proxy.golang.org : github.com/axiomhq/fsst
Package fsst provides fast string compression via learned symbol tables. FSST (Fast Static Symbol Table) is a compression algorithm optimized for strings with repetitive patterns. It learns up to 255 symbols (1-8 bytes each) from training data and encodes text by replacing matches with single-byte codes. FSST excels at compressing: Typical compression ratios: 1.5x to 3x, depending on repetitiveness. FSST is not suitable for: Compared to gzip/zstd: Compared to LZ4: Training: O(n × k) where n is input size, k is number of rounds (5) Encoding: O(m) where m is output size, ~200-500 MB/s Decoding: O(m) where m is output size, ~1-2 GB/s (table lookup) The table is ~2-8KB and encodes/decodes millions of strings per second.
Registry
-
Source
- Documentation
- JSON
- codemeta.json
purl: pkg:golang/github.com/axiomhq/fsst
License: MIT
Latest release: about 2 months ago
First release: about 2 months ago
Namespace: github.com/axiomhq
Stars: 12 on GitHub
Forks: 0 on GitHub
See more repository details: repos.ecosyste.ms
Last synced: 25 days ago