4/4 perf: alphabet ranking, SWAR key packing, and doubling-round overhead - #6

Open

BenjaminDEMAILLE wants to merge 16 commits into

COMBINE-lab:mainfrom

BenjaminDEMAILLE:stack/4-key-packing

Commits on Aug 13, 2026

feat: radix-seeded prefix doubling for the plain in-memory SA

BenjaminDEMAILLE
and
claude
committed
ffe3680
feat: `verify_sa`, an O(n) independent suffix-array check

BenjaminDEMAILLE
and
claude
committed
148f266
perf: parallelise grouping and rank publication in the doubling path

BenjaminDEMAILLE
and
claude
committed
9a39247
bench: add a reproducible chr21 harness

BenjaminDEMAILLE
and
claude
committed
9369034
docs: document the fast path, and correct two stale claims

BenjaminDEMAILLE
and
claude
committed
cda07cb
perf: replace the seed comparison sort with an MSD counting sort

BenjaminDEMAILLE
and
claude
committed
f5ef8e2
feat: extend the doubling path to subsets and to in-memory sample sort

BenjaminDEMAILLE
and
claude
committed
fe411d8
docs: refresh the fast-path numbers and coverage

BenjaminDEMAILLE
and
claude
committed
14f3107
perf: rank the alphabet before packing keys

BenjaminDEMAILLE
and
claude
committed
ef6a78b
docs: record the alphabet-ranking result

BenjaminDEMAILLE
and
claude
committed
f9c42c9
perf: pack keys with a SWAR gather over a pre-ranked text

BenjaminDEMAILLE
and
claude
committed
83f721e
docs: refresh in-memory numbers after the SWAR key packing

BenjaminDEMAILLE
and
claude
committed
cd28bfb
perf: drop the per-group allocation and prepass from doubling rounds

BenjaminDEMAILLE
and
claude
committed
54b6c57
docs: refresh in-memory numbers after the doubling-round cleanup

BenjaminDEMAILLE
and
claude
committed
3400b1c
refactor: drop the external-memory seed from the doubling module

BenjaminDEMAILLE
and
claude
committed
5101601
perf: stop materialising the seed key array, and narrow the group bounds

BenjaminDEMAILLE
and
claude
committed
94921f6

Navigation Menu

Uh oh!

4/4 perf: alphabet ranking, SWAR key packing, and doubling-round overhead - #6

4/4 perf: alphabet ranking, SWAR key packing, and doubling-round overhead#6
BenjaminDEMAILLE wants to merge 16 commits into
COMBINE-lab:mainfrom
BenjaminDEMAILLE:stack/4-key-packing

Commits on Aug 13, 2026

feat: radix-seeded prefix doubling for the plain in-memory SA

feat: `verify_sa`, an O(n) independent suffix-array check

perf: parallelise grouping and rank publication in the doubling path

bench: add a reproducible chr21 harness

docs: document the fast path, and correct two stale claims

perf: replace the seed comparison sort with an MSD counting sort

feat: extend the doubling path to subsets and to in-memory sample sort

docs: refresh the fast-path numbers and coverage

perf: rank the alphabet before packing keys

docs: record the alphabet-ranking result

perf: pack keys with a SWAR gather over a pre-ranked text

docs: refresh in-memory numbers after the SWAR key packing

perf: drop the per-group allocation and prepass from doubling rounds

docs: refresh in-memory numbers after the doubling-round cleanup

refactor: drop the external-memory seed from the doubling module

perf: stop materialising the seed key array, and narrow the group bounds