Benchmarks

Every number, measured.

Six datasets, eight tools and settings, every archive extracted and compared byte for byte with the original. Pick a dataset to see how Zipling's Arca engine does against Finder, 7-Zip, zstd and xz.

Show this dataset as a table
ToolSize% of originalCompressExtract

Summary

Against 7-Zip Ultra, set by set.

7-Zip's strongest setting is the toughest common competitor, and the engine Keka and BetterZip use. Size change is Zipling's archive against 7-Zip's; speed compares Zipling's default setting with 7-Zip Ultra.

DatasetSize at DefaultSize at MaxCompress (Default)Extract (Default)

Method

How we measured.

  • Apple M1 Pro (8 performance + 2 efficiency cores, 32 GB), macOS 27, warm file cache.
  • Wall-clock time for every tool, and every round trip compared with diff -r against the source. All were identical.
  • Versions: 7-Zip 26.02 (7z -mx9), zstd 1.5.7, XZ Utils 5.8.4, and macOS's ditto, which is what Finder's Compress uses.
  • Zipling and 7-Zip were measured together on an idle machine. Rows marked † come from an earlier session, so compare their times with care.
  • Back-to-back runs vary by about ±10%, so treat small differences as ties.
  • Public and system data only. No personal files were read.

Honesty

Where Zipling is still behind.

  • Max extraction of PDF and PNG folders is slower than 7-Zip (1.2 s against 0.8 s), for an archive 55% smaller. Re-creating a PNG that holds one huge image can't be split across cores.
  • Max compression of PDFs and source code is slower than 7-Zip. It is within 5% on the mixed folder.
  • The default setting on apps, code and generic data trades size for speed. It is 15–32% larger than 7-Zip Ultra there, but about 10× faster to compress. Max matches or beats 7-Zip on all of them.
  • Ultra helps only prose-heavy data and extracts about 2× slower, so it stays opt-in.

Full tables, raw data and the benchmark harness are in the repository.