Hi,

I have only just discovered these forums and I registered to see if anyone can offer any positive suggestions about a subject that I have been looking into for the last couple of weeks.
I have run a search of these forums and found some past discussion that touches on alignment but my circumstances are different so I am starting a new thread.
If I do go over old ground then I apologise in advance.

My OS is XP-SP3. My PC has 3 hard drives, 2 internal, 1 external, all 3 are IDE, and all 3 are partitioned. I have a copy of Norton Ghost 10 which I have not had installed on my present computer as when I tried to move it from a previous computer I had repeated registration problems, by the time they finally got sorted out I had installed Acronis True Image, although to start with I ran NG 10 from the CD.

A couple of weeks ago I first heard about the subject of disk alignment when I discovered the Paragon Alignment Tool, after reading up on it I downloaded a copy and gave it a try.
I will not bore you with all the details but I have finished up with all my partitions aligned except for C:, the system partition.
In spite of trying numerous different ideas and prolonged correspondence with Paragon all attempts to align this partition and get the OS working have failed.
Using a combination of PAT and GParted I can align the partition but while the OS will start to load afterwards it crashes before it gets to the desktop.
If I restore an Acronis image it removes the alignment and goes back to the XP default.
If I try to restore Windows as files, some will restore but others are throwing up error messages about the originals having the wrong sector start point.

So my two main questions are:
1. Is there any way of getting NG 10 to do what Acronis will not, i.e. restore to an aligned drive without removing the alignment?
2. Can anyone suggest a way of getting beyond my present situation without doing a clean install which is not a practical option for the foreseeable future, if ever?

I read on another thread on this forum that XP-SP3 will install to an aligned partition and retain the alignment, unfortunately when this OS was installed it was SP2, although I have since made an SP3 CD.

Finally, those of you who have read this far are probably wondering why I don't just settle for what I've done so far?
The answer is, because of the results.

I have read much about what, if any, improvement alignment makes, on my PC the answer is phenomenal.
Before the alignments that I have managed to make if I was copying or moving a large file, 1GB+, I would set it up and go for a walk.
Since however I have copied an 8.8GB file from one aligned partition to another in 4 minutes 25 seconds.
To my mind it is definitely worth the effort.

@ Pilgrim

Interesting observations. I haven't noted any performance differences between cylinder aligned or 2048 sector aligned partitions on my SATA HDs. I know you can expect a performance improvement (at times) with Solid State Drives but you have IDE HDs. I'd be interested to hear more about your tests.

Just a few questions so that I understand what you are doing. Are you using 2048 sector aligned partitions? Except for the WinXP partition. Which tool do you use to determine alignment? Are you using RAID? Is DMA enabled on your HDs?

You will be able to produce a 2048 sector aligned WinXP partition with BING.

Hi Brian,

To start with I need to explain a couple of general things.
I am a practical person not a technician.
Until I bought my first PC five years ago I had rarely used a computer let alone owned one.
A year later, when the guarantee ran out, I was so sick of both retailers and manufacturers selling computers and then not wanting to know that I stripped it out and rebuilt it to its present spec less the second internal hard drive and a graphics card both of which I have added since.
As I have no technical training a lot of terms come up that I have no idea about, DMA is an example. When that happens I look it up. To answer your question about it being enabled, I have no idea.
The point being that what knowledge I have gained is based on specific experience not general computing so if some of my comments seem strange it is simply my limited technical vocabulary.

The story of what I have been doing is on another forum and to save writing it all out again I hope you will not mind if I give you the link:
http://www.raymond.cc/forum/software/20476-need-review-for-partition-alignment-software.html
That should answer some of your questions.

With regard to cylinder alignment or sector alignment, you will see from the link that I had trouble with the first partition on my second internal drive as well as the first, when I used GParted if the 'align to cylinders' box was ticked the partition would not align. When I unticked it, it aligned.

Paragon Alignment Tool is what I use to determine alignment but when I went into System Information I found that after alignment all the partition sizes on the PC were divisible by 4096. That may no longer be true because I found that the alignment had left small amounts, 1-2mb, of free space between partitions and I used GParted to extend the partition in front of the space to fill it.

No, I am not using RAID.

To reply to your comment on SATA drives:
Both the drives for my Netbook are SATA and I have not noticed anything like the performance improvement that I have got on the PC. However I think there might be a couple of reasons for that:
1. The Netbook,s C: drive was already aligned. (If you look at my edit at the end of the third post on the link you will see why I was interested in the information I found on this forum about the difference between XP Service Packs.)
2. Far less resources. (1.6Ghz Atom and 2GB RAM against 3.73Ghz D945 and 4GB RAM.)

Could you expand on your last comment. The only BING that I know of is a search engine?

Just to make one point very clear, my biggest issue is trying to find a way of aligning my PC's C: drive without reinstalling.

@ Pilgrim

BING is BootIt NG.

http://www.terabyteunlimited.com/bootit-next-generation.htm

There is a one month trial usage. Download. Unzip the file and make a boot CD.

double click makedisk.exe, next
dot in I accept the agreement, next
dot in Mouse Support Enabled, next
dot in VESA Video, next
dot in Partition Work (Don't put a dot in Normal), next
don't choose any Default Device Options (if necessary, these can be chosen in BING), next
leave Registration strings blank, next
select your CD burner drive letter (you can use a CD-RW or a CD-R disc)
Finish

I'll talk you through converting your WinXP partition to 2048 sector aligned.

Let me say that I've used WinXP and Win7 on IDE and SATA HDs with both partition alignments and have not seen one iota of performance difference due to the alignments. I haven't seen any reports of a performance difference except on SSDs and RAID. But I'm interested in what you have done. Let me know when you have made the BING CD.

Could you post a partinfo so we can view your partitions?

http://www.terabyteunlimited.com/downloads-free-software.htm

Double click partinfg.exe  Then click File, Export to text file. Please attach that file to your post.

Hi Brian,

One of those days today.

Since I posted earlier I have received the latest 'experimental' version of the Alignment Tool from Paragon. It still cannot do the job.

Then I went to the TeraByte site and the first thing that caught my eye was BurnCDCC which I have been using for the past couple of years but I was not aware of their other programs.

I am attaching the PartInfg.txt file and I am just going to create the CD.
My internal CD/DVD drive has been having problems recently due to a very high room temperature so I might have to use an external one.

I have had a quick read of the BING manual but as you know the program I am quite happy to follow your instructions, the offer of which is much appreciated.

I would prefer not to affect the second partition on the system drive if it can be avoided.
If there is a question of alignment at the end of the C: partition I can if necessary create some free space there before we start and fill in any free space afterwards.

To reply to your comment about improved performance I have read figures from 0 to over 300% on all types of drives,
the main reason I noticed the difference on my PC was because the improvement was so big.


This is getting ridiculous.

While I have been writing this Paragon have sent me an ISO image of their latest effort to try it offline.

I will be back.

I have created the BING CD but I will not in any event be able to do anything before tomorrow.

The latest Paragon ISO did no better than the installation.
I did find out that if you do have an aligned XP partition their current version of imaging software is likely to remove the alignment. I was assured however that the next version should keep it.
What has interested me from the beginning is that not once has the word 'BETA' been mentioned, strange that.

Pilgrim wrote on Jul 1^st, 2010 at 7:36am:

I am attaching the PartInfg.txt file

I can't see it yet.


Pilgrim wrote on Jul 1^st, 2010 at 7:36am:

I would prefer not to affect the second partition on the system drive if it can be avoided.

We won't touch that partition.

Pilgrim,

"As I have no technical training a lot of terms come up that I have no idea about, [...] what knowledge I have gained is based on specific experience not general computing so if some of my comments seem strange it is simply my limited technical vocabulary."

Acknowledged.  But understand you are involving yourself in technical topics here, so it helps if you try to be precise and unambiguous.  For instance:

"Using a combination of PAT and GParted I can align the partition but [...] If I restore an Acronis image it removes the alignment [...] Is there any way of getting NG 10 to do what Acronis will not, i.e. restore to an aligned drive without removing the alignment?"

One of the things Brian was getting at in Reply #1 was what kind of alignment are you talking about?  The two most prevalent types are cylinder alignment (common on older systems) and 2048-sector alignment--aka, Megabyte-alignment, which is being used on most new systems.  It appears you're talking about MB-alignment, but you don't want to confound people or make them read between the lines to guess what you really meant to say.  To infer older systems are "not aligned" is akin to newbies who improperly assert their store-bought Windows machine is "not partitioned", and just as muddling to follow.

As for your immediate question, it might help some people to know more of the details of what you did:  Which version of True Image did you use?  Was it run from the hard disk, or from CD?  Was your image of the whole disk, or just the one partition?

I think Brian's plan is to have you restore with TI and then readjust the partition boundaries with BING.  But I think before this is over you'll want to know if this will happen again the next time you want to image/restore.  To understand that, I think the key is recognizing whether or not your tool can restore a partition image to an existing partition.

For example, Ghost 2003 can use the boundaries of an existing partition--it simply "pours" the contents of the image into a pre-existing partition, so even though Ghost 2003 is an old, DOS-based tool that knows nothing about MB-aligned partitions, it actually works.  It simply uses whatever you give it, and doesn't question how it is aligned.  Other alternatives that I suspect take a similar approach might include the Windows-based Casper-XP, DriveImage XML, or the linux-based "dd" tool.

OTOH, some utilities can only restore into unallocated space.  If there is an existing partition in that space--such as your carefully, PAT/GParted-prepared, first partition--the utility will delete it, create its own new partition, and then restore the contents.  BING is one such tool.  However, BING understands both cylinder-alignment and MB-alignment.  Other tools may not.  If the tool only understands cylinder-alignment, then when it deletes/recreates the partition prior to filling it, it will naturally create a cylinder-aligned partition.

Note Ghost 10 is an older version that won't create a MB-aligned partition.  That may also be the case with your version of TI.


Brian wrote:

"Interesting observations. I haven't noted any performance differences between cylinder aligned or 2048 sector aligned partitions on my SATA HDs. I know you can expect a performance improvement (at times) with Solid State Drives but you have IDE HDs. I'd be interested to hear more about your tests."

I, too, would be interested in hearing more about how Pilgrim ran his tests.  Like others,  I've considered the advantages of MB-alignment to be more theoretical than observable.  Maybe we just haven't copied big enough files?

Hi,

Thank you both for your replies.

Brian, I have no idea what happened with the file, when I posted yesterday it was showing as being attached?
I will try attaching it to this post and then start another to address the other issues raised, that way if there is a further problem I can try and sort it out.

---

@ Pilgrim

Thanks for that. On HD0, MBR slot 1 is 2048 sector aligned at the start but not at the end. I assume you want to align the partition in MBR slot 0 to be 2048 sector aligned.

Disconnect your external HD.

Boot from the BING CD. Click Close on the Work with Partitions window and Click Settings. Remove the tick from..

Align on End
CHS Alternative

Put a tick in..

Align MBR End HS
Align 2048

Then click  OK and click Partition Work.

Select the WinXP partition on HD0 and click Resize (in the Actions column)
OK (it will then check for errors)
In the New Size field, reduce the value by 5 MB, OK
Continue
Close

Select the WinXP partition and click Slide (in the Actions column)
Make the Free Space Before 2 MB
OK
Continue (this could take a while depending on how much data has to be moved)
Close (when completed)

Select the WinXP partition and click Resize (in the Actions column)
OK
Leave everything alone and click OK
Continue
Close

Select the WinXP partition and click Properties
Divide the LBA Information Start by 2048. You should get an integer.
Divide the LBA Information (End +1) by 2048. You should get an integer.

Close Work with Partitions
Remove the CD and click Reboot

Did it work out?

Dan,

If you will forgive me I need to make an observation about the appearance of your post.
I have set the forums appearance to 'default' to enable me to see it clearly, your quotes appear in yellow in my browser and are impossible to read against a white background.
I have had to copy the entire post into Wordpad and am reading it from there. Sorry.

In reply to your first quote, the reason I wrote that was so that if anything I said was not clear I could be asked for more details. My intention was not to be ambiguous but exactly the opposite.
To take up something you say further down, my intention has never been to confound people or make them guess at what I meant.
Both you and Brian obviously know a lot more about what we are discussing than I do, my lack of knowledge was what prompted me to post in the first place.
I will not apologise because I do not know as much as you appear to and it is sad to see you refer to 'newbies' with such apparent contempt.
If there is anything I said that was not understood or required more details all that was necessary was to ask, allowing for the fact that I may not know enough to answer in the 'precise' way you are looking for.
There was no need to start your post by implying that I am (deliberately?) being awkward.
Having got that out of the way I will address the main issues.

What kind of alignment?
The only way I can answer that is by taking you through what I did.
When I first used the Alignment Tool analysis on my PC it showed that none of the partitions were aligned, after I let it run it showed that all the partitions were aligned except for C and E, the first partitions on the 2 internal drives.
As I have already said, C remains unaligned. The following is how I aligned E.

I had read that it was possible to align a difficult drive by changing the starting offset slightly so that was what I looked into.
When I originally partitioned the drives I had used Acronis Disk Director Suite 10 but I knew that the smallest movement I could get on that was 7.84mb which was too much.
I had read an article about alignment which suggested using GParted, a program I had never used before. In the article it suggested using the command prompt from the GParted CD to actually align a partition but I could not get it to work.
In a number of places 2mb was suggested as a practical offset so I used GParted to create it, (I left the box 'align to cylinders' unchecked as had been advised) on finding that it was still unaligned I ran PAT which aligned it.
I then found that all the partitions which had been aligned had a small amount of free space in front of them so I used GParted to extend the partitions in front of the spaces to fill them.

The version of True Image that I have is 11.0 build 8101.
I only ever create images of individual partitions, mainly C.
If I am restoring a complete image I use the CD, if I am restoring separate files/folders I do it from the hard drive.

I should add that I use the same version of TI on my Netbook on which the C partition appears to have been aligned from new and has been restored using TI without losing the alignment.
There is however one major difference in the two OS installations.
The OEM installation on the Netbook was XP-SP3. The original installation on the PC was SP2 and then updated when SP3 came out.
From what I have read it is possible to clean install SP3 to an aligned partition and retain the alignment but not SP2. Adding SP3 at a later date would obviously not change that.

That answers your questions as far as I am able, if you need any more details please ask.
There is of course the information on the link I previously posted.

Before I end for now I have learned something more about improved performance with alignment, it is not precise but it is reasonably accurate.
I backup both my computers to an internal partition every day that I use them, once a month I backup a complete partition image to the respective external drives.
Yesterday I backed up the Netbook in that way for the first time since I aligned all the partitions.
The size of the images has not varied by more than a few 10's of mb's over many months so any decrease in creation time should not be great.
Yesterday it took close to 12 minutes as opposed to the usual 15, i.e. an improvement of around 20%.

While I have been writing this I see that there has been another reply, I will post this first and then check it.

Pilgrim wrote on Jul 2^nd, 2010 at 4:42am:

My intention was not to be ambiguous but exactly the opposite. To take up something you say further down, my intention has never been to confound people or make them guess at what I meant. Both you and Brian obviously know a lot more about what we are discussing than I do, my lack of knowledge was what prompted me to post in the first place. I will not apologise because I do not know as much as you appear to and it is sad to see you refer to 'newbies' with such apparent contempt. If there is anything I said that was not understood or required more details all that was necessary was to ask, allowing for the fact that I may not know enough to answer in the 'precise' way you are looking for. There was no need to start your post by implying that I am (deliberately?) being awkward.

I don't see where I implied you were deliberately being ambiguous or awkward.  I said that when you (or anyone) are seeking to discuss technical topics, you (and everyone) need to be as precise as you can to avoid confusion.  Everyone needs to understand terms the same way, and I was making the point that your understanding of the term "alignment" seems to be somewhat skewed.  I'm sorry that offended you.

My "newbies" comment was drawing a comparison to common inquiries about partitioning problems (not necessarily here, but widespread enough in other forums to be familiar to everyone here) from users who don't understand the distinction between a disk partitioned with a single partition and a disk that is unpartitioned.  If the basic concepts are misunderstood, it's going to be hard to communicate in a technical discussion in any meaningful way.  I may either attempt to educate or take a pass and let someone else try--but either way, I can't help if the reader doesn't understand (or, in some cases, won't learn) the concepts central to the topic.  If you choose to extrapolate that to be contempt, that's your prerogative.  I can live with it.

Brian,

Thank you for the instructions, I will try it as soon as I have sufficient time which might not be for a few days.

It seems that Bing works in a similar way to GParted although that does not have the 'align' option built in which is why I had to use PAT as well.
Having one program that will do both makes me more optimistic about the outcome especially as you have successfully used this method before.

One thing that I would like to ask is about the question of how long it will take.
In the instructions I have for GParted it suggests reducing the partition to used space plus 1GB before you start, so that there is not so much to move, and then expanding it after to fill the free space.
Would that make sense with Bing as well, and if so would it make a difference to any of your instructions, 'Align on End' for instance?

If this works and I am pretty confident that it will, and taking everything else into account, I am reasonably optimistic that I should be able to continue to use TI provided that I only use images that are made after the partition is aligned. If the BING trial lasts for 30 days I think it will be worth trying it while I still have the option to realign if I need to.

Incidently, what happens to the BING CD after 30 days, does it just stop working?

I started writing this immediately after my last post but before I could finish it we had a power cut and nearly an hour on we still have no power.
I am writing this on my Netbook but I will not be able to post it until the power comes on as I have no internet connection.

Power cuts are not frequent where I am but nor are they uncommon, we get around 3 or 4 most years.
This is the 21st Century isn't it?


Dan,

Peace.

I meant no more offence than you did.

If I might briefly pick up on two points in your last post:

Quote:

Everyone needs to understand terms the same way

When you have two very different levels of understanding that might only be achieved after the one with the greater understanding explains things.


Quote:

users who don't understand the distinction between a disk partitioned with a single partition and a disk that is unpartitioned.

On this I would have to plead guilty not because I do not understand the difference but because of the way I think about it.
To me a partitioned disk is one with more than one partition.

Precision in any form is not a requirement in my day to day life any more than it is in, I suspect, the majority of people's lives.
I explained my lack of knowledge so that anyone reading this thread who was knowledgeable would not expect me to be able to communicate on their level without at least a degree of explanation on their part.
There are a whole range of subjects that have their own terms and definitions, to expect someone who has admitted to having little knowledge to use them as well as someone who does ......?

I have accepted your apology and now I hope that you will accept mine.
I also hope that it will not deter you from contributing further to this thread if you have something to add.

@ Pilgrim


Pilgrim wrote on Jul 2^nd, 2010 at 7:32am:

Would that make sense with Bing as well, and if so would it make a difference to any of your instructions, 'Align on End' for instance?

No, just follow the instructions without any embellishments. After succeeding, don't try to change the few MB of unallocated free space at either end of the WinXP partition.


Pilgrim wrote on Jul 2^nd, 2010 at 7:32am:

One thing that I would like to ask is about the question of how long it will take.

This depends on your computer, not on BING. Allow roughly 1 minute for every 1 to 2 GB of data in the partition. It is the data amount, not the partition size that is important.


Pilgrim wrote on Jul 2^nd, 2010 at 7:32am:

what happens to the BING CD after 30 days, does it just stop working?

I bought it before the 30 day limit and now have BING installed on the HD. It is one of my most useful apps.


Pilgrim wrote on Jul 2^nd, 2010 at 7:32am:

I am writing this on my Netbook

Which Netbook do you have? I have an Asus 1000HE and love it. BING is installed on the HD of course.

Pilgrim, I've read a lot of Dan's posts and I can assure you he meant no malice. It was purely constructive criticism. I thought about making the same points after reading your first post but being a lazy typist I "left it till later".

I looked at your partitions on HD1 in that partinfo. The start of each is on a 2048 sector boundary but the partitions in slots 2 and 3 don't finish on a boundary. Slots 0 and 1 are fine.

I haven't tested recent Acronis True Image versions as to whether 2048 sector alignment is preserved after an image restore. Please let us know. All of the TeraByte Unlimited apps (BING, IFW, IFD, IFL) will preserve whichever alignment you choose when restoring to the same or another HD.

"On this I would have to plead guilty not because I do not understand the difference but because of the way I think about it.  To me a partitioned disk is one with more than one partition."

That's exactly the kind of position I have a problem with.  If one comes into a technical discussion and says, "My HDD is unpartitioned," I immediately conclude, "Oh, he doesn't have an MBR or partition table."  It just wastes everyone time until it's revealed, "No, what I really meant was..."  Relevant terms must be used correctly or we can't communicate.  It's especially inexcusable if one understands the difference yet willfully chooses to use terms incorrectly.  Call it contempt if you wish, but I opt out of those debates so I don't waste my time.

"When you have two very different levels of understanding that might only be achieved after the one with the greater understanding explains things."

Exactly.  But it doesn't work if the pupil takes offense when explanations are provided.  As I explained, you need to understand the term "alignment" correctly.  I'm not just being pedantic here--your partinfg.txt report reveals you have some real issues.  Some partition boundaries are MB-aligned, some are CHS-aligned, and some are unaligned.  If that term means different things to each other, we aren't going to be able to communicate.

As it is, I'm already confused.  Supposedly there are three HDDs but the partinfg.txt report only shows two.  Yet the HDD brands, sizes or where they are connected hasn't been stated, so I don't know which comments to associate with which evidence.  Brian says, "MBR slot 1 is 2048 sector aligned at the start but not at the end."  I can't tell what he's referring to, but he's probably a better mind-reader than I am.  For me, I would need more precise details and less ambiguity.

I read the Paragon Alignment Tool pdf and the only HDs discussed were the new 4K sector HDs. Not our stock standard HDs.
The following is from TeraByte Support and is an interesting read about 4K sector HDs and alignment.


Quote:

For some reason many users are thinking there is something special about these new drives with internal 4K sectors and that they require "advanced formatting" to function properly.  Let me just nip it in the bud now - this is mainly marketing BS. This should all be obvious to anyone with basic knowledge of drives/file systems and the link.  Here's some simple brief information: What's the deal with the 4K sectors? A sector is the smallest unit/block of data a drive can transfer.  A traditional hard drive has used 512 byte sectors and the plan to change to a larger internal sector size has been in the works for years and now there are drives that are using 4K sectors.  The larger sector sizes allow there to be more data on the same platter because there is a gap between each sector (including the ECC data).  These drives are still fully compatibly with all existing software and operating systems because they (properly) provide a 512 byte sector view and internally handle read/writing in 4K chunks using read-modify-write.   With proper caching technology (built into the drive) the affects of this should be negligible on typical desktops. What about this alignment stuff? Alignment has to do with where the partitions and data are located on the drive.  Obviously the "optimal" location is where all data to be read/written are contained in the least number of sectors; however, it's marketed simply as aligning partitions to the size of a sector (or block for devices not using sectors).  The reality is you may have a more optimal configuration or setup if the partitions are not aligned to the size of a sector - what matters is where and how IO occurs on the drive which is dependant on the applications (including OS) use of data, file system in use, the file system driver implementation, and any cache manager implementation. On the application side of things, one given is the OS paging file which can be heavily used by the OS.  The OS will read/write an entire page which is 4096 bytes (4K) so having that file aligned to a sector size is probably the biggest single alignment factor on a typical desktop computer.  For that file to be aligned, the clusters of the file system need to be aligned (see below).  While we're talking about the page file, if your file system cluster size is < 4K you'd also want to ensure that the page file is not fragmented or that could affect performance as well. For the file systems, NTFS clusters (allocation units) occur starting at the start of a partition so if that partition is aligned on a sector size, so will a cluster.  For FAT the data may not be aligned even if the partition is aligned on a sector size.  However, you can use BootIt Next eneration (BING) to format a FAT/FAT32 partition using "align for NTFS" which will be aligned; you can also use BING to align an existing FAT/FAT32 partition using its "slide" / "align for NTFS only" option.   BING is located at http://www.terabyteunlimited.com/bootit-next-generation.htm However, even if a cluster is sector aligned, it doesn't mean the file system driver is going to read/write entire clusters at a time (in fact that would normally be a waste of time and slow file access way down).  The only time you will see that is if either the file system driver is poorly implemented, or for security the writes on newly allocated clusters are zero padded (to end of new cluster) for security. Another factor is a cache manager.  If this cache manager reads or writes entire clusters (or multiple of clusters) then having a cluster aligned with a sector size will be optimal for the cache manager and hence anything that uses it. What's the bottom line on alignment and Windows? Due to the cache manager and paging file it's probably best that you align your partitions (particularly on heavy used servers/systems).  You can use BootIt Next Generation to align your new or existing partitions by enabling its Align 2048 setting (you should also disable Align on End, enable Align MBR End HS - see manual for details at http://www.terabyteunlimited.com/support-bootit-next-generation.htm ) before creating the partition or before using its slide (move) option to move the partition location. However, if your partitions are not aligned or you don't want to align them then simply don't worry about it because the system will continue to function.  If you find that the performance is not as expected then align them at your convenience to see if that was the cause of your performance issues.

Brian,

In answer to reply #13:
The simple bit first, I have a Samsung NC10. Although my experience of different computers is extremely limited I am impressed by what it can do.
I used to use a friend's computer that had a 1.3Ghz CPU and you frequently had to sit and wait for things to happen, I was a bit concerned that I might find the same with this especially compared to what I was used to on my PC but I have been very pleasantly surprised.
I have separate external hardware for both computers, except for a printer and scanner which are rarely used and both are attached to the PC, my eventual intention is to set up the Netbook in another room as the heart of a completely independant system.

Sector boundaries: The reason that the 2 partitions you mention don't finish on a boundary might well be because I filled up the free space that was left after alignment. There are probably others as well.
The reason I filled up those spaces is because of True Image.
I have moved nearly 11GB of less frequently used applications from Documents and Settings and Program Files on C to similarly named folders on D.
After I had got to where I am now with the alignment I did not initially realise that the free spaces existed.
The alignment disrupted the file system on D so I needed to restore an image to it as it is the only other partition that I make an image of, for all the other files/folders that I back up I either copy them directly or use SyncToy.
The restoration went without a problem as usual but at the end of it I was astonished to see a window saying that TI was aligning the partition it had just restored with the system.
As the system partition was never aligned that meant that it removed the alignment from D and moved it back into the free space.
I realigned D and then instead of restoring the image again I copied the files across from another partition, which is how I came to be moving an 8.8GB file.
Looking back my next thoughts and action might well have been misguided.
I concluded that if there had been no free space in front of D it would have had nowhere to go so I went round all the partitions and filled it up, with the exception of the last 6 partitions on the Netbook's external drive which are extended/logical partitions and would not move.
Based on what I know now I suspect that if I were to try that again with the same image the restoration would either fail or TI would try to move the partition in the opposite direction to align it.
While writing this it has occurred to me that it might be easier to first experiment with TI on D once C is aligned as I have already done it once and it will be easier to put right should it prove necessary.
If it is successful then it should also work on C, if it is not I will leave C alone and find another means of back-up.

When we get my C partition sorted out if you would be prepared to advise me how to set the rest of them up correctly it would be much appreciated but as what I already have is so much better than what it was I will happily settle for that if necessary.

I have no problem whatsoever with Dan, in fact I suspect that as people we have quite a bit in common. I have more than once found myself on the opposite side of the sort of discussion we have been having when it is about a subject that I know well.
When I finally realised some of the effects my attitude was having on other people, even though it was unintentional, it troubled me, so I gave it some serious thought.
One thing which I realised fairly early on was that people were being deterred from asking me for advice or help through fear of my response, even though at times it was badly needed.
I finally concluded that we all had the right to be the people that we are and that none of us had the right to expect other people to be like us or to think in the same way that we do.
But like everyone else I am human and there are still times when I come into contact with people who remind me of myself that instead of letting things pass I react, which is what I have been doing.
Mea culpa, mea culpa, mea maxima culpa. And I am sorry.

Dan,

It is not about willfully choosing to use terms incorrectly, it is simply about not thinking like you do.
Your first post in this thread did not start with an explanation but with criticism, even if it was constructive as Brian described it, it was still criticism and as such I considered it unwarranted.

My understanding of the term alignment is very basic and comes from reading the Paragon White Paper on the subject.
By the time I had got very far into it I was lost, particularly when they started talking about servers and RAID, two things of which I have no experience and absolutely no understanding.
But as I have said I am a practical person and I understood enough to realise that from a practical point of view to have your disks/partitions aligned made a lot of sense, even if it offerred no improvement in performance it should make the operation of the system more efficient.

My interest in things is that they should work efficiently, and that does not always require an understanding of how they work, although I am the first to admit that at times it can be an advantage.
I started out as a mechanic and I have always believed that my understanding of how a vehicle works has made me a better driver but that does not make somebody without that knowledge a bad driver.

What I am about to say you will probably totally disagree with but again it is my way of looking at it.
From what little I understand about alignment it is more important for the beginning of a partition to be aligned than the end because as I understand it, it is that which affects how sectors are read or written to. The relatively small spaces at the end of a partition are far less important and in my case probably never used, this may not be 100% correct technically but then neither am I.
If this assumption is wrong I would be pleased to have someone explain.

As far as the number of hard drives is concerned there are 3 as I stated previously, the external one is permanently connected to the PC but only switched on when I am going to use it, approximately half a dozen times a month on average.
Brian was asking for information about my C partition so I saw no reason to turn the external drive on. The information provided therefore was for the 2 internal drives and I think that even I understand enough to know that HD0 has to be the system drive on a Windows system.
If either of you want information now or later on my external drive I will simply create another report, if you like I will do so anyway after I align C so that you will have before and after readings on it.


You have both mentioned the issue of testing and Dan raised the question of file sizes, I think that is a very important point and that file sizes should be relavent to the system being tested.
In the two observations I have made about improved performance my assessments have been subjective not objective.
I have not used a stopwatch or any software to measure differences I have simply compared the time it used to take to do something and the time it now takes.
In respect of the 8.8GB file the only thing I could say about before was that it used to take a very long time to copy/move any large file, when I noted the time after alignment I was sitting in front of the computer the whole time and was astonished at how quickly it finished, but I have no direct comparison.
The time it took to create the image was something different, for one thing the file size, 5GB, has remained pretty constant over many months, for another there are several reasons why I should notice the times both before and after.
Until a short time ago both computers took almost the same time to create an image of their C partition, 15 minutes, even though the PC had 3 times the amount of data it has far more resources.
Besides the alignment I have also, as I said, recently moved a large amount of files off of the PC's C partition and both times have dropped almost the same amount to around 12 minutes.
Even if my assessment of the difference is a whole minute out, which I am certain it is not, that would still leave a significant improvement.

Because I have looked at very large files, and periods of minutes rather than seconds, it has made comparison that much easier and differences more obvious, even if my assessments lack accuracy.
If I were setting up a test on any other computer and wished to be more accurate I would pick a file size that corresponded with how long it took the test machine to copy/move it, and a time of around 15 minutes would make sense to me. It is short enough not to drive you mad waiting but long enough to make any difference clear.
If I had used much smaller files I would probably not have noticed much, if any, difference. In fact in Reply #2 I said that uptil then I had not noticed much improvement with the Netbook, it was only when I got up to 5GB with the image that the difference was great enough not to be ignored.


I am not a lazy typist, just very slow. It has taken me 3 hours to type these 2 posts, I need a rest and I have other things to do.
I will try to sort out C before I post again.

@ Pilgrim

I'll try that too. Copy a 5 GB file from a system with cylinder aligned partitions and note the time. Convert the partitions to 2048 sector aligned and copy the file again. Later today.

My test computer has a 40 GB (HD0) HD and a 20 GB  (HD1) HD. WinXP (4 GB) and a 8 GB empty primary partition were on HD0. A 9 GB empty primary partition was on HD1. All partitions were cylinder aligned. A 4.79 GB file was copied back and forth between the two non OS partitions. The copy times in seconds were…

HD0 to HD1  281, 278
HD1 to HD0  294, 292

All three partitions were converted to MB aligned. The copy times in seconds were…

HD0 to HD1  289, 281, 279
HD1 to HD0  300, 295, 297

There is no significant difference in copy times between the two alignments in this test.

Brian,

I've not got the time to look at recent posts at the moment, I have a problem.

I followed your instructions to the letter and everything went as you said it would until I rebooted, apart from the fact that it is getting slightly further it is doing the same as it was when I used the other software. This time it is getting right to the point where the Desktop should open and then crashing.
Also like last time I cannot access the installed Recovery Console so as I am writing this on the Netbook I am running chkdsk /r from the Windows installation disk although that made no difference before, I did try entering fixmbr and was informed that I had a non-standard or invalid MBR and if I continued I risked losing access to all partitions on the disk.

My emails are all set up on the PC so as I cannot receive automatic notifications of a reply I will keep watching this thread for the next couple of hours.

I realise that we are in very different time zones so if I do not hear from you I will do a restore and post again tomorrow as I need the PC up and running by then.

Pilgrim,

I can't explain your problem. I'd restore the image too.


Pilgrim wrote on Jul 4^th, 2010 at 3:40am:

I did try entering fixmbr and was informed that I had a non-standard or invalid MBR and if I continued I risked losing access to all partitions on the disk.

I see that every time I try fixmbr. I assume it's a standard message because if you run fixmbr one minute later you see the same message. But there is nothing wrong with your MBR.

Just to keep you up to date, chkdsk finished running and I rebooted with the same result.
The window with the progress bar opens for the same time as it normally does and then the screen goes black, again as normal, nothing happens for several seconds and then I get the BSOD.
I tried accessing Safe Mode but cannot get into it.

I have looked at your test results and the only thing that I can say is that the one thing evident in all the test results that I looked at was that there was no consistency.
The tests I saw were for all types of disk and as I said in an earlier post the results ranged from no improvement at all to one person claiming a 300% improvement on an SSD.

I hope Dan has an explanation for your non loading OS. Did you note the BSOD error?

Hi Brian,

I just caught your post.

Like a lot of other things my knowledge of the Recovery Console is not extensive, is there anything in there that you can think of that might be worth trying before I restore?
It will be a shame to have to give up having got this far again.

I also wondered if there was anything that could be done if I was able to get into the hard drive with a different boot disk, among the stuff that I got from Paragon was a Recovery CD which was supposed to be used if after running PAT you could not boot, if you have no further ideas I think I will give it a try and see what options it offers.

It seems as if Windows is trying to open but fails right at the last.

Brian wrote on Jul 4^th, 2010 at 4:24am:

I hope Dan has an explanation for your non loading OS. Did you note the BSOD error?

No but I will run it again before I try anything else.

BSOD:

STOP: 0x0000008E (0xC0000005,0x00000000,0xA9DEAC9C,0x00000000)

Might as well be heiroglyphics to me.
Need an MS Rosetta Stone.

I had a look at the recovery CD. Nothing on it that would help you that I could see. Your WinXP boots but it fails to load.

http://support.microsoft.com/kb/315335

I guess it's worth removing RAM sticks one at a time to see if the error goes away. But your computer doesn't like a MB aligned OS partition.

Do you have the latest BIOS?

Paragon CD never changed anything.
Just tried LKGC, BSOD now reads:

STOP: 0x0000008E (0xC0000005,0x00000000,0xB6BD7C9C,0x00000000)

Going to put the end of the partition back to where it was and then restore.

I'll look back on here tomorrow to see if anyone has any ideas.

Yes to the latest BIOS.
Opening up the PC not an option at the moment but I recently ran a memory check and it showed no faults.
The problem only happens after the partition is moved.
I keep thinking about what I read about the differences between an original installation of SP2 against an original installation of SP3 and the question of whether it somehow 'locks' to the hard drive. That's one for the technicians, it's beyond me.

You can use BING to re-create a cylinder aligned partition. In Settings, remove the two ticks you entered and tick the two boxes that were originally ticked. Actually, when you boot from the BING CD the ticks should be in the default position. Slide the start of the partition a few MB. Then do the Resize exercise on the end of the partition. I hope that should be enough.

Sorry, lost my internet connection for a couple of minutes while I switched the power off.

I got down on the floor to have a look at the RAM, I cannot take it out without disconnecting everything and putting the tower on its side on a table, I did manage to loosen the clips and reseat all 4, but nothing changed. I am confident in my own mind that is not the problem.

I have got TI running without altering the partition, it will almost certainly reset the start of it and I am not too worried about the end. As long as C is not aligned I cannot restore D with TI anyway, which was my main concern.

I need to break off now, when I get everything back in place I will send you a PartInfo file including the external drive. I have made one of the Netbook as well. Perhaps you could have a look at them and see if you can spot any other issues besides C.

TI just finished, rebooted without problems and ran PAT because it was the easiest, as expected C is showing as not aligned.

Got to go.

Nice to hear WinXP is working again.

I am attaching the 2 PartInfo files I have just made so that you can see where everything is.
My intention was that all the partitions on any single disk should be the same size leaving as little free space as possible.
Recently, using software that is more accurate than Acronis DD with which I originally created the partitions it appears that I succeeded better in some places than in others.


I would like to sum up where I think we are with the PC's C partition and ask for comments:

1. The events of earlier today largely matched my other attempts, the partition can be aligned but once it is the OS will not run on it.
    I refer here to sector alignment so there is no misunderstanding.

2. Mention has been made of cylinder alignment and sector alignment but is there a state where a partition is simply 'not aligned'?
    The reason I ask is that when I run an analysis with PAT it has three options:
    1. Aligned.
    2. Not aligned.
    3. Unable to align.
    Or does the 'not aligned' only apply to the type of alignment which PAT uses, which I assume is sector alignment.

3. Staying with PAT for a moment.
    Through my contact with Paragon I have tried several versions of this program, 3 were installations, 3 were CD's.
    A couple of the installations matched a couple of the CD's but as far as I can work out without version numbers 4 of them were different.
    On all versions except one the only information that I got about a partition was 'aligned' or 'not aligned'.
    The other one produced an error message which said it had failed due to 'cross-linked files'.
    As far as I can make out this is being quoted as the cause on most, if not all, the systems that PAT fails on.
    Never having heard of this I looked it up and among other things I found a reference to .chk files.
    I ran a search for these using Everything and found that both my computers had 3 of them.
    They are all in different Windows folders, they are all 8kb, they are all called edb.chk, and if they are deleted they reappear at the next reboot, 1 of them with what I can only presume was an installation date, the other 2 with the current date and time.
    I have been able to make no understandable connection between these files and the failure to successfully align C.

4. I was aware before I started this thread that different imaging software had different effects on a partition when restoring a full partition image, Dan has also referred to it in one of his posts.
    I would like to know if anyone knows of a program that will restore a non-aligned image to an aligned partition without removing the alignment. If that is, such a thing is possible.
    It would need to have a free trial and it would need to run from a CD.
    My idea being to create an image, then align the partition, then restore the image and see if it would run.
    I would also make a current TI image so that I could get back to where I started.

That's it, I am out of ideas.

---

@ Pilgrim


Pilgrim wrote on Jul 4^th, 2010 at 10:27am:

The reason I ask is that when I run an analysis with PAT it has three options: 1. Aligned. 2. Not aligned. 3. Unable to align.

I tried the PAT on my test computer. The three options I saw were
1. Optimally aligned
2. Not optimally aligned
3. Partition cannot be aligned.

I ran the tool and my cylinder aligned WinXP and Data partitions became MB aligned. I don't understand what happens to your WinXP partition when it becomes MB aligned.

The Paragon White paper on alignment doesn't mention the HDs you and I are using. The paper discusses the new 4K sector SATA HDs. I know my test and my main computer perform the same with both alignments and I suspect non 4K sector SATA HDs are alignment agnostic as far as performance is concerned.

Brian,

When I listed the PAT options in my last post I never had the program open in front of me and I could not recall the exact wording only what it implied.


Quote:

I ran the tool and my cylinder aligned WinXP and Data partitions became MB aligned.

This is why I asked the question if it was possible for a partition to be 'not aligned'.
PAT is changing your partitions from one type of alignment to another, what if I am starting with my partition in a third state?
As I say I do not know if this is possible which is why I asked the question.

Dan referred to cylinder alignment and sector alignment as the two most prevalent types of alignment.
As he is the one who has made a point of precision in definitions, and I mean no sarcasm in that, is he not saying that there are other states?

Like many of the articles I have looked at the Paragon White Paper goes way beyond my understanding so I will not pretend to have read it in detail. What I did understand was some of the diagrams which laid the basics out graphically.

Let me go back to something I have mentioned a couple of times that has not been picked up, the differences in installations of XP with different service packs.

Paragon have admitted that they started developing PAT after SP3 came out and have never tested it on what is, or in my case was, an SP2 installation.

Have any of the systems which you have successfully used BING on been SP2 installations?

The service pack issue might be irrelevant but I have seen it raised in a couple of places neither of which offered a definitive answer.

There is also the question of 'cross-linked files' to which Paragon attribute most of PAT's failures.
This is something else that I do not fully understand.
Apart from a reference to .chk files the only other thing I could make sense of is that they are files which are connected to more than one point but I do not understand exactly what they mean by that. The only thing that comes to my mind are hard or symbolic links and while I have the software to create them I have never used it because by the time I got to moving files from one partition to another I had picked up a program that enabled me to change the paths in the registry to the new locations all in one go.

@ Pilgrim

I've been using WinXP SP3 since it was released so I haven't tested with SP2. Are you using SP2?

On HD1, your partitions in MBR slots 2 and 3 don't finish on a cylinder or a MB boundary so they are unaligned. I'm interested to hear from Dan about the significance of a non aligned end.


Pilgrim wrote on Jul 5^th, 2010 at 3:11am:

There is also the question of 'cross-linked files' to which Paragon attribute most of PAT's failures.

I've no idea.

To align the end of your partitions to a MB boundary, use the Resize option in BING. Just click OK but remember to select the appropriate choices in Settings.

Brian,


Quote:

Are you using SP2?

No I am using SP3 but the point I am making is about what was originally installed.

My Netbook was installed with SP3 from new and its C partition has always shown as aligned.
My PC was originally installed with SP2 and later updated.

What I read was that SP2 would not install in an aligned position and updating to SP3 did not change that.
The same article suggested that with SP2 there was some sort of interaction between the OS and the disk that prevented the OS, and therefore the partition, from being successfully moved.
This was apparently resolved in SP3 but only applies to clean installs not updated systems like mine.

As I am still in touch with Paragon I have written to them and brought them up to date with events.
I have also asked them if they can provide further information on 'cross-linked files'.
These are being linked to the problems PAT is having with other operating systems as well as XP.

As I have only used BING once I am not completely confident with using it.
I will have to have a look at the manual.

As you only mention 2 partitions are all the rest, on both computers, as they should be?

Response from Paragon:

Quote:

thank you for update. I have requested from developers identification of cross linked files within the program working process but got answer that stated - it is impossible to give exact paths or information about the cross linked files. Cross linked files means that files located on different Windows directories are pointing to the same record in MFT of the file system where only one file should be written to. If you use any defragging tool please try to compact MFT and this could fix the problem (sometimes it does for sure). In my opinion the problem happens when MFT is not renewed automatically and its records are not updated so deleting one file does not erase the record and system tries to write there another file which leads to cross link error. Our software checks the file location on file system and records in MFT that compares where which file must be stored. And if it finds problems it stops any operation to avoid possible problems with data.

My Reply:

Quote:

To pick up your point about defrag tools, I use Defraggler if I just wish to defrag individual files or small numbers of them but my main defrag tool is PerfectDisk 10 Pro. I run this at the end of most days especially if I have installed/updated any programs. It has a very efficient Boot Time Defrag for system files which I run when the program indicates that system file fragmentation has reached a certain point which is still only equivalent to a tiny percentage of the whole, when I mentioned on the 2nd the effects of running your last CD that was the first time that it has proved less than 100% effective.

Paragon have just come back with the suggestion that I should try compacting the MFT, unfortunately none of the software I have offers that option.

Pilgrim, concerning “cross-linked files,” I believe Paragon is refering to “hardlinks." Please see the Microsoft FSUTIL utility for further information.

Pleonasm,

Love the definition of your username.


Quote:

Pilgrim, concerning “cross-linked files,” I believe Paragon is referring to “hardlinks." Please see the Microsoft FSUTIL utility for further information.

If you look at the bottom of Reply #35 you will see that I had the same idea.
Earlier I have been looking into the question of whether it is possible to locate them but everything that I found states that you need to know the file name first. Something which according to Paragon it is impossible to find out.

I see that I have another reply waiting for me from Paragon, if I learn any more I will come back.

Pilgrim, one option to identify entries in the $MFT with multiple hardlinks is to use WinHex, but it appears to require a forensic class license (expensive).

P.S.:  I did not notice the reference to hardlinks in your prior post. 

Pleonasm,

Buying software licenses at any price is not on my list of priorities at any time in the foreseeable future. (If ever!)
Thank you for the information though.

As far as it goes what I am trying to do is not something that has got to be done nor would it be worth throwing money at (even if I had any spare) but as a practical person when something that should work fails to I try to find out why.

As I am posting again I will offer a few thoughts to see if anyone has any ideas:

It seems to me that the heart of this problem is being able to move the data at all, in particular the OS, whether or not it does align, and keep it running.

The problem of cross-linked files is not system specific nor is it OS specific.

An OS that does not have them can be moved and can be aligned.

An OS that does have them will work without any problems provided that you do not try to move it.

Ergo; something is happening/changing when you move such a system, what?

I keep coming back to the question of imaging files:
With all the imaging software that I have any knowledge of, apart from a small number of exceptions like an hibernation file a full image will copy all the data necessary for that system to run, so that it can even be restored to a different disk.
If there is an imaging program that ignores the question of alignment should it not be possible to restore any image, that was working when it was made, to an aligned partition and still have it work?
The question is if there is such a program?


Quote:

Cross linked files means that files located on different Windows directories are pointing to the same record in MFT of the file system where only one file should be written to.

The first time I looked at this it made me think of the fact that there are numerous instances within Windows as a whole where you have more than one file with the same name, this could apply to any number of them.
I can even think of one instance on my PC, explorer.exe, where I have 3 files of the same name when in fact one is different from the other two.

As I said these are just my thoughts on a subject that, as everybody who has followed this post will have realised long ago, is largely over my head.


Re your P.S. Don't worry about it, I have trouble keeping up with everything in my posts, and I wrote them.

By the way I have been thinking about the definition of your username again.
Isn't that another description of a conversation?

Just to keep anybody who is following this thread updated I have not given up on getting C: sector aligned and working, I am awaiting developments from others.

Paragon gave me a fuller explanation of cross-linked files as follows:

Quote:

The alignment operation moves the data proportionally to a new position. As you know the MFT has a table of data allocation and every file has its record. If you assign the wrong records to the wrong files you will get a total mess in the data. This is the main risk of the failure you can get resizing the partition that has errors in the file system. A simple example: in the MFT you have allocation records such as 1 2 3 4 5 6 and have data this way 1 2 4 5 6 (as you see there is no 3), now when you move the data your MFR will point as 1 - 1; 2 - 2; 3 - 4; 4 - 5; 5 - 6.

As I understand it if you attempt to align/move the partition the space that was occupied by a file that is no longer there is closed, if you create an image you copy the spaces (errors) along with the data.

Paragon are at present working on the development of a new version of their imaging software which it is intended will offer the option of partition alignment when restoring an image, regardless of how it was aligned when the image was taken.
As this will restore both data and errors in the same order as they were in originally it should overcome the problems that are caused by cross-linked files.
It will not remove them but it seems that except in very rare circumstances they do not affect the running of the system.