The big XP defragmenter test

[Darlis]

I've started a project to finally answer the question "Which defragmenter is the best?".

I've compared 6 defragmenters so far and that's just to top of the iceberg. There is still a lot of work to do. Everyone is welcome to discuss and criticize the defragmenters, results, test setup and methodology.

>>> The big XP defragmenter test <<<

[smurphdude]

Interesting results and the positive impact on boot time and application launch from mydefrag certainly tie in with my experience. I've had been trying to benchmark impact of frequency of defrag on my system (trying to decide if weekly or daily is best). Results were not conclusive, but it appears (not unexpectedly) that boot time improvements are maintained after defrag for at least a week, but application launch times start to degrade within a day or two of last defrag. This is on a single disk XP laptop that is used for a lot of video/TV so very large files are regularly created and deleted.

I'll be interested in further results from your tests.

[Darlis]

Results were not conclusive, but it appears (not unexpectedly) that boot time improvements are maintained after defrag for at least a week, but application launch times start to degrade within a day or two of last defrag.

It think that it highly depends on the program. If no program files get changed or moved, it should stay the same. Creating and deleting movies does usually not move program files. But, for example, updating the data from TV-Browser would definitely slow it down, since these files are read during startup.

I'll be interested in further results from your tests.

If you are interested in a particular defragmenter, you can vote for it here. 

[jeroen]

My congratulations on your project. It is quite extensive with lot's of different measurements, unlike some other comparisons I have seen, and I can see you have put in a lot of time. Very good, and I am not just saying that because MyDefrag comes out on top! There are of course lot's of other things that could be tested, but I think your comparison already covers a lot and gives a good idea to readers which program is the best. Perhaps I should mention that even better optimization results can be achieved than with the generic monthly MyDefrag script, by using the scripting abilities of MyDefrag. It allows for fine-tuning a disk for whatever use, not possible with any other defragmenter/optimizer as far as I know.

[Darlis]

Perhaps I should mention that even better optimization results can be achieved than with the generic monthly MyDefrag script, by using the scripting abilities of MyDefrag. It allows for fine-tuning a disk for whatever use, not possible with any other defragmenter/optimizer as far as I know.

An I'm going to make use of that, too.
I've planned to test my OptimizedSystemDisk scripts too, to see if these can reduce the optimization time and data shuffling without reducing the performance.

[Kasuha]

I'd like to know how Monthly optimization results differ from using just Daily or Weekly optimizations in MyDefrag.

[Darlis]

I haven't tested these (yet). I think I'll test the Weekly too, not the Daily. The Daily is not that thorough and intended mostly for maintenance.
I would guess that the results will look like those from JKDefrag (-a 3), maybe a tad better since MyDefrag can consolidate the prefetch files.

[smurphdude]

It think that it highly depends on the program. If no program files get changed or moved, it should stay the same. Creating and deleting movies does usually not move program files. But, for example, updating the data from TV-Browser would definitely slow it down, since these files are read during startup.

I think the initial apparent degradation is to do with gaps near the beginning of the disc filling up during general usage so that any temp files and the like that an application writes on launch end up being written to slower parts of the disc. So for me the main advantage of a daily defrag over a weekly is gap maintenance.

[Kasuha]

I haven't tested these (yet). I think I'll test the Weekly too, not the Daily. The Daily is not that thorough and intended mostly for maintenance.

There's minimum difference between daily and weekly scripts. Using FastFill() or using FastFill(WithShuffling) is only going to move files closer to the start of the disk by negligible amount... maybe except the spacehogs zone.
However, I recommend to run the script twice during the test to work around fastfill's end-of-zone problems. If there are none the script will not move any files in the second run.

[Darlis]

The difference between the Daily and Weekly is that the Daily will leave more gaps. I think the performance will be almost equal but the Daily is likely to produce more fragmented files. That's why I'm going for the Weekly first. I think I'll test the Daily, too at a later time.
I'm also pondering if I should do a separate benchmark page just for MyDefrag. (Curse you Jeroen for making that scripting language  )

However, I recommend to run the script twice during the test to work around fastfill's end-of-zone problems. If there are none the script will not move any files in the second run.

Sorry, but I'm not going to run a defragmenter twice in one test. If MyDefrag (or the script) can't handle such situations and thus deliver bad results in the tests then that's a problem of the program (or the script). I'm treating every defragmenter equal, MyDefrag is not getting bonus points.

[Kasuha]

Sorry, but I'm not going to run a defragmenter twice in one test. If MyDefrag (or the script) can't handle such situations and thus deliver bad results in the tests then that's a problem of the program (or the script). I'm treating every defragmenter equal, MyDefrag is not getting bonus points.

As you noted before, Daily and Weekly scripts are intended for maintenance and they assume that the disk is at least slightly pre-optimized. If you run them once on a completely unsorted disk, many files may end up out of their zones. Usually just one additional run gets rid of that problem, which would get more or less sorted out for normal user the next time they run the script.

But do whatever you think is best, I believe the performance will not suffer much on that.

[Darlis]

As you noted before, Daily and Weekly scripts are intended for maintenance and they assume that the disk is at least slightly pre-optimized. If you run them once on a completely unsorted disk, many files may end up out of their zones.

I said that the Daily is for maintenance, not the Weekly. The weekly moves all files to it's zones because of FastFill(WithShuffling).

[Kasuha]

I said that the Daily is for maintenance, not the Weekly.

If we consider result of Monthly script the desired layout then Weekly script is only maintenance, too.

The weekly moves all files to it's zones because of FastFill(WithShuffling).

That's true only on a disk that is not used for anything else while optimizing. I believe that's good enough assumption for your test, though.

I am sure the performance after Daily/Weekly will be worse than after Monthly. I'm just curious how big will the difference be.

[BloodySword]

Why don't you test on Windows 7?
This year, even the support of XP SP3 will end.

[Darlis]

I'm going to do that too.
I chose XP first because there is more to optimize. I would guess that the results for booting and application start on 7 won't differ that much from an unoptimized system (with enabled SuperFetch). Plus, this is more like a test run. I'll probably add some more tests or change the existing ones a bit. For example, I'm not happy with the antivirus scan test and I'll probably replace this with another one.

And the support for XP SP3 ends in 2014, not 2011: http://support.microsoft.com/lifecycle/?LN=en-gb&x=16&y=12&C2=1173