Seed TotK prewarm from the recorded M2 iPad route (#261)

Add 1,537 shader, 2,512 render-pipeline and 43 compute-pipeline rows from
the update 1.4.3 capture, preserving first-use order. Document the
author-reported route and results, and limit the coverage claim to
that route.

This profile-only capture is separate from the shared CPU and loader
optimizations. Validation checks the row counts, uniqueness and exact
data/order preservation; no new cold-start or gameplay run here.

Co-authored-by: draco-cong <71372970+draco-cong@users.noreply.github.com>
Co-authored-by: laobeibei <laobeibei@users.noreply.github.com>
This commit is contained in:
committed 2026-10-26 15:28:59 +08:00
1 parent 5632203274
commit 58f74014f4
2 files changed
+4095

No files matched your search

+1
View File
@@ -59,6 +59,7 @@ Game-specific findings. The general host changes they led to are listed under [F
## Files and shaders
- **Current: prewarm capture reported in [PR #261](https://github.com/seedds/nxnative/pull/261).** The seed in `pso_manifest.tsv` contains 1,537 shader, 2,512 render-pipeline and 43 compute-pipeline rows, recorded on an M2 iPad Pro with update 1.4.3 on 2026-10-25 over about 33 minutes: boot, title, a save load and overworld exploration. The author reports no pipeline compilations in the second session's long frames with this list. This covers the recorded route, not the whole game; no new gameplay or cold-install validation here. The capture is separate from the shared deswizzle and symbol-index optimizations.
- Most RomFS files are zstd with dictionaries (`*.zs`); the dictionaries are `Pack/ZsDic.pack.zs`, which is also the profile's `romfs_marker`.
- Shaders: `Shader/*.bfsha.zs` (BNSH; `material.bfsha` holds most) and the layout shaders in `Shader/ApplicationPackage.Nin_NX_NVN.release.sarc.zs`.
- **Resolved in the host: duplicate pipeline keys from unused colour state.** MacNativeGames commit `f3e2f66`, reported in [PR #196](https://github.com/seedds/nxnative/pull/196), found 8,479 of 24,146 recorded render pipelines (35%) differed only in unused state. The general normalization and prewarm deduplication are ported; this is the source project's measurement, with no new gameplay or prewarm-count measurement on this game here.
File diff suppressed because it is too large. Load diff