How to use benchmarks to cut through marketing hype

It’s not your imagination: your Mac’s overall performance can slow down over time. Most often this happens because we gradually add more apps and background processes, have more and more documents and browser tabs open, and don’t restart very often. All these things take a cumulative toll on your Mac’s performance. Cutting back on the number of things you have open is therefore one of the easiest strategies for putting some zip back into your Mac. Adding RAM (if your Mac supports it), switching from a hard drive to an SSD, and keeping your software up to date are also effective quick fixes for performance problems.

But if I may rant for a moment, I want to call your attention to two oft-touted remedies for slow Mac performance that usually have so little effect as not to be worth the effort: freeing up disk space and defragmenting hard drives. That’s not to say these procedures are pointless or that they never help, but as with repairing permissions, their curative properties are greatly overestimated.

Fortunately, you can prove (or disprove) the effectiveness of such remedies using science! Benchmarking utilities can provide you with cold, hard, numerical facts—if you take the time to use them properly.

Space invaders

Lots of utilities can locate and delete duplicate or excessively large files, old caches and logs, unwanted apps (and their supporting files scattered all over the place), unused fonts, and countless other types of data that may be cluttering up your disk. I’m all for tidiness and saving disk space, and I appreciate the time and effort such apps save me.

But what bugs me about much of this software is the claim, repeated endlessly in ads and marketing copy, that deleting all this digital detritus will speed up your Mac tremendously. The implication is that there’s a direct correlation between performance and disk space used.

There’s a kernel of truth in this claim. The true part is that OS X needs some breathing room to store things like virtual memory swap files; temporary files used when installing software; RAM images created when you put your laptop to sleep; and scratch files for audio-, video-, and photo-editing apps. If you run out of breathing room—which happens only when your disk is quite close to being full—OS X will indeed slow down, sometimes to the point of being unusable. Free up enough space, and performance should return to normal. (The need for breathing room is as true for solid-state storage as it is for hard disks, although SSDs should exhibit less pronounced speed reductions as you approach maximum capacity.)

Exactly how much free space you need to prevent performance degradation depends on quite a few variables. As a rough rule of thumb, I recommend 4 GB plus the amount of physical RAM you have installed as a reasonable minimum. But notice that this figure is independent of the size of your disk. In other words, if you have a MacBook Pro with 16GB of RAM and a 1TB disk with 990GB occupied, you’re in the danger zone. But put the same files on a 2TB disk and you have loads of breathing room. In the first instance, pruning 100GB of unneeded files might have a miraculous effect on speed, while in the second, you probably won’t notice any improvement at all.

There’s also the matter of what you delete. If your Mac is running slowly because it has insufficient disk space for virtual memory swap files, then deleting a couple of big files might help a lot. But if it’s running slowly because a particular buggy app is out of control, then only deleting (or disabling) that app will help. If you let a utility uninstall dozens of apps, disable login items, and clear caches, that might help your speed problem—but not necessarily for the reason you think.

Fragments of truth

When your Mac writes a file to a hard disk, there may not be enough contiguous space to store the whole file as a single unit. Instead, your Mac stores a piece here, a piece there, and keeps a record of where all the pieces are so that they can be reassembled when you need to open the file. This all happens transparently and almost instantly. In addition, OS X automatically defragments smaller files (under 20MB) in the background.

But conventional wisdom has it that since fragmentation only increases with time, eventually disk access will slow down because the read/write head has to physically jump around so much to reach all the pieces of each file. And for that reason, several utilities can defragment your disk, rearranging all the pieces of each file so they can be read in a single pass. Defragmentation can be extremely time-consuming, and while it’s happening, your Mac will definitely be much slower than usual because of the constant heavy disk access. (As a side note, I should mention that SSDs don’t require defragmentation, and in fact, attempting to defragment an SSD can reduce its lifespan.)

But is defragmentation worth it? Again, it depends. All things being equal, the less free disk space you have, the greater the likelihood of fragmentation, and the greater its impact on your Mac’s performance. If you have a large, fast hard drive that’s nowhere close to being full, it will still have some fragmentation, but the real-world performance gains from defragmenting the drive will probably be trivial.

Put it to the test

If you encounter a process that purports to speed up your Mac (whether deleting files, defragmenting, or something else), you could try it and then make a subjective assessment as to whether it helped. But a much better approach is to arm yourself with facts. You can use a benchmarking utility to measure it before and after making a change and compare the numbers.

The two most popular benchmarking tools for Macs are Spiny Software’s Xbench (free) and Primate Labs’ Geekbench (free for basic 32-bit benchmarks, $10 for the standard version, or $100 for the Pro version). They’re simple to use—a single click will run a predefined suite of tests and give you an overall numeric score plus individual scores for various tests.

But before you jump in, remember that we’re trying to be scientific, so you must take steps to ensure that your measurements are valid. Here’s what I recommend:

• Eliminate any extraneous factors that might influence the results. Ideally, quit all apps, close all windows, and turn off any background processes (such as backup software) that might change your Mac’s resource usage during the test. (For good measure, I like to restart—without reopening any apps—right before running a benchmark.)

• Run the benchmarking software and record your scores.

• Change just one thing. This is the hard part! If you run the test, make lots of changes, and run it again, you won’t know which change was responsible for your new score. So delete some large files, or uninstall an app, or turn off a background process or whatever—but do nothing else. Then restart your Mac if whatever you changed involves software that runs automatically.

• Now rerun the benchmarking software and again record your scores. Small changes are to be expected for any of numerous random reasons and aren’t significant. If you see numbers go way up or way down, whatever you changed was most likely the cause.

Of course, the fact that a benchmark number goes up significantly doesn’t mean your Mac will necessarily feel faster or make you more productive. But if the numbers don’t move significantly, you’ll know whatever you changed doesn’t affect its performance, and you can save time and effort by not worrying about that thing in the future.