Cleaning For Performance - Is It Real - Check Some Tests!

In the world of computing you'll find many people out their claiming that cleaning up your systems temp files will improve performance. Don't believe the hype because there is more to performance then simply cleaning, there are to many other factors that come into play when it deals with system performance, and they both work at the hardware and software level.

Next time you are tempted to purchase or use something, go out there and review it and then even go a step further and TEST it!

Like Mama always said the Proof is in the Pudding...

So with this in mind I've run a test to prove that you won't typically see any differences after cleaning temp files, or even after defragging the drive. (In many cases you might see a difference after defragging the drive, but then again you might not, more times then often it's not noticed.)

First I ran CCleaner to show the system temp files are cleaned out:

http://i248.photobucket.com/albums/gg190/Alohapcs/ccleaner.jpg

Next Disk Defragmenter is run showing some fragmentation:

http://i248.photobucket.com/albums/gg190/Alohapcs/Disk_Defragmenter_1.jpg

Next Disk Defragmenter is run to clean up fragmentation:

http://i248.photobucket.com/albums/gg190/Alohapcs/Disk_Defragmenter_2.jpg

Next the system is rebooted...

After the reboot HD Tuner Pro is run to do a read test on the hard drive:
(REMEMEBER THIS DRIVE IS NOW CLEANED AND DEFRAGGED!)

http://i248.photobucket.com/albums/gg190/Alohapcs/HD_Tune_clean.jpg

Now it's time to fill the drive up, like simulating the 'Temp' files, and frag the drive and even go one step further which will reduce performance, FILL the drive up almost to max capacaity, beyond the recommended 10%-15% free space.

Here the files are placed in a users Local Settings Temp directory:

http://i248.photobucket.com/albums/gg190/Alohapcs/Temp_files.jpg

Disk Deframenter is checked, this time showing the hard drive needs defragging:

http://i248.photobucket.com/albums/gg190/Alohapcs/Disk_Defragmenter_1-1.jpg

Disk Defragmenter shows in this picture how fragmented the drive is, also NOTE that there is only 226MB of 'Free Space':

http://i248.photobucket.com/albums/gg190/Alohapcs/Disk_Defragmenter_2-1.jpg

HD Tune is now run on the fragged out drive that is also filled close to capacity:

http://i248.photobucket.com/albums/gg190/Alohapcs/HD_Tune_fragged.jpg


TEST RESULTS SIDE BY SIDE:

Clean defragged drive specs given by HD Tune:

Transfer Rate Min: (15.2 MB/Sec) - Max. (41.4 MB/Sec) Avg. (35.8 MB/Sec)
Access Time: 14.8ms
Burst Rate: 60.0 MB/Sec
CPU Usage: 2.6%


Fragmented drive specs given by HD Tune:

Transfer Rate Min: (6.4 MB/Sec) - Max. (41.4 MB/Sec) Avg. (35.6 MB/Sec)
Access Time: 14.7ms
Burst Rate: 60.0 MB/Sec
CPU Usage: 3.2%


CONCLUSION:

With the fragmented drive you'll notice there is a drop with the minimum transfer rate, BUT OVERALL the averages between the the clean and fragged drive is only .2 MB/Sec, that is POINT 2...

Access times only show a .1 difference.

FINAL WORDS:

The worst case scenario is when you fill your drive to capacity. Most Techs agree 10%-15% is a good rule of thumb...

I intentionally filled the drive to almost max capacity beyond the recommendation of free space to show that even with a bad fragmented drive and filled up, that the transfer and access times didn't degrade performance by much, and not enough for anyone to even notice.

TAKE THE TEST YOURSELF:

1. Run your favorite cleaner
2. Defrag the hard drive
3. Reboot
4. After reboot run either HD Tune or another favorite benchmark tool.
5. Fill up your drive with files. (DVDs or Avi will do nice and quick)
6. Check Disk Defragmenter and notice the fragmentation
7. Reboot
8. Run HD Tune, or your favorite tool, and now compare results of before and after...

Have fun testing, and REMEMBER there is no substitute for benchmarks, especially when someone tells you that a particular program improves performance, get out your trusty benchmark tool and test it, or look for others that have done the same...

P.S. Your mileage my vary, but this typically only happens in extreme cases of neglect. As a Computer Tech that has spent many years cleaning computers for clients, I've worked on systems that were never cleaned, ever in the life of the computer, and we are talking like anywhere from 2-3 years in temporary files and a hard drive that never got defragged, and when I was done cleaning these systems performance wasn't seen for simply cleaning. In other words, it takes more then just cleaning out temp files to see a performance increase...

 

Hi DasFox,

although I agree that cleaning and defragmenting the hard drive in general auser will not see much performance improovment, it seems to me that your test is a bit flawed and does not reflect the real fragmentation that can take place in a year.

For explaining better:
1. When you perform a copy/paste of a file windows already knows the files size and will place them in one contigues piece if it has enough space available.
2. by defragmenting first the drive you already have optimised the OSes files/directories. The directory fragmentation has more impact in generall performance than the file fragmentation.
3. It is not needed to add files for "adding" fragmentation on the drive; just select you program files directory and use the ntfs compression. After that uncompress it again and then reboot your system. You will be surprised with the impact you will have when you will lanch the various applications.

Panagiotis

 

Fragmentation over a period of time is not different. Point in case, it doesn't matter where fragmentation occurs in one area, or spread all over the drive over time, it's all the same and doesn't change anything.

All drives have imperfections, so even though they are not reporting bad sectors and clusters there are bad sections, and when Windows places data on a drive there will be bad sections marked and this then breaks up files into non contiguous pieces.

Not sure what you mean about the compression other then causing fragmentation. Compression will cause applcaitions to respond slower, nothing to do with fragmentation.

Forgot to mention compressing and uncompressing directories and then rebooting and running the applications will not cause for any slow downs.

Something I want to stress here, no matter how the fragmentation and file build occurs, if someone notices a big difference, then there is more likely an issue with hardware, not enough ram on the system, buggy software, to many apps competing in RAM, and running in the background, also hard drive access happening constantly because of Swap file access...

Good hardware and a tweaked system rarely see differences...

 

I disagree on this one. If fragmentation occures in the system exes, dlls, etc. you will see an impact in the totall performance of your OS. If it happens in archives, videos, etc. you will not notice it unless you try to extract the archive or you perform seeking operations on the video.

Fragmentation does not impact much the generall performance of the disk itself, but it impacts the initial lanch of an application untill the necesarry files are loaded in the memory.

I have seen drives with bad clusters but I do not understand why do you say that all drives have imperfections?

If you have little free space and you perform an ntfs compression/uncompression your files become fragmented. I did not said to test the compressed files I said to compress them, after that uncompress them again and then lanch an app. (but it will function only if you do not have plenty off free space on the drive or if the files are large enough).

Better yet use raxco scrambling toolsto cause fragmentation.

 

All I've been trying to say is that real performance seen and noticed happens in many ways across many different levels with hardware and software, and that by simply cleaning out your temp files and defragging your drive depending on your setup, etc., you are most likely not going to notice anything.

Now the reason there are a lot of people that claim to see this is because they don't have either good hardware or a properly set up system, and run to much junk software that is also sucking up the system.

Anyone that just goes out and buys a Dell computer as an example, fires it up and says this is how good computing works, doesn't understand, and with a DEFAULT system like this as I will call it, you are more likely going to run into problems and see them.

 

DasFox on my first reply I said that I agree with your statement that cleaning and defragmenting will not increase performance on a general use system.

On the other hand if you run a server or have a pc dedicated to multimedia editing, defragmentation means alot.

For making an example, I defragg my parents pc once every 3-4 months and the performance gained is 0%. On the other hand I defrag my main pc (system partition) every 2 weeks and the performance boost in seeking times is 10-15%. It all depends on how and for what purpose you use a system.

Panagiotis

 

Let's have a look at other factors in cleaning out temp files, especially gigabyte amounts.

So tell me, how much longer does a defrag or an AV scan take and also the making and size of an image.

If a defrag or AV scan takes 5 minutes with temp files gone instead of an hour I would say that's a huge speed gain.

Images are a quarter the size and way quicker to make/restore being another speed gain with way less wear and tear.

 

Yes of course if you're doing something hard drive intensive you need to keep things clean and fragged for better performance.

 

I'm not sure I understand what you are saying.

Defrag and AV scans can take longer simply because you have more data and have nothing to do with how badly a drive is fragmented.

Sorry the subject of the post was about cleaning out temp files and defragging to see performance gains, not sure what images have to do with this either...

 

In my years of use I would say there is no doubt in my mind that with sufficient free space, fragmentation can noticably affect to a small degree bootup and some key system usages, like when the dll's are all over the place. Especially after many updates.

The most benefit I see from defragging is directory walking. I can definately tell a perceptible difference in that above all else. Other than that I don't percieve any other improvements from defragging.

I can also say that when working on peeps computers, a defrag will almost always give them a perceptible speed up. At least the ones who don't even know it existed.

As for temp files, I don't worry about them with ample free space. So I pretty much agree that the performance gains are minimal indeed. An advanced user probably spends more time exploring directories or creating scripts etc to do that, so it is those who would see the directory difference.

How about right click context menu speed? Have you noticed if a defrag helps that? I have. Again, it is relative, but I can tell the difference. On that note cleaning up application registry entries that use context menus is even more effective, as well as just plain deleting some that slow things down.

Sul.

 

Do you mean by that to use only 10% of the available space, or is there no significant effect on responsiveness unless you use over 90% of the space?

 

No I'm saying to leave 10%...

There seems to be a lot reasons for needing some extra free space, the simplest reasons would be that some programs need Temp space to run...

 

The extent of fragmentation can determine the time for defragmentation as well as AV scans. Other factors remaining the same, a drive with less/no fragmentation ought to run an AV scan faster than one which is fragmented.
http://files.diskeeper.com/pdf/Antivirus_and_Defragmentation_White_Paper_channel.pdf

BTW, does the HD benchmarking software mentioned in the OP do raw reads sequentially from the start to the end of the disk? If so, then (de)fragmentation related effects may not be really reflected by the benchmark I think.

If the free space drops below ~12.5%, it might result in population of the MFT reserved zone (~12.5% of the volume) with files, possibly leading to fragmentation of the MFT.


Best regards

 

HD Tune tests hardware, not system, you should use some software benchmark. Cleaning PC will not upgrade hardware, though the benefit might be similar. The point is, that it will speed up aplications. I had see quite a difference on my old PC, even on my new one. Eg I installed a game, when I ran it, it lagged a lot, then I defraged HDD and ran it again and guess what, almost no lagging at all. And there many other things like to improve performance, eg unpacking game files will noticable increases its performance, because it does not have to unpack it in realtime, like Quake 4, so the files are access via RAM, it is like disabling pagefile. Cleaning and defragmenting registry can speeds up system start by a few sec, 40 MB registry size is better than 50 MB, is it not? Of course deleting any files will do nothing for performance, but cleaning temp or so can solve many problems and so fort increase aplications response time. CCleaner helped me on quite a few occasions. But I agree, that on these days PC, even a 10% perfomance boost does not make a difference for a common user, but it might do for a gamer, who is willing to pay even for 1 FPS more. Gamers pay for about 10 times more expensive DDR3 or fast DDR2, though its performance increase vs common DDR2 is about 1-5%. Anyway, paying for those "miracle" aplications, which clean up PC and claim to get even as much as 20% is a lie. People would benefit more by buying RAM. But I definitelly agree, that using cleaning aplications is dangerous, even simple log files cleaning can result in problems, registry cleaning can be deadly, literally, it can result in system reinstal. So for a common user would do better not to try to clean PC, but it does not hurt to defrag, but Windows does it by default, right?!

 

I think it has been shown in many tests that defraggers, registry cleaners and all the hundreds of system 'optimizers' are pretty much snake oil, and at best will offer minimal performance gains. 99% of those perf gains can be had by using builtin Windows tools, and 100% by freeware. Vista will do a lot of it automatically as well.

I esp like the extravagant full page color ads by Diskeeper which make it seem that without their product you are basically doomed

 

I think I got a measured 10% decrease in performance from ONLY fragmented files, when I ran my tests (I recreated 1 month of REAL disk usage on a Office desktop).

The performance hit of all the files regularly used was below that of the deviation of my test results (average of 16 runs, disguarding fastest and slowest results).
In other words, in one month of 9 to 5 daily usage my work desktop machine suffered a small amount (1% or less) fragmentation , but the performance hit was not measureable.

Part of this is XP's built in prefetch and automatic defragmentation of the prefetch, but also performance loss is mitigated by the random accesss patterns of standard desktop usage and extensive caching of the filesystem.

Cleaning temp files can make a huge difference if your drive is close to full (last 10 to 20% depending on size of partition) OR you have massive freespace fragmentation, as you get more/larger areas of contigous freespace.

Plus cleaning temp files is just a good house keeping habbit to get into, as though most programs clean up temp files well, if something crashes, the temp files get left behind.

One of my criticms of on-the-fly / continous defragmentation is that temp files are amongst those that get fragmented the most, but as they are temp, why bother defragging them ?
Same applies for recycle bin files, why defrag files that have a good chance of being deleted.
A smart defrag tool will treat these kind of files differently.

On a side note, though I have no tests to prove, I did notice on my development machine, which suffers far more fragmentation that the rate of fragmentation gradually decreases (after a defragmentation) after some time and I suspect that this is related to a very similar set of temp files being created/deleted each day (I tend to only do a few set tasks on my development box).

 

Hi DasFox,

I remember seeing people run HDTune (free version) on new drives w/o any data, in order to compare the read/write performances for various drives & drive models.

If that's so, then doesn't that put into question if actual files on the harddrive (and whether they're fragmented or not) affect HDTune's results?

Thanks.

 

... and do not forget - every week - making manual cleaning for performance on : C:\System Volume Information\_restore...\ blue restore points - all except the last!
Very very good for performance!
Your Pro-clean

 

Or not using system restore at all! Best for performance.
Mrk

 

Yes Mrkvonic, I prefer search of the mushrooms, with nice Autun ... Best for MY performance ? ...

Thanks, PROROOTECT

 

Hello Mrkvonic,

If I delete System Restore, I would download something else that will affect my Use Memory ... Yes? ...

Thanks, PROROOTECT

 

Mrk,
If I get rid of System Restore what would I replace it with on an XP Pro system?
Hugger

 

Hi,

A good imaging solution. System restore requires Windows access, works partially both in terms of success and areas it covers. Imaging solution = entire partition.

prorootect, I'm not sure I understand your question ... You mean your memory will be taken up by another process - or the hard disk space?

Mrk

 

Use Memory will increase with other software, if I want to replace System Restore ...

Thanks ...

 

Not sure...

Oh by the way the TOPIC of the post was ----> 'Cleaning For Performance', and we've gotten off topic here with other areas of performance factors.

I meant the post to show that just cleaning out files isn't going to improve performance, also even defragging a drive might not increase any too, unless there is some really bad fragmentation going on, or your hard drive is almost full to max capacity. Like I said before I've dealt with customers computers that never saw a defrag for 3 years and when I completed it I didn't notice any speed differences at all, and 3 years without defragging, I'd say that was a pretty extreme case. All I can say is it depends on what software you use, your hardware and other things going on with the OS that affect this performance, because I know with a well tuned system, tweaked up, running very little resources, and running good fast hardware, these things attributed to better system performance then just simply seeing it occur after a defrag....

On the subject of System restore, it's JUNK!!! I've seen it kill an OS and then the system had to be reinstalled!!!

If someone needs a EASY and CHEAP restore solution, then get something like Image For Windows:

http://www.terabyteunlimited.com/image-for-windows.htm