# Benchmark

An HTML-to-Markdown benchmark for htmd-cli, [Turndown.js](https://github.com/mixmark-io/turndown) and [Pandoc](https://github.com/jgm/pandoc).

### What does it do?

- Fetch 200 page links from [Wikipedia - Rust](<https://en.wikipedia.org/wiki/Rust_(programming_language)>)
- Fetch all 200 pages and save them as HTML files
- Bench using [hyperfine](https://github.com/sharkdp/hyperfine) and following command:
  ```
  hyperfine --warmup 3 --runs 5 \
  'cargo run --release -- ./bench/bench-pages -o ./bench/bench-out/htmd' \
  'bun bench/turndownCli.ts ./bench/bench-pages' \
  'bun bench/pandocBatchCli.ts ./bench/bench-pages'
  ```

# Environment

System: linux x64 #23~22.04.1-Ubuntu SMP Thu May  9 17:59:24 UTC 2024

CPUs: AMD EPYC 7763 64-Core Processor x 4

Memory: 15.6 GB

# Versions

Bun: 1.1.17

Hyperfine: 1.18.0

htmd-cli: 0.3.3

Turndown.js: 7.2.0

Pandoc: 3.2

# Inputs

File count: 200

Total size: 61.84 MB

# Results

```
Benchmark 1: cargo run --release -- ./bench/bench-pages -o ./bench/bench-out/htmd
  Time (mean ± σ):     899.9 ms ±   4.0 ms    [User: 3117.6 ms, System: 192.0 ms]
  Range (min … max):   893.0 ms … 902.5 ms    5 runs
 
Benchmark 2: bun bench/turndownCli.ts ./bench/bench-pages
  Time (mean ± σ):     11.616 s ±  0.340 s    [User: 13.380 s, System: 0.791 s]
  Range (min … max):   11.166 s … 12.111 s    5 runs
 
Benchmark 3: bun bench/pandocBatchCli.ts ./bench/bench-pages
  Time (mean ± σ):     42.064 s ±  0.614 s    [User: 145.672 s, System: 18.846 s]
  Range (min … max):   41.487 s … 43.056 s    5 runs
 
Summary
  cargo run --release -- ./bench/bench-pages -o ./bench/bench-out/htmd ran
   12.91 ± 0.38 times faster than bun bench/turndownCli.ts ./bench/bench-pages
   46.74 ± 0.71 times faster than bun bench/pandocBatchCli.ts ./bench/bench-pages

```

*Updated at Thu, 27 Jun 2024 06:39:38 GMT*

*Generated by [bench.ts](bench.ts)*