jan/rrt
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Preserve Tier2 header crossover matrix

Browse source
This commit is contained in:
Jan Petykiewicz 2026-04-21 18:42:19 -07:00
commit 9c6acbd4b3
1 changed files with 30 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

33
docs/rehost-queue/tier2-candidate-row-families-2026-04-21.md
View file

@ -55,6 +55,30 @@ The checked header-cluster export keeps the root scan bounded too:
So the candidate-table header side is no longer an unbounded per-map surface either.
## Header / Row / Trailer Crossover
The header-cluster export and named-run export now join cleanly enough to keep one more matrix
explicit:
- `0x00000000 / 0x00000000 :: 35/43 :: 0x00000001 -> 19`
- `0x00000000 / 0x00000000 :: 35/43 :: 0x00000000 -> 2`
- `0x00000000 / 0x00000000 :: 10/18 :: 0x00000001 -> 2`
- `0x00000000 / 0x00000000 :: 10/18 :: 0x00000000 -> 4`
- `0xcdcdcdcd / 0xcdcdcdcd :: 35/43 :: 0x00000001 -> 5`
- `0xcdcdcdcd / 0xcdcdcdcd :: 35/43 :: 0x00000000 -> 3`
- `0xcdcdcdcd / 0xcdcdcdcd :: 10/18 :: 0x00000001 -> 1`
- `0x10000000 / 0x00009000 :: 35/43 :: 0x00000001 -> 1`
That keeps the mixed families concrete:
- the off-row `10/18` subset is mostly the dominant `0x00000000 / 0x00000000` header family,
but it keeps one `0xcdcdcdcd / 0xcdcdcdcd` outlier
- the ordinary-row `35/43 :: 0x00000000` subset is mixed across both main header families
- `Alternate USA.gmp` stays the lone `0x10000000 / 0x00009000` outlier and still lands on the
ordinary `35/43 :: 0x00000001` side
So the row/trailer split still does not collapse to one header family either.
## Crossover Reading
The new crossover field keeps the row-family and trailer-family interaction explicit too:
@ -97,6 +121,9 @@ This keeps the Tier-2 frontier narrower than a generic “candidate table varies
probe-bearing maps
- only the `00` pair moves, and it does so in one small `7`-map family
- the numbered trailer split stays mixed across both row-pair families instead of collapsing to
the off-row subset, so the next owner question is still the earlier seed or projection seam
that feeds nonzero `[candidate+0xba/+0xbb]` into `0x00412d70`, not the mere existence of the
stable named rows themselves
the off-row subset
- the row/trailer split stays mixed across both main header families too, instead of collapsing to
one header word pair
- so the next owner question is still the earlier seed or projection seam that feeds nonzero
`[candidate+0xba/+0xbb]` into `0x00412d70`, not the mere existence of the stable named rows or
one fixed header family by itself