Reproduce a result

Pick a test and follow its setup. The benchmarks use several different environments.

Choose the comparison

Each link includes the hardware, tool versions, file sizes and settings for that test. The other comparisons on All results have their own setups too. Similar hardware and conditions should give a useful comparison; exact times will vary, especially across networks and NFS services.

Run it yourself

The syq-bench repository contains a recipe and configuration templates for these tests, including how to build the measured syq 0.5.2 version.

From a checkout of the repository, copy the template for your chosen test to my-test.toml. Set the paths to your tools and empty scratch folders; for a remote test, also set the SSH destination. With Python 3.13 or later and uv installed, preview the commands, then run:

uv run syq-bench run my-test.toml --dry-run
uv run syq-bench run my-test.toml --out results/my-test.json

Use otherwise idle machines and match the filesystem, file sizes and cache settings described in the recipe. The templates reproduce the copy jobs and tool options; the recipe also describes the original run order and extra diagnostics.

Before sharing your results, remove any private hostnames or paths from the output.