RayoMD vs Pandoc 2026 06 18

RayoMD vs Pandoc Local Export Comparison

Run date: 2026-06-18.

This is a local end-to-end PDF export comparison, not a universal benchmark. The goal is to show the cost difference between RayoMD's native lightweight subset and a full Pandoc-to-XeLaTeX PDF pipeline on simple Markdown documents.

Method

• Platform: Windows, Microsoft Windows NT 10.0.26100.0.
• CPU identifier: AMD64 Family 25 Model 33 Stepping 0, AuthenticAMD.
• RayoMD: rayomd 0.1.0, Windows release binary, 2,364,928 bytes.
• Pandoc: pandoc 3.9.0.1.
• PDF engine for Pandoc: MiKTeX-XeTeX 4.16 (MiKTeX 25.12).
• RayoMD command shape: rayomd.exe --export input.md output.pdf native modern normal.
• Pandoc command shape: pandoc input.md -o output.pdf --pdf-engine=xelatex --no-highlight -V mainfont=Arial -V monofont="Courier New".
• Each case uses one uncounted warm-up run per tool, then reports the median of 5 measured runs.
• Timings include process startup, Markdown parsing, PDF generation, and file writing.
• Inputs are synthetic, deterministic Markdown files with no remote images or network-dependent content.

Pandoc is a much more complete document system than RayoMD. These numbers should not be read as a claim that RayoMD replaces Pandoc. They show the expected advantage of a narrow native renderer for simple documents where the lightweight subset is enough.

Results

Case	Input	RayoMD median	Pandoc median	Ratio	RayoMD PDF	Pandoc PDF
large_table_500_rows	45,437 bytes	38.98 ms	7113.89 ms	182.5x	500,973	94,515
nested_lists_code	30,685 bytes	20.21 ms	6504.84 ms	321.9x	123,612	91,043
mixed_100kb	102,476 bytes	52.45 ms	10614.53 ms	202.4x	1,100,237	215,143

Reproduce

python tools/benchmark.py compare -- `
  --rayomd build\release-check-windows-escalated\rayomd.exe `
  --root benchmark-output\pandoc-comparison-windows `
  --runs 5

The script writes the generated corpus, raw outputs, JSON record, and Markdown summary under the ignored benchmark-output/ tree.