Benchmark

Do memory cleaners help, do nothing, or make a Mac slower? We are testing it, and we will publish the method with the data so you can repeat it.

The method

This is the plan, and it is the same whether or not results are in yet.

• Macs. A MacBook Air with 8 GB first, then 16 GB. Fresh user account for each run, same macOS build across a comparison, plugged in, Spotlight indexing finished.
• Workload. Chrome with fixed tabs, an Electron app that leaks at a known rate, a call, an editor with a mid-size project, a build, and photos. Every two minutes, the script switches app and types.
• Four conditions. A: do nothing. B: empty the file cache every ten minutes, which is what a cleaner app's main button does. C: a popular cleaner app, using its clean button every ten minutes. D: Hedroom with its default rules, plus one relaunch that a person accepts when the leak is found.
• Repeats. Five runs of 45 minutes per condition, in random order, with a reboot between runs. We report the median and the spread.
• What is measured. Memory pressure, swap written, how long an app takes to come to the front, how long an app takes to reopen, build time, energy, and Hedroom's own CPU and memory.
• What we publish. The method, the hardware, the raw files, the scripts, and every result, including any that do not show what we expected. If the data disagrees with us, it goes up anyway.

Until then, this guide shows a test you can run yourself in Activity Monitor.