Putting Registry-/system-cleanup apps to the test

By Fred Langa The most contentious software category has to be PC-system/Registry cleaners. Some users find them invaluable; other users consider them worse than useless. A series of controlled experiments puts these apps to the test and turns up some surprises. Many Windows programs are still sloppy about their uninstall process, leaving behind digital debris. In older versions of Windows, this situation was a known cause of trouble. Leftover, "orphaned" files waste space on your hard drive and create extra work for Windows when it performs indexing, searching, defragging, backups, or other file-related operations. Leftover Registry entries similarly inflate the Registry to no useful purpose. Erroneous Registry entries can create system instabilities and crashes, and a bloated Registry might needlessly slow down system startup and shutdown. Registry- and system-cleaning software is designed to correct these problems by finding and removing orphaned files, useless Registry entries, and other junk. The end result is supposed to be a leaner, cleaner, more stable system. There's plenty of anecdotal evidence to suggest that cleanup software can do what it claims. But most of this information derives from older versions of Windows, which were notorious for not cleaning up after and policing third-party software. Windows 7, on the other hand, has more self-protective features than any previous version of Windows. So, are Registry- and system-cleaning tools still worthwhile in a Windows 7 world? I decided to find out.

Designing the tests and the baseline system

First, let me be perfectly clear about this report: my tests were not designed to find the best Registry cleaner or even a comparative analysis of one cleanup tool versus another. My sole point was to see whether the general principles of system and Registry cleaning deliver quantifiable, measurable improvements to Windows 7. Do specialized cleanup tools really let you remove more junk than Windows 7's own built-in tools? Do system and Registry cleanups provide any real-life, practical benefits, such as faster boot times? To find out, I started with a known-good, plain-vanilla, up-todate, fully normal Windows 7 setup in an Oracle VirtualBox

(site) PC. (See Figure 1.) VirtualBox systems can be cloned easily, which meant I'd be able to use exactly the same system as a starting point for different cleaning tests.

Figure 1. The test system was a plain-vanilla, minimalist Windows 7 VirtualBox setup with very little extra software installed, as shown. Because this system was to be the starting point for all that followed, I documented several key variables. First, I timed how long the unmodified system took to boot. I used a stopwatch to measure the time from the moment I switched on the system to the initial appearance of the Windows sign-on password dialog box. I then paused the stopwatch and entered my password. On hitting Enter, I restarted the stopwatch and continued timing to the point of full boot; when the full, normal Windows desktop appeared and the system was stable and ready for use. I also timed how long the system took to turn off from the moment I clicked Shut down on a stable, idle system to all lights out. I ran the tests several times in succession, with a full power-off shutdown in between, and then I averaged the results to help smooth out any human timing errors or other random variables. These initial timing numbers would let me see whether my cleaning experiments would have any effect on startup and shutdown.

I measured the size of the test system's Registry by exporting its full contents (via Windows' Registry editor, Regedit) to a text file and noting the size of the file. I also used Windows Explorer to record the aggregate size of all the files on the hard drive. These Registry-size and file-size numbers would let me see the effects, if any, of various cleanup techniques and tools, regardless of what the tools themselves might report. (Many cleanup tools tend to overstate their own effectiveness.)

Looking for trouble: adding 20 popular downloads

With the baseline measurements established, my next step was to create a system with lots of orphaned files. That meant installing and uninstalling applications that might leave digital detritus. To choose software commonly used by real-world users, I consulted CNET's list of the "20 most popular Windows downloads." I installed all 20 apps on the test system as inexperienced Windows users might over time layering on the software without regard to consequence and accepting all default settings, including the offers of extra toolbars, download managers, and so on. As you'd expect, the system ended up a mess. (See Figure 2.) After loading the programs, Windows was much less responsive; several subsystems (Internet and sound, for example) stopped working entirely.

Figure 2. Installing 20 popular apps brought Windows to its knees. I measured the boot/shut-down times and file sizes on the bloated system. Incredibly, the time for a full boot went from under a minute (39 seconds) on the clean system to almost 10.5 minutes (629 seconds) for the bloated configuration. The Registry ballooned 170 percent from 99MB to 169MB. Table 1 shows the results.

Table 1: Adding 20 popular downloads to the test system created a bloated setup whose full boot time went from 39 to 629 seconds over 10 minutes!

Removing the bloat: Uninstalling the 20 apps

As planned, the test system was now desperately in need of cleanup. The first step to clean any bloated system is to simply uninstall unneeded software. I uninstalled each of the 20 programs in the normal way using the control panel's Uninstall a program applet. This removed much of the bloat as it should and also resolved whatever conflicts had caused the sound and

networking failures. My next round of timing and size measurements showed that, as happens all too often in the Windows world, uninstalling programs left behind various files and Registry settings it did not fully restore the test system to its initial condition. Table 2 shows the results. After standard uninstallation of the popular software, the test system did not regain all its initial startup and shutdown speed. Also, 1.4GB of orphaned files and 6MB of Registry entries were left behind.

Table 2: After I uninstalled the sample apps, the test system still contained leftover files and suffered from reduced performance. I was now ready to see what the cleanup tools could do.

Windows Disk Cleanup versus third-party cleaners

I had planned to run three different cleaning tests (I'll explain them in a moment), so I cloned three, identical copies of my test system. These cloned systems would give each of my clean-up tests precisely the same starting point. On the first cloned setup, I downloaded and ran the standard edition of Piriform's free CCleaner (site), an immensely popular and easy-to-use system- and Registry-cleaning tool. (See Figure 3.)

Figure 3. Piriform's CCleaner is designed for quick, easy, and routine removal of junk files and useless or broken Registry entries. I chose CCleaner because I'd recommended it many times in the past and wanted to see whether my recommendations still held. CCleaner also represents a class of tools primarily intended for frequent, routine cleanups rather than a deep, targeted spelunking of the system's internals. I ran CCleaner's file and Registry cleanup routines in their default settings, letting the software decide what to clean on the test system. I then rebooted the system and made new timing and file-size measurements. I'll discuss the results in a moment. On the second cloned setup, I downloaded and ran Macecraft's jv16 PowerTools 2011 (site; free, fully functional trial for 50 days, $30 thereafter). This software represents a heavierduty class of tools that can do routine cleaning but also offers much more power and configurability for expert users.

Figure 4. Macecraft's jv16 PowerTools 2011 includes features and functions aimed at advanced users. Using jv16 PowerTools' default settings, I ran its Registry Cleaner, Registry Compactor (a type of tool unavailable in CCleaner), and File Cleaner. I then rebooted the system and recorded times and file size. On the third, identical clone setup, I used Windows' built-in Disk Cleanup app, a system-cleanup tool that's been included in every version of Windows from Win98 onward. (You can enter cleanmgr into the Search programs and files box to access the tool in its most basic configuration.) Win7's Disk Cleanup is actually surprisingly powerful and complete; it's one of those unheralded tools that have been quietly improved with each iteration of Windows. I've always found Disk Cleanup to be safe and reliable, but Microsoft in an abundance of caution, perhaps has always made Disk Cleanup's most potent cleaning functions a little hard to get at. In fact, to use the tool to its best effect, you have to enter cleanmgr in a Command Prompt window or from a command line, and that's what I did. When run that way, Win7's Disk Cleanup is actually surprisingly powerful and complete; it's one of those unheralded tools that have been quietly improved with each iteration of Windows. Disk Cleanup's basic commands haven't changed in almost a decade, and long-time readers may recall the April 4, 2002, article, "Sageset unlocks CleanMgr's power." The how-to instructions in that story still work perfectly in Windows 7.

Figure 5. Windows 7's built-in Disk Cleanup can delete about 20 different kinds of junk files. I ran cleanmgr exactly as described in that article. When it was done, I rebooted the system and again noted the startup and shutdown times, the Registry size, and the overall disk space used just as I'd done with CCleaner and jv16 PowerTools. Table 3 shows the results of these tests.

Table 3: All three tested cleanup methods reduced bloat and helped improve system performance. These results make it clear that Windows 7 can indeed benefit from use of cleanup tools!

Bottom-line conclusions and caveats

The primary takeaway from these tests is that use of any cleanup tool even the free, built-in cleanmgr can help fight bloat and improve your system performance over what you get if you simply uninstall an application. Although I was surprised that no tool removed all the junk files and leftover Registry entries, they all even the lowly cleanmgr, if launched from a command line with all its cleanup options enabled reduced the startup and shutdown times to those of my original, clean system. The real surprise? jv16 PowerTools actually made the cleaned-up system slightly faster than the original, unmodified, baseline machine! It seems that use of a more advanced tool can yield greater benefits than using simpler, less-powerful tools. But while system cleanup clearly is worthwhile, I urge you not to get carried away, obsessing over a few seconds of speed or a bit of extra disk space. Reducing boot time from 629 seconds to 32 seconds (by uninstalling unneeded software) is truly worth pursuing; going from 32 seconds to 33 seconds (for cleanmgr) or 30 seconds (for jv16 PowerTools) is almost meaningless in real life. And although I encountered no problems from use of the cleaners in the above tests, it must be said that the more advanced, expert-level cleaning tools can royally mess up a system if they're used improperly or too aggressively. So don't risk destabilizing a solid system for a trivial gain. Stay within your own comfort and skill zone and always, always, always make a backup before using any cleaning tool. For me, the bottom line is this: I'll continue using and recommending lightweight tools (such as command-line cleanmgr and CCleaner) for routine cleanups and expert-level tools (such as jv16 PowerTools) when simpler software isn't enough. And now, if you'll excuse me, I'm off to clean up my personal-use PCs. I suggest you do likewise!

Filename: Using registry cleaners Directory: C:\Users\viejo\Desktop Template: C:\Users\viejo\AppData\Roaming\Microsoft\Templates\Normal.dotm Title: Subject: Author: viejo Keywords: Comments: Creation Date: 11/10/2011 8:12:00 AM Change Number: 1 Last Saved On: 11/10/2011 8:13:00 AM Last Saved By: viejo Total Editing Time: 14 Minutes Last Printed On: 11/10/2011 8:34:00 AM As of Last Complete Printing Number of Pages: 7 Number of Words: 1,904 (approx.) Number of Characters: 10,853 (approx.)