Lotus
Experimental lossless compression — enwik9 scores
—
—
—
Dataset: enwik9 — the first 1,000,000,000 bytes of English Wikipedia from the Large Text Compression Benchmark. Rankings are only officially comparable on that same file.
Same enwik9 input. Neighbors are the five LTCB compressors just smaller and just larger than Lotus. Lotus is measured on this server at deploy time.
| Codec | Mode | Compressed size | Ratio |
|---|
MK1–MK9 wait on one shared 256→18 cover and hex bag. Packing workers start after intern; each model then gets its own progress.
| Version | State | .lts | Bag | Total | Compress | Decompress |
|---|
O1 covers 256→18. O2 covers 256→26, then reverse-merges adjacent tiles. O3 reuses existing tiles before adding leftover chunks. Models in this range pack in parallel with the MK-line.
| Version | State | .lts | Bag | Total | Compress | Decompress |
|---|
| Bag length | Entries | Hex | Binary | Delta | Encoding | DB3 |
|---|
Full enwik9 as O1, plus gzip -9 as a baseline. Lower is better on every chart. Memory is peak resident set size (RSS).
25 sequential 10 MB imports from enwik9: cover first so the chart fills, then reverse optimization as a second interruptible job.
Current Lotus development score: —
This ladder is estimated from published LTCB compressed sizes. It is not an official ranking until Lotus is listed on that benchmark.
Lotus is an experimental lossless compression system exploring alternative ways to represent and transmit large amounts of information efficiently.