nmuidi Benchmarks

This is supporting documentation for this video.

Setup

To be able to run these benchmarks you will need:

• Hyperfine
• Python 3
• nmuidi
• A WSL 1 Ubuntu image (either called "Ubuntu_WSL1" or update the script accordingly)
• Busybox
• An empty directory called empty for robocopy to use

Once you have all of these installed you should be ready to run the benchmark.

Benchmark

The way this hyperfine command works is by setting a "prepare" step which runs the file_generator.py Python script which generates the sample data into a folder called files. This prepare step is run before every command is executed. Commands must be wrapped in quotes and separated by spaces. Each command is run twice as set by the --runs 2 option. If you want to increase the run count higher you should be warned that this will wear on your hard drive/SSD, probably not much but worth considering. In my testing the results are pretty consistent. I used a drive that is not my main Windows drive (i.e. not my C drive) which will reduce the amount of background noise that can affect your results.

hyperfine.exe -p "python ./file_generator.py" "nmuidi files" "busybox.exe rm -rf files" "rmdir /s /q files" "del /f /s /q files > nul && rmdir /s /q files > nul" "wsl -d Ubuntu_WSL1 -e rm -rf files" -i "robocopy empty files /MIR /mt:128 /log:log.txt" --export-markdown results-5MBfiles.md --runs 2

Results

You can find the latest results here

Contributions

If you can think of a better way to delete files on windows, open a pull request and I'd be happy to test it out and include it in the benchmarks.