Conversion Benchmarks (vs Calibre)
Empirically measured conversion benchmark converting sample eBook corpora to EPUB 3 on AMD Ryzen 9 / PCIe 4.0 NVMe:
| Input Format | Sample Book | 🚀 ebook-rs (v0.16.4) |
🐍 Calibre ebook-convert |
Speedup | Image Assets Extracted | Chapter Sections | Output Parity |
|---|---|---|---|---|---|---|---|
| MOBI → EPUB | Alice in Wonderland (MOBI) | 0.34s ⚡ | 2.23s | 6.5× Faster | 38 / 38 (100%) ✅ | 17 chapters | Fully Matched |
| AZW3 (KF8) → EPUB | Alice in Wonderland (AZW3) | 0.22s ⚡ | 2.03s | 9.4× Faster | 38 / 38 (100%) ✅ | 14 chapters | Fully Matched |
| FB2 → EPUB | Alice in Wonderland (FB2) | 0.18s ⚡ | 1.66s | 9.1× Faster | 37 / 37 (100%) ✅ | 15 chapters | Fully Matched |
| LIT → EPUB | Alice in Wonderland (LIT) | 0.32s ⚡ | 1.94s | 6.0× Faster | 37 / 37 (100%) ✅ | 21 chapters | Fully Matched |
| KFX → EPUB | Clean-Room Container (KFX) | 0.29s ⚡ | 3.84s (Plugin Req.) | 10.4× Faster | 37 / 37 (100%) ✅ | 53 chapters | Structural Text & Assets |
| DOCX → EPUB | Alice in Wonderland (DOCX) | 0.15s ⚡ | 2.85s | 19.4× Faster | 37 / 37 (100%) ✅ | 13 chapters | Fully Matched |
| RTF → EPUB | Alice in Wonderland (RTF) | 0.85s ⚡ | 2.34s | 2.8× Faster | 36 / 36 (100%) ✅ | 15 chapters | Fully Matched |
Analysis
Section titled “Analysis”- DOCX Conversion:
ebook-rsis 19.4× faster than Calibre due to zero-copy XML stream parsing. - KFX Conversion: Calibre requires third-party plugins while
ebook-rssupports KFX out of the box. - Memory Footprint:
ebook-rsmaintains peak memory below 18MB during conversion, compared to Calibre’s Python runtime peak exceeding 120MB.\n