bbs.profullstack/logicsrc
1
0
Fork 0
mirror of https://github.com/profullstack/logicsrc.git synced 2026-10-01 20:33:50 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 4
logicsrc/docs/openstream/reports/2026-09-12-nixamp-0.17.1-synthetic.md
Anthony Ettinger 9f42ce222a
docs: OpenStream benchmark reports, published per release (#150) 
Adds a reports section to the OpenStream spec so its claims rest on a
reproducible measurement rather than an assertion. Each report is a run of
the envelope over a defined corpus on real hardware: proof that
decompression restores every byte, that an incompressible input costs only
the framing overhead, that a compressible one saves what it claims against
the complete wire size, and how long each codec takes.

- docs/openstream/reports/ holds a machine-readable <id>.json (canonical,
  with a versioned schema) and a rendered <id>.md per report, plus a README
  on the shape and on submitting one. The seed report is nixamp 0.17.1 over
  the synthetic corpus, labelled synthetic so no one reads a padded-fixture
  number as production.
- The site renders them at /docs/openstream/reports (index) and
  /docs/openstream/reports/<id> (one report), under the dynamic /docs/[slug]
  tree so the reports routes never shadow a spec's own doc page. A small
  lib/reports.ts reads the JSON at build time; REPORTED_SPECS keeps the
  route surface explicit. sitemap includes the index and every report.
- The spec doc gains a Benchmark reports section linking there, and repeats
  the honest caveats: OpenStream frames Zstandard and gzip rather than being
  a new algorithm, synthetic padding flatters a codec, an efficient real
  feed saves little, and round-trip exactness is the one pass/fail.

The report format is produced by `nixamp compression benchmark` (in the
nixamp repo); a release runs it and commits the two files here.


Claude-Session: https://claude.ai/code/session_01MxNif5tsYq4LczgG7aE8Jp

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
2026-09-12 06:23:46 -07:00

2 KiB
Raw Permalink Blame History

OpenStream benchmark — nixamp 0.17.1

Generated 2026-09-12T13:13:15.713Z · envelope NXS1 · schema 1

Environment. bun 1.4.0, zstd f8745da6ff1ad1e7bab384bd1f9d742439278e99, zlib 12731092979c6d07f42da27da673a9f6c7b13586, on linux 7.0.0-30-generic x64, 8× DO-Premium-Intel, 15.6 GiB.

Policy. block 262144 B; eligible at 3% and 512 B; zstd levels 1, 3, 9.

Corpus

sample kind bytes container best mode saves
random-1mib synthetic 1048576 bytes stored stored
zeros-1mib synthetic 1048576 bytes zstd L9 99.96%
text-repeat-1mib synthetic 1048576 bytes zstd L9 99.95%
ts-padded-50pct synthetic 752000 mpegts zstd L9 50.09%
ts-unpadded synthetic 752000 mpegts stored stored
tiny-3b synthetic 3 bytes stored stored
empty synthetic 0 bytes stored stored

Aggregate, per mode across the corpus

mode wire bytes saving enc ms dec ms round trip
stored 4651091 -0.03% 0 0 ok
gzip L6 2189399 +52.91% 96.34 13.78 ok
zstd L1 2180200 +53.11% 18.84 23.12 ok
zstd L3 2177968 +53.16% 24.23 11.67 ok
zstd L9 2177476 +53.17% 53.51 13.5 ok
ts-zstd L1 1127961 +25% 15.4 9.8 ok

Caveats

  • Envelope v1: a 16-byte stream header, a 48-byte header per frame, plus one end frame.
  • OpenStream is a framing envelope over Zstandard and gzip, not a new compression algorithm; these numbers are those codecs at the block boundary, honestly framed.
  • Synthetic samples do not predict production savings. A padded transport stream flatters a codec by its padding; an efficient real feed saves far less. Use --corpus with authorized real samples for numbers that mean something.
  • Timings are wall-clock on the machine and runtime named in environment and do not transfer to other hardware.
  • roundTrip:false in any row is a failure of exactness and must block a release.