Benchmark Results

This page is a viewer for the generated benchmark snapshot committed with the docs.

Local numbers win

The interactive chart reads docs/.vitepress/data/benchmarks.json, which is rewritten by make bench. Treat that generated JSON file and the chart below as the source of truth for the current repository snapshot.

Interactive Performance Chart

Performance Comparison

Dataset:Text-like (10 MiB)Repetitive (10 MiB)Small dictionary-like sample

Metric:Encode SpeedDecode SpeedCompression Ratio

C++

Go

Rust

Huffman

16

21

11

Arithmetic

6.7

5.6

7.9

Range

RLE

Encode Speed (MiB/s)

Generated Data Model

JSON location	Fields	Meaning
top level	generated	Date of the benchmark snapshot used by the docs
each results[] row	algorithm, language, dataset	Benchmark coordinate shown in the chart
each results[] row	encodeTime, decodeTime	Wall-clock milliseconds
each results[] row	encodeSpeed, decodeSpeed	Throughput in MiB/s
each results[] row	compressionRatio, throughput	Size ratio plus the coarse throughput label used by the UI

Current Dataset Coverage

Dataset key	Chart label	Why it appears
textlike_10MiB	Text-like (10 MiB)	Main comparison input for Huffman and Arithmetic
repetitive_10MiB	Repetitive (10 MiB)	High-repeat input where RLE is meaningful
small_dictionary_like	Small dictionary-like sample	Reduced Range workload kept below the known large-file decode limitation

Because the chart reads the generated JSON directly, local reruns replace the snapshot without any manual table editing on this page.

Reading the Snapshot Safely

• Compare implementations on the same dataset before drawing language conclusions.
• Range results are intentionally shown on a smaller input; do not read them as apples-to-apples throughput versus the 10 MiB datasets.
• RLE is most meaningful on repetitive inputs and may expand other data.

Refresh the Source of Truth

make bench
npm run docs:build

• make bench rewrites reports/ and docs/.vitepress/data/benchmarks.json
• npm run docs:build verifies that the docs portal renders the refreshed dataset
• If you commit benchmark updates, commit the generated JSON file rather than hand-editing narrative numbers on this page

See Also

• How to Run Benchmarks — Detailed instructions
• Cross-Language Testing — Verification methodology
• Algorithm Guide — Comparison and selection guide