How to benchmark your storage using Stibium

As experience is building with my free utility Stibium for measuring read and write speeds of SSDs in and connected to Macs, here’s a basic tutorial on how to benchmark your Mac’s storage in just a few minutes. Here’s my Gold Standard protocol, which should return robust and meaningful results.

Stibium version 1.0b6 is available from here: stibium10b6

1. Prepare your Mac and its storage

Choose which disk you’re going to benchmark, and create on it, in a folder to which you have read and write permissions, a new folder to contain the test files. For example, if you want to test the internal SSD holding your Documents folder, create a new folder in that named 0StibiumTest or similar.

Then restart your Mac, and leave it a minute or two for everything startup to settle. Ensure that it’s not due to undertake any scheduled activity including making backups, and leave all apps other than Finder closed. When you’re ready, open Stibium and set it up so:


The important controls to set are the popup menu in the top row, which should be set to 0x41, No Cache next to it which should be ticked √, and the number in the repeats box below, which should be 10. Unless you want to see each individual result, leave the Verbose checkbox empty, unticked.

2. Run the write test

When you’re happy those are correct, click on the Write… button in the middle row, next to the words Series tests:. This will open a window in which you should select your test folder and then click Open.


Once you’ve clicked Open, leave your mouse/trackpad alone so that you don’t disturb your Mac again until text appears in the lower scrolling text view in Stibium, which contains the results.


The write box in the lower row of controls then displays the write speed for that test, and below that is a detailed report. Use the Save Report… command in the File menu to save that in a text file, but not inside your folder full of test files.

3. Run the read test

Now change the number in the repeats box from 10 to 1 ready for the read test, and quit Stibium. Restart your Mac, give it a minute or two to settle again, open Stibium, check that it’s set ready for the read test with the repeats set to 1. Then click on the Read… button in the middle row of controls (between Write… and Mixed… buttons).


In the window which then opens, select the folder containing the 160 test files which Stibium wrote previously and click Open to run the read test. Leave your mouse/trackpad alone so that you don’t disturb your Mac again until text appears in the lower scrolling text view in Stibium, which then contains the results.


Use the Save Report command again to save your read results. Note that Stibium automatically suggests a file name containing the date and time stamp of the tests, so you shouldn’t have to change it to avoid replacing the previous report. You’re now free to open up your apps and resume normal work again. Unless you want to perform any more read tests on the original 160 test files, you can now drag that folder to the Trash.

4. Check the results

In each of the result reports, look for the following:

  • Check whether it’s the read or write test first.
  • Check that the number of files (n) is 160 in each.
  • Look down through median transfer speeds given by file size (starting with 2000000 and ending with 2000000000). These may start rather low and rise by half way down the list. No median should be greatly different from its neighbours.
  • There are three different estimates of overall transfer speed: the most reliable is the Regressed rate, which is calculated from the raw time results to yield an overall speed; the median should also be fairly reliable, although usually less than that from regression. Least reliable or representative in this test is the Average.
  • Look for evidence of outliers. Sometimes individual tests run very slowly or worryingly fast. Next to the median is the range across all 160 measurements. Check that those extremes don’t look crazy. If they do, it may well be worth repeating the test(s) to see if they don’t reduce and provide more reliable results.

If your results are looking good, they should be representative of the real-world performance of that disk.

5. Further tests

Try running the same tests using an external disk. To do that, create the test file folder on that disk, and select it when running the write and read tests.

Try running the write and read tests without restarting between them. On many Macs, you can get away with doing this, but on some you’ll see very fast read speeds because macOS has cached some of the files written during the write test, so instead of reading it from disk, it just cheats and copies the cached files from memory.

Try using random file sizes instead of the standard ones. To do that, set the repeats box to 2 and the length to 80. Then tick the Random Sizes √ checkbox and run the Write… test as you did before. That will write 160 files of random size to your test folder. Once complete, set the repeats to 1 again and click on the Read… button in the middle row of controls. Size-grouped medians will then be a very long list, as the test files are of different sizes, and the transfer speeds might differ from those of the Gold Standard test detailed above.

Stibium’s extensive controls let you design and run tests as you wish: it’s the à la carte storage benchmarking app.