As a reference, the performance of dirhash is benchmarked against the shell command:

find path/to/folder -type f -print0 | sort -z | xargs -0 md5 | md5

(top answer for the SO-question: Linux: compute a single hash for a given folder & contents?)

Each test case contains 1 GiB of random data, split equally into 8, 1k or 32k files, in a flat or nested (binary tree of depth 8) structure.

For a fair comparison, the CLI version of dirhash was used (including startup time for loading of python modules etc.).

For full details/reproducibility see/run the run.py script for which the output is found in results.csv. These results were generated on a MacBook Pro (2018):

• 2,2 GHz Intel Core i7 (sysctl -n hw.physicalcpu hw.logicalcpu-> 6, 12)
• 16 GB 2400 MHz DDR4
• APPLE SSD AP0512M