The latest articles on DEV Community by Sihwan Lee (@leeshiny7). https://dev.to/leeshiny7 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4232760%2F2c1ad450-3fd7-4406-9574-c3b78a9f8232.png https://dev.to/leeshiny7 en Sihwan Lee Sat, 07 Nov 2026 20:23:59 +0000 https://dev.to/leeshiny7/i-rewrote-7-zip-in-rust-heres-what-i-learned-about-memory-2nj7 https://dev.to/leeshiny7/i-rewrote-7-zip-in-rust-heres-what-i-learned-about-memory-2nj7 <p><a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fq6f8kchhwo0txf4dlaqc.png" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fq6f8kchhwo0txf4dlaqc.png" alt=" " width="799" height="537"></a><br> 7-Zip is legendary for compression ratio and speed. It's also legendary for<br> eating RAM: compressing a 500 MB log file at 16 threads peaks at ~1.8 GB.</p> <p>I wondered: how much of that is the algorithm, and how much is the<br> implementation? So I wrote <code>r7z</code>, a Rust clone of the core <code>a</code>/<code>x</code> commands,<br> and measured.</p> <h2> The numbers </h2> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Metric</th> <th>r7z</th> <th>7z</th> </tr> </thead> <tbody> <tr> <td>Time (500 MB text, 1 thread)</td> <td>68.4 s</td> <td>19.4 s</td> </tr> <tr> <td>Peak RSS (same)</td> <td>85 MB</td> <td>376 MB</td> </tr> <tr> <td>Peak RSS (16 threads)</td> <td>85 MB</td> <td>1.8 GB</td> </tr> <tr> <td>Ratio</td> <td>0.02%</td> <td>0.02%</td> </tr> </tbody> </table></div> <h2> Takeaways </h2> <ol> <li> <strong>Ratios are essentially identical</strong> — the algorithm dominates.</li> <li> <strong>Memory is an implementation choice.</strong> 7z's solid-block arenas and per-thread buffers scale with thread count. r7z's bounded buffers don't.</li> <li> <strong>Speed is where C++ still wins</strong> — SIMD range coder, tuned match finder, and a custom work-stealing pool. Pure Rust can't match that yet without <code>target-cpu=native</code>, SIMD intrinsics, or FFI to liblzma.</li> </ol> <h2> What's next </h2> <ul> <li>FFI path via <code>xz2</code> for parity speed</li> <li>SIMD CRC-32 / match finder</li> <li>Real multi-threading (currently single-threaded encoder)</li> <li>Solid-block size knob</li> </ul> <p>Repo: <a href="https://github.com/WhAnci/r7z" rel="noopener noreferrer">https://github.com/WhAnci/r7z</a> (English/Korean/Chinese README)</p> rust performance compression cli