013 | 992 | | `contrib/jabbaxipil-rtx-4070-linux-dlaa` | @jabbaxipil | NVIDIA RTX 4070, driver 615.71.09, Ubuntu 26.04.1 (Wayland) | nightly v2.1.0-rc.3-46, 2560x1440, DLAA (model K), frame generation 3x | child, by day: Kokiri Forest out to Gerudo Valley | 960 | 778 | 710 | | `contrib/disgustingusting-web-intel-uhd-intro` | @disgustingusting-web | Intel UHD (i3-1115G4, integrated), driver 32.0.101.7088, Windows 11 | v2.1.0, 1920x1080, no frame generation | the intro only | 9 | 6 | 0 (the tour's list already has all 6) | All three are version 2 lists (`HKSTATE2`). The merged list grows from 25,793 to 27,441 pipelines (1,648 new; the two big lists share 54). I kept the Intel list even though it adds nothing today: the tour's list is replaced whole each time, and this one holds pipelines no other contributed list has. ## What's left out - @pmwork-personal's first list, @yeetbot90's first list and @Imasvpp's list: already in `main` (#2075 and #2082). - Lists from @Brett-D (#2143), @pmwork-personal (#2133) and @jarynclouatre (#2064): each is in its contributor's own open pull request. - @disgustingusting-web's `hookshot_log.txt`: it's a log, not a list. ## How it was checked - I downloaded each zip into its own empty folder outside the repository and listed it without extracting. Each one held a single `hookshot_pipeline_states.bin`: no paths, no symlinks, nothing encrypted, and sane sizes and compression ratios. Only that one file was extracted, with a size cap. - Each file starts with `HKSTATE2`, uses 552-byte records, has a whole number of records, and has no duplicates. - I also checked every field of every record: mesh, sample, depth and flag fields in range, no attributes or answers past their counts, padding zero, and each variant matching its answers' hash. The only exceptions were records written with no answers, which `tour.bin.xz` has in the same form. - `python -I nx2/pipeline_states.py check`: passes. The merge gives the same bytes twice. - Tests: `tests/test_pipeline_state_sources.py`, `test_ship_state_list.py`, `test_pipeline_states.py`, `test_pipeline_notes_open.py`, `test_pipeline_state_counts.py` and `test_pipeline_state_portable.py` all pass. ## Not tested - I didn't run the game: no first-start shader preparation with the new shipped list, no game build, and no check that these pipelines remove stutter on these GPUs.