Windows XP Home SP3 booting slow, but runs fast once booted

Hi ArTech.

You are obviously a knowledgeable technician, as well as Ztruker and I. Nobody here is arguing the facts behind the technical details of what defrag does, what fragmentation is, and how it can increase latency or improve performance. The question here is about how much of this has any real impact on an average user's computer experience. Absolutely there are performance gains to be had by defragmenting, but under what circumstances are those gains even slightly noticeable?

Millions of people download defragmentation tools because they are led to believe that this is something they should do on a regular basis to keep their computers running optimally. This information stems from days gone by where hard drives were very little, and very slow, and file fragmentation was a very real problem. Big software companies would love to have you keep believing that so they can keep rolling in the dough with useless utilities like real-time defraggers that only serve to slow the computer down more. You can use the facts you provided to try and claim that defragging is necessary, but it does not, it only means that defragging is a popular, even if misinformed, idea. And, even if you use "free" utilities, there are, without a doubt, monetary goals in the picture. It is no different than convincing people they need to run registry cleaners.

You see, you can talk about hundreds of thousands of file fragmentations but really most of those files are hardly ever read from or written to. Only a relatively few files are regularly accessed where fragmentation might cause some significant impact on performance. Files like those that hold the registry, and the pagefile are not sequential files so there is very little impact by fragmentation. By this I mean Windows is randomly reading and writing to various sections of the file so the hard drive is always seeking out different locations. It would be no different than if the operating system made an attempt to read the file from start to finish and had to skip around to various fragmentations. Even if the files are defragged windows will never realize a significant improvement in performance because it doesn't read or write to those files in that way.

When Mark Russinovich say “Paging and Registry file fragmentation can be one of the leading causes of performance degradation related to file fragmentation in a system” that isn’t feeling for me.

Mark is definitely a large contributor of tools and information for the Windows operating system. He is well known and respected, I use his tools myself. However, his quote says specifically that paging and file fragmentation can be one of the leading causes of performance degradation related to file fragmentation in a system. Specifically he is saying IF there is any performance degradation due to file fragmentation the leading cause would probably be from those specific files being fragmented. This is obvious because those are the primary files that are continually read from and written to. This quote DOES NOT say file fragmentation can be one of the leading causes of performance degradation, period. Mark knows better than that, and I would tell you that such a respected person would be out of his mind to even suggest such a thing, so I'm glad he didn't.

In reality, the grand scheme of things so to say, there are many other LARGE factors to system performance issues. The number one reason woud be a lack of enough physical RAM, the second would be too many programs running at the same time or loading at startup. Every other performance issue except for serious hardware or program issues has a far smaller impact on overall system performance. File fragmentation, in my list, would be at the absolute bottom of the entire list. On a daily basis I make computers run like they are new again when they come to me running dog slow. I don't reformat the hard drives, I fix the issues causing the computer to run slow. And I speak honestly when those problems are NEVER file fragmentation. I have, on very rare occassion chosen to defrag a hard drive on an old computer that is several years old and has NEVER had the hard drive defragged. After I get done working my magic I will start the defrag. But, you know what, even under those extreme circumstances I, a skilled technician, notice absolutely NO gain in performance whatsoever above that of what my other techniques provided. An example would be a computer that has a 20 or 30 gigabyte hard drive running Windows XP, 256MB of RAM, a 700Mhz processor and hasn't been defragged in 7 or 8 years. I will say again, even under those supposedly extreme circumstances, there is absolutely NO noticeable improvement after defragging the hard drive. Yes, I know they are there, somewhere, but where I'll never know. Maybe if I got my stopwatch out I could find them.

An example would be a computer that has a 20 or 30 gigabyte hard drive running Windows XP, 256MB of RAM, a 700Mhz processor and hasn't been defragged in 7 or 8 years. I will say again, even under those supposedly extreme circumstances, there is absolutely NO noticeable improvement after defragging the hard drive. Yes, I know they are there, somewhere, but where I'll never know. Maybe if I got my stopwatch out I could find them.

My old laptop might serve as an example. Windows XP, 512 MB of RAM, 1.8 GHz Celeron, 80 GB disk with only 3.5 GB free. The largest DBX file (Outlook Express mail store) was around 2 GB and consisted of more than 120000 fragments. To initially load/open that folder, Outlook Express took around 15 seconds. After clearing out some space and defragging the drive for the first time in about 6 years, the initial load time of said folder within OE is down to a few seconds.

That's a good example Don't Spam Me. In this particular case, with a file that large on an older machine there would probably be measurable improvements. However, 15 seconds down to a few seconds still sounds a bit extreme for any defrag, I wonder what else was improved by "cleaning out some space?"

That's a good example Don't Spam Me. In this particular case, with a file that large on an older machine there would probably be measurable improvements. However, 15 seconds down to a few seconds still sounds a bit extreme for any defrag, I wonder what else was improved by "cleaning out some space?"

At the same time I also used Ccleaner to deactivate various startup entries and to clear out misc. junk (e.g., 1.x GB of ancient SpyBot S&D logs, ancient ZoneAlarm logs, etc.). As a result, there might also have been more free memory which in return would have helped OE to better handle creating an index of said large mail archive, etc. I guess it's hardly ever that people take the time to just change one thing and then sit down with a stop watch. And unless one would take that approach it is rather difficult to determine which one of the steps taken had the most impact. All in all, the various steps in combination proved beneficial, though.

Before I forget it, you might have noticed in the PCPitStop report that there's a rather low transfer rate reported for all of my external drives. They are all connected through one and the same powered Dynex USB2.0 7-port hub. Shouldn't I normally get a considerably higher throughput or is this the result of the additional cable length and the additional plugs involved? (If this should be asked in a separate/new thread, rather than in this one, please let me know).

Hello Appleoddity, Ztruker

Appleoddity

I believe in your story and respect your rich daily experience here on forum and on your business.

I am electrical engineer with less computer software/hardware experience than you. PC is only my tool.

I have my rules here on the forum. My recommendation have 3 levels:
1. by the book (Microsoft support, well known experts and books, manufacturer recommendation,…).
2. my experience (until now only XP)
3. methods and some third-party tools beyond native windows tools that I heard or read about and suppose can help, but never try by myself

I am trying to notice which level is my recommendation, so that user can easier decide what to do. I prefer level 1. The best results are with advices from level 1 + level 2. Levels 2 and 3 shouldn’t be opposite with level 1.

As you said defragging is a popular, millions of people download defragmentation tools, and our member asks and many will ask which defragmenters you could recommend.

Again, I respect your rich experience but your advice is subjective and has level 2.

Level 1 can be:

Microsoft Support: Maintenance on a computer running Windows XP

This article contains information for the basic maintenance of computers running Windows XP. In this article, you will learn how to perform the following tasks:
… Defragmenting the computer hard drive to improve performance

or

Contig v1.55 By Mark Russinovich, published: September 30, 2008

There are a number of NT disk defraggers on the market, including Winternals Defrag Manager. These tools are useful for performing a general defragmentation of disks, but while most files are defragmented on drives processed by these utilities, some files may not be. In addition, it is difficult to ensure that particular files that are frequently used are defragmented - they may remain fragmented for reasons that are specific to the defragmentation algorithms used by the defragging product that has been applied. Finally, even if all files have been defragmented, subsequent changes to critical files could cause them to become fragmented. Only by running an entire defrag operation can one hope that they might be defragmented again.
Contig is a single-file defragmenter that attempts to make files contiguous on disk. Its perfect for quickly optimizing files that are continuously becoming fragmented, or that you want to ensure are in as few fragments as possible.

Instead of conclusion - the questions: Why is Mark so worried if some files remain fragmented? Why he made and promotes Contig?

Hi arTech. I really don't want to drag this thread out to the point where there is contention between members. I do respect your desire to follow the advice of highly respected developers and authors. Again, there is no question that defragging has its purpose, and yes it absolutey can improve performance. In this profession I have learned that many times personal experience is more valuable than that of what is published in books even by respected authors. Mainly because they have the responsibility to present theory for the purpose of understanding. But, they are unable to take into consideration the hundreds if not thousands of variables that come into play in the real world. It just wouldn't fit in a book. This is why a college education with no real world experience is much less desired, in my opinion, than someone who has been working on computers for 20 years and hasn't had a single college course. I'm not arguing with anything that you have said about defrag or what Mark says. I am simply stating that under most circumstances, defragging is highly overrated. In my real world experience I have found that the effect of defragging a hard drive makes such a little improvement in performance, if any, that it is not worth me wasting my time doing it. I've given it a shot, because I too once thought that defragging a hard drive really did something. I remember those good 'ol days when we had to have special third party tools to defrag hard drives with DOS on them. I remember how annoying it was to defrag a hard drive with Windows 95 because it would constantly start over from one thing or another accessing the drive. I used to waste a ton of time defragging hard drives. Back then I would say it was probably a much needed maintenance task. But now a days there is just very little to be gained by defragging as compared to the much greater impact that other variables have on system performance. Can defrag improve performance? Yes. But, it certainly shouldn't be considered THE deciding factor in computer performance, because you can defrag to your hearts desire and if you ignore several other factors your computer will run very poorly. On the contrary, you can have a computer that runs absolutely great and never defrag whatsoever. As far as pagefile and registry fragmentation goes, if your computer has enough RAM (the main deciding factor in computer performance), the operating system will need the pagefile very little and it will load most of the registry into RAM. So instead of worrying about defragging "in-use" files, you should be upgrading your RAM. The negative effect on performance when the computer has to use the pagefile AT ALL is many, many times that of some file fragmentation. When it comes to gaining maximum optimization of your computer it should be defragged, but when it comes to troubleshooting why a computer is unordinarily slow, file fragmentation cannot even be considered. That is the whole purpose of me even having this conversation. Finally, when time is money like it is for me, wasting what can be sometimes several hours to defrag a hard drive to gain a few feet after I already have gone a mile is just not worth it. Neither for me, nor the end user.

Don't Spam Me

My old laptop might serve as an example. Windows XP, 512 MB of RAM, 1.8 GHz Celeron, 80 GB disk with only 3.5 GB free. The largest DBX file (Outlook Express mail store) was around 2 GB and consisted of more than 120000 fragments. To initially load/open that folder, Outlook Express took around 15 seconds. After clearing out some space and defragging the drive for the first time in about 6 years, the initial load time of said folder within OE is down to a few seconds.

Is it your intuition or knowledge, but you are done good job.

Look what Microsoft Support recommend with comparable problems with Outlook Express and lack of free disk space (true with older Windows, but the method is universal):
1. Clearing out disk space (CCleaner is my favorite)
2. Defragging the drive

My opinion: open new Topic because your problem - low transfer rate for all your external drives, will be hidden here.

Is it your intuition or knowledge, but you are done good job.

Thanks. I used to be a Unix and NT admin, among other things, in a previous life.

Before I forget it, you might have noticed in the PCPitStop report that there's a rather low transfer rate reported for all of my external drives. They are all connected through one and the same powered Dynex USB2.0 7-port hub. Shouldn't I normally get a considerably higher throughput or is this the result of the additional cable length and the additional plugs involved?

(If this should be asked in a separate/new thread, rather than in this one, please let me know).

Hi Dont Spam Me..

I did take the opportunity to run a PCPitStop test on my newly built Windows 7 machine with an external USB drive hooked up.

Here's the results: http://pcpitstop.com/betapit/sec.asp?conid=23356811

I did notice that all your drive speeds were kind of slow. I am used to seeing, on average, about 35 to 40 mb/s on the C: drive. Yours is at 25. However, this is not so low that I would suspect any problems. It is most likely a slow hard drive or some other bottle neck in the hardware itself because 25 is not unheard of, it is just below average.

On my USB drives I got 22 mb/s, but this is a high quality western digital drive. I'm not sure what you are running but it would seem to me that the most likely explanation of your slow USB performance would be the limitations of the USB ports themselves. Possibly having too many plugged in at the same time effects speed, but I would lean more towards just a lower transfer rate due to USB limitations. They are certainly not going to rate the same as internal drives, and with your internal drive being relatively slow, its not really surprising that on average your USB is significantly slower. Hard to say there.

The data transfer rate of your external hard disk is significantly lower than internal HDD. In Windows XP you can increase the performance of external hard drives by enabling the write caching to increase the performance. By default write caching is disabled, because removable devices could not lose data if you unplug them quickly. But when you enabled this option then you should disconnect the removable drive in proper way.

My computer
Right click on the external hard disk and select properties
Select the Hardware tab
Under All Disk Drives, mark your external hard disk drive and select Properties
Select the Policies tab and select Optimize for performance.
OK twice
Reboot

Note: use the Safely Remove Hardware icon in the taskbar to eject the drive – otherwise you’ll risk corrupting your data by just removing the drive from the port.

I can't test - have no external hard disk.

The data transfer rate of your external hard disk is significantly lower than internal HDD. In Windows XP you can increase the performance of external hard drives by enabling the write caching to increase the performance.

Thanks for the advice, arTech. Unfortunately write caching is already activated on each of these drives and it also does not make any difference whether I connect them directly to the laptop or through the powered USB 2.0 hub.

Write caching is actually ON by default and external drives including flash drives and such have to be safely removed. It is user preference to turn write caching off so they can remove the drives without "safely removing hardware." But, again, a USB hard drive is never going to get the same speed as an internal hard drive.

There might be an explanation in Microsoft's knowledge base for the behavior that I am observing:
http://support.microsoft.com/kb/332023

On Policies tab we can see: Optimize for performance. This settings write cashing in Windows to improve disk performance. Shortly you can try with Enabled and Disabled write caching to see any differences. With better option please tell me: 1. How much is average transfer rate in megabytes/sec for your external disks respectively when transfer data from and to external disks – example large *.avi file (would be a large compact or compressed file, without applications running, small CPU usage before tests)? Can I guess – average 26 megabytes/sec. 2. Are there any differences in transfer rate with your external disks on other PC? Note: USB 2.0 has a maximum theoretical transfer of 60 megabytes/sec.