Gizmos Freeware Reviews  

Go Back   Gizmo's Freeware Forum > Debating Chamber > System

Reply
 
Thread Tools Display Modes
Old 11. Feb 2011, 05:02 PM   #1 (permalink)
Site Manager
 
MidnightCowboy's Avatar
 
Join Date: Aug 2008
Location: South American Banana Republic, third bunch from the left
Posts: 15,285
Default Discussion about Advanced Partitioning

The following posts have been copied across from the main site article: Best Free Partition Manager. Although informative they are too long for the comments section.

This was the original trigger:

by tousavelo (not verified) on 27. January 2011 - 23:27 (#65339)

Hello,

It will be interesting to document if those tools do allow alignment of partitions for Advanced Format.

Thank you for the great work
__________________
Buy a Hoover and prove technology sucks.
MidnightCowboy is offline   Reply With Quote
Old 11. Feb 2011, 05:03 PM   #2 (permalink)
Site Manager
 
MidnightCowboy's Avatar
 
Join Date: Aug 2008
Location: South American Banana Republic, third bunch from the left
Posts: 15,285
Default

by crank on 3. February 2011 - 15:21 (#65778)

Thank you for raising this issue, I was unaware that these disks were being sold. A little research led me to discover the 3 disks I recently purchased were indeed advanced format drives, the Western Digital Green 'EARS'. Very good performing drive, at $80/2 TB, a very good deal.

I tested MiniTool, Easeus, and GParted with one of these, I don't have one in a 32_bit machine so I can't test Paragon's just now. Only GParted handled the new format properly, starting the first partition at sector 2048, as did Windows7. Both MiniTool and Easeus still use the old standard 63. As this is not divisible by 8, the partition is misaligned, as were subsequent ones, and this leads to significant performance penalties.

I used the new Easeus, 7.0.1. In both Easeus and MiniTool, I set the cluster size to 4k, this did nothing as far as I could tell, whether default or 4k. This should have properly restricted the partitioning to minimum of 8 old sector sizes, 512 bytes, to be minimum allocation unit, but it didn't. GParted, on the other hand, I set to the 'no alignment' setting, and even though you would think this would sabotage the partitioning, GParted correctly picked a 2048 start for the first partition, and subsequent one's sizes were all divisible by 8, not so with either MiniTool or Easeus.

To do-Plan is to install a 32_bit system in one of the PCs with an EARS, test the Paragon offering, try to contact the publishers and see what they have to say about this. Again, I appreciate you bringing this to our attention, and I apologize for taking this long to reply, I am not that fast on the learning and testing. Maybe I am mis-aligned?

See the wiki and its external links for more info, also go to Western Digital for free alignment tools, though I think those only work on their disks.
http://en.wikipedia.org/wiki/Advanced_Format

http://support.wdc.com/product/downl...id=805&lang=en

2/4/2011 0646 central time Addendum:

Well, Paragon does the same, starts the 1st partition at 63, has a partition with an uneven number of sectors, all with it set for 4k clusters. I got a quick reply from Easeus, which is admirable, unfortunately it was to suggest I use the alignment tool for best performance and they will work to make the product better. Stay tuned.
__________________
Buy a Hoover and prove technology sucks.
MidnightCowboy is offline   Reply With Quote
Old 11. Feb 2011, 05:03 PM   #3 (permalink)
Site Manager
 
MidnightCowboy's Avatar
 
Join Date: Aug 2008
Location: South American Banana Republic, third bunch from the left
Posts: 15,285
Default

by - bill (not verified) on 7. February 2011 - 0:45 (#65949)

1. There is no 'proper' way to handle the new 4KB sector drives, because 1) there is no real standard for determining where to start a partition so as to cause the data clusters within a partition to fall on natural 4 KB disk boundaries (assuming that the clusters are at least 4 KB in size themselves) - while NTFS and the ext(x) Linux file systems start their cluster regions at a multiple of 4 KB from the partition's start, FAT32 cluster regions begin at an offset from the partition's start determined by the size of the partition (and hence of its FATs) and 2) there is no way for the partitioning software to know the environment in which the disk will be used (e.g., using standard Windows XP Disk Management partitioning tools on a disk partitioned using the new Vista/Win7 alignment can cause data loss, and vice versa - both acknowledged, albeit somewhat reluctantly, by Microsoft itself, as described in some detail at http://www.multibooters.co.uk).

2. Thus it is up to the user to tell any partitioning software how to partition a disk if they want something other than the behavior that is standard in the environment under which the partition manager is running. If the manager is running under XP or any earlier edition of Windows, the standard behavior is to use logical-cylinder alignment, regardless of the sector size on the disk. If running under Vista or Win7, the new standard behavior is to use 1 MB alignment, regardless of the sector size on the disk. Only if running under Linux (e.g., gparted), which appears to be more willing to be guided by existing partition alignment on a disk than either XP/earlier or Win7/Vista are, is it reasonable to expect the partition manager to be guided by the sector size on an empty disk.

3. Just how much non-aligned file clusters affect the performance of 4 KB sector disks varies with both workload and disk vendor. Read workloads are virtually unaffected. Large write workloads SHOULD be virtually unaffected. Small sequential write workloads should be virtually unaffected as long as the disk's write-back cache is enabled. Only small random write workloads should be significantly affected, and these only by a factor of about 2. However, multiple testers have found that the WD 'EARS' 2 TB (and perhaps other size) drives appear to have a brain-dead implementation of their internal facilities for dealing with unaligned writes (even good-sized ones) and can be affected much more than, e.g., the Samsung 2 TB 5400 rpm or the Seagate 2 TB 5900 rpm 4 KB sector drives.

4. The file system cluster size has, very properly, no impact whatsoever on partition alignment on the disk (as should be obvious from the fact that 4 KB file clusters have been the most common size for decades).

5. The best way to use Advanced Format (4 KB sector) disks that emulate 512-byte sector disks (as all currently do) in XP and earlier environments is to use their traditional partitioning facilities to manually align partitions (or, in the case of FAT32 partition, their internal file clusters) on 4 KB boundaries, since this will avoid potential data loss if traditional mechanisms are later used to modify the partition layout on the disk. E.g., if you want a primary NTFS (or extx) partition at the start of the disk, first create a dummy primary partition of 62 MB (8 logical cylinders when rounded up, necessary because the normal logical cylinder has a odd number of sectors in it and the first logical track of the disk is occupied by the Master Boot Record so you can't begin the partition at the very first sector on the disk), then create the partition you actually want, then delete the dummy, and the result will be a partition (and internal file clusters) aligned to 4 KB disk boundaries. If you want a FAT32 partition, first create the size you want, use EASEUS Partition Master to find the offset of its first internal data sector, and use that offset to determine where you have to start the partition in order for its first data sector to start on a 4 KB disk boundary. Logical partition alignment is complicated by the fact that each logical partition is preceded by an Extended Partition Boot Record that occupies what would otherwise be the first logical track of the partition (much as a primary partition the very start of the disk begins one logical track later than you'd expect it to due to the presence of the MBR), causing you to need to start the partition one logical cylinder later than you're otherwise expect to. This approach produces a bog-standard traditional partition layout without using any of the legal but unusual traditional alignment tweaks that not all traditional partitioning software may understand: all you're doing is carefully choosing sizes within that standard layout such that the result aligns file clusters on 4 KB disk boundaries, so no potential for confusing old-style partitioning utilities should exist (and you shouldn't be operating on the disk using any new-style partitioning facilities such as Disk Management in Vista or Win7, unless those facilities are capable of honoring the old-style layout - which in fact Vista and Win7 can be told to do by special Registry adjustments).

6. The best way to use Advanced Format disks in Vista/Win7 (and probably Linux) environments is much simpler: do nothing and they'll work just fine as long as they are never operated upon by old-style partitioning mechanisms.
__________________
Buy a Hoover and prove technology sucks.
MidnightCowboy is offline   Reply With Quote
Old 11. Feb 2011, 05:04 PM   #4 (permalink)
Site Manager
 
MidnightCowboy's Avatar
 
Join Date: Aug 2008
Location: South American Banana Republic, third bunch from the left
Posts: 15,285
Default

by crank on 9. February 2011 - 7:24 (#66077)

Thanks for the heads-up. By 'improper', I meant that used right-out-of-the-box [defaults], with a new disk, these partitioning managers mis-align advanced format drives by starting with first partitions at sector 63. The WD 'EARS' drives have a jumper arrangement to fix that for XP users, it supposedly works well with a basic partitioning, no extended, etc. Otherwise, you need to use their tool afterward to ensure alignment. You will definitely suffer a significant performance penalty otherwise, at least with the WDs.

I can't quite glean what exactly is going on with the XP/7 data loss problem, and it isn't just a partitioning problem, you lose restore points and other shadow-copy related data whenever you dual-boot with XP after 7.

Clusters only involve the data files I see, I haven't needed to think about them before, I thought the restriction worked on any writes. If that were the case, there wouldn't be a problem, but I misconstrued its applicability.

I tried the 62 MB dummy partition scheme and I ended up with a partition that started with sector 112,455, or 456/8 is an even divide, it works! I think your other ideas require a finer-grained control of the partition placement than these tools allow, maybe I don't understand?

#6: Very good, better to ignore XP, it is approaching iron-age relic status, onward thru the fog! [sorry Oats]. You must be a Brit, that 'bog-standard' bit threw me, I kept pondering what techie arcana was I missing? Google google, ah, britlander slang. Cheerio!
__________________
Buy a Hoover and prove technology sucks.
MidnightCowboy is offline   Reply With Quote
Old 11. Feb 2011, 05:04 PM   #5 (permalink)
Site Manager
 
MidnightCowboy's Avatar
 
Join Date: Aug 2008
Location: South American Banana Republic, third bunch from the left
Posts: 15,285
Default

by - bill (not verified) on 10. February 2011 - 22:31 (#66152)

Addendum:

Well, this has been interesting.

1. The two apparently identical disks for which Paragon Partition Manager 8.5 (which is sufficiently recent to support Vista and thus presumably Win7 and which I therefore haven't bothered to update; since I'm using a Win2K system to experiment with in this particular instance that shouldn't have mattered anyway) was reporting different logical geometries (255 heads for one, 16 heads for the other) both show up as having 16 logical heads in the system BIOS (possibly a nod to the ATA spec, which limits the head count to 16 even though BIOSes routinely used to accept head counts up to 256 in INT13 - or 255 to accommodate a Microsoft bug - in order to support larger disks without having to resort to logical-block access that they might not support). Swapping the SATA ports that they use or using the PPM bootable CD (so that it's not running in the Windows environment) makes no difference to what it reports for each disk. Copying a 255-head-cylinder-aligned partition from the start of the '255-head' disk to the start of the '16-head' disk resulted in PPM resizing the partition to 16-head cylinder alignment on the output side.

2. It's worth noting that the '16-head' disk had, as I alluded to earlier, previously been populated with 255-head geometry partitions (though those at the disk's start had since been deleted), so PPM had been reporting their start locations at strange boundaries (not beginning with head 1 on track 0 or 1 of a '16-head' logical cylinder). More troubling, though, was its reporting that a couple of MB of available space existed at the end of the disk and under-reporting the size of the last partition by the same amount: it appeared to be rounding down the partition end-point to a 16-head cylinder boundary, despite the fact that the partition originally extended beyond that point and might have contained data there, which definitely suggested trouble if one tried to create another partition after it.

3. Just to see if anything changed I used Seagate's Disc Wizard (version 11, actually an Acronis product) to restore a partition at the start of the '16-head' disk that had originally been backed up aligned on 255-head cylinder boundaries. Disc Wizard appeared to retain the original partition size/alignment. As best I can remember PPM 8.5 then reported that disk as having 240 heads: it seems to have been taking some kind of cue from the first partition on the disk to override any information from the BIOS.

4. Being more than somewhat disenchanted with PPM by now I tried EASEUS Partition Master. I think it reported something other than 255-head geometry initially for the '16-head' disk, but then when the partition at the start of the disk was deleted reported 255-head geometry. At least it was willing to copy a 255-head-aligned partition from the 255-head disk to the other disk without resizing it, and after that point both EPM and PPM reported both disks as having 255 heads.

5. At some point during the proceedings the first logical (NTFS) partition on the (had been '16-head') disk became unformated, despite the fact that it (and the extended partition containing it) began several GB beyond any of the earlier activity that I described above. I suspect PPM but can't be certain of exactly when that occurred.

How much of the above stemmed from Microsoft's casual screwing around with previously-stable practices I don't know, but it strongly suggests extreme caution when using current partitioning products in pre-Vista environments - even if no new-style partitions have ever been created on the disk. Making sure that your partitioning product considers the disk you're about to modify to have 255-head geometry should usually suffice, though.

(While this addendum grew out of the 4KB-sector discussion it's not specific to it.)
__________________
Buy a Hoover and prove technology sucks.
MidnightCowboy is offline   Reply With Quote
Old 11. Feb 2011, 05:05 PM   #6 (permalink)
Site Manager
 
MidnightCowboy's Avatar
 
Join Date: Aug 2008
Location: South American Banana Republic, third bunch from the left
Posts: 15,285
Default

by crank on 11. February 2011 - 5:44 (#66172)

I've barely begun to absorb your earlier response! You're hurting my head, doesn't help that as I typed 'You're' my other desktop BSODed on me, something got trashed a few days ago, runs for a minute or two, or a day or two, then either freezes or BSOD, hate it when that happens]

OK, I thought the whole cyl/head thing was irrelevant these days, and for some time? It is a legacy/compatibility issue then. Maybe there is a reason MS wants to leave all that behind. Tis a philosophical conundrum, when do you decide to ditch the buggy whips? I don't mean to be too snarky, I really like 7 and even Vista, the pros for me outweighed the headaches. BUT, MS should have tried harder [at all!] for compatibility. The VirtualXP is a nice touch and from what I have heard, it does what it is supposed to, after you cough up the Xtra $$ for Pro or Ultimate. Have you tried running your model T's in VirtualBox or other virtual environments?

My other system immediately BSOD in safe mode! That one is going to be fun, probably just go back to older image and forget about figuring out what happened.

I did my little test where I got first partitions starting on Sector 63 in System 7, that is no good, the responses I got from Paragon and Easeus were essentially unhelpful [use an alignment tool, which we will happily sell you [paragon]]. I understand it's impossible to know the intended use of a disk, but wouldn't the default for an advanced format drive in a System7 PC NOT be the XP version of a partition?

I have no idea what I was thinking in the 62MB dummy partition test. I just redid the test, using Paragon Partition Manager 11, the other was Easeus, and I have a starting partition of 128,520, which is 0 mod 8 putting it obliquely mathematically. That should be ok, no? Start with sector 0, so a partition starting at sector 63 is starting on the 64 sector, the last [8th] 512 sector of the 8th 4k sector, making it mis-aligned by that 1 512 sector. If the start sector is div by 8, then the number of sectors before it is div by 8 because of the +1 needed to correct for starting with 0.

I am being painstaking here because I am beginning to confuse myself, a short trip most days. Also because the alignment tool reports that both the 62 MB partition and the 1.819 TB partition are misaligned. [WD Align-Powered by Acronis, they have another free version that is 'powered' by Paragon, they are so similar must be derived from the same program.][And I must say, using my brand new 2 TB drive as a crash test dummy/chew toy somehow wasn't in the plans I had for it]

This won't look right here, but aligning made this:
+-----------------------------------------------------------------------------¦
¦Pos Idx Type/Name Size Boot Hide Start Sector Total Sectors DL Vol Label ¦
+--- --- --------- ---- ---- ---- -------------- -------------- -- -----------¦
¦ 1 1 07-NTFS 63M No No 63 128,457 I: New Volume ¦
¦ 2 2 07-NTFS 1.8T No No 128,520 3,906,895,545 J: New Volume ¦
+-----------------------------------------------------------------------------+

into this:

+--- --- --------- ---- ---- ---- -------------- -------------- -- -----------¦
¦ 1 1 07-NTFS 61M No No 2,048 124,928 ¦
¦ 2 2 07-NTFS 1.8T Yes No 126,976 3,906,895,872 New Volume ¦
+-----------------------------------------------------------------------------+

I should have deleted the 62 MB partition, so redoing it, and I still get the align tool reporting mis-aligned for the single partition, which is now:

+-----------------------------------------------------------------------------¦
¦Pos Idx Type/Name Size Boot Hide Start Sector Total Sectors DL Vol Label ¦
+--- --- --------- ---- ---- ---- -------------- -------------- -- -----------¦
¦ 1 2 07-NTFS 1.8T No No 128,520 3,906,895,545 J: New Volume ¦
+-----------------------------------------------------------------------------+

And after the chiropracty:

+-----------------------------------------------------------------------------¦
¦Pos Idx Type/Name Size Boot Hide Start Sector Total Sectors DL Vol Label ¦
+--- --- --------- ---- ---- ---- -------------- -------------- -- -----------¦
¦ 1 2 07-NTFS 1.8T Yes No 126,976 3,906,895,872 New Volume ¦
+-----------------------------------------------------------------------------+

Which the tool considers aligned now. And 976 is 0 mod 8. Is that some cylinder/head thing, 126,976 vs 128,520? hmmmm =256*8*62? Paragon reports a 255/63 Hd/Sec geometry. This is Cyl 8, Hd 0, sec 1, bingo. [255*8*63 It also lists Root entries:27 Sectors per boot: 8. Don't know about those, all moot for new systems.

Sorry if this is boring you, I appreciate the effort you have put in trying to clear up some issues with partitioning in windows. I wanted to detail the above in case you thought it useful how the latest partitioning packages were treating advanced format drives in a Windows 7 environment and to get any feedback you might have. I need to put in a warning in the reviews about these pitfalls. I thought GPT partitioning to be my primary worry for updating the reviews, I wasn't aware of so many compatibility issues existing now, and with stuff out for years now.

I think it wise for users to heed your warning, be very careful using products across windows versions. Or just ditch the stagecoach, try flying coach!
__________________
Buy a Hoover and prove technology sucks.
MidnightCowboy is offline   Reply With Quote
Old 11. Feb 2011, 05:06 PM   #7 (permalink)
Site Manager
 
MidnightCowboy's Avatar
 
Join Date: Aug 2008
Location: South American Banana Republic, third bunch from the left
Posts: 15,285
Default

by - bill (not verified) on 10. February 2011 - 9:50 (#66123)

As I already noted, starting the first primary partition on the disk by default at sector 63 is entirely proper when running under XP. You might well find that the same partition managers start it at sector 2048 when running under Vista or Win7 (at least that's the behavior that I'd make the default there, because that's the 'proper' default behavior in those environments).

I also noted that the WD tool and/or jumper are not the only ways to get a different alignment under XP if you want it (and in fact without knowing exactly what the WD tool does I'd be a lot more confident in the manual approach that I described: without testing its output I wouldn't count on the WD tool - which may be made by Acronis - to do more than reposition the partitions as if they had been created under Win7/Vista, which could lead to problems down the road in XP environments that later modified the partition structure using traditional mechanisms).

The loss of Vista/Win7 restore points when accessing an NTFS partition with XP is entirely unrelated to the potential for data loss when modifying a disk's partition layout with Win7/Vista mechanisms when it was set up using traditional (e.g., XP) mechanisms, or vice versa. The first is an internal NTFS file system problem. The second is a partition management problem unrelated to the file system involved.

Directories use clusters too - in fact, all files do, including the MFT. The MFT uses 1 KB file entries, though (at least by default), and may well access them at that granularity in pre-Vista environments. In general, while I'd assume that Vista and later environments accessed everything in multiples of 4 KB (in anticipation of the new 4 KB sector disks), earlier Windows and other environments felt free to issue small accesses at 512-byte granularity even though they ALLOCATED space in integral units of (usually 4 KB) clusters.

So if XP is accessing the MFT and small files at less-than-4-KB granularity writes will be slower in those cases even if the clusters themselves are suitably aligned (though MFT updates may be mitigated if the NTFS transaction log is sufficiently suave to gather them up to write back in batches rather than on every file update). I'd check this using Sysinternals DiskMon or ProcMon if the former were working and the latter weren't crashing my system at the moment...

I'm afraid that the 62 MB trick did not 'work' for you. Back in my pre-XP days I mostly used Partition Magic and after discovering that it didn't always get along with XP NTFS partitions started using Paragon Partition Manager, both of which round UP partition sizes to the next logical cylinder boundary. Whatever you used apparently rounded DOWN, so instead of creating an 8-cylinder partition (minus the initial track stolen for the MBR) it created a 7-cylinder one (minus the MBR track), which left you misaligned (the start sector of the following partition wants to be integrally divisible by 8).

In the course of investigating this I tried creating a dummy partition on an eSATA external drive and received another surprise: while for time immemorial (almost) virtually all logical cylinders (save for a few exceptions on things like old IBM Thinkpads) have been 16065 sectors in size (255 logical heads * 63 sectors per logical track), for some reason my test created logical cylinders with only 16 logical heads, resulting in logical cylinders with only 16*255 = 4080 sectors - a convenient number for primary partitions since you need only a 2 MB dummy partition following the MBR track to get your target primary partition 4-KB-aligned, but most unfortunate for logical partition which you can NEVER get aligned using traditional layouts). This was all the more mystifying because my system drive - the same make, model, and size as the eSATA drive - got partitioned using the normal 255-head logical geometry. Since other pre-existing partitions on the eSATA drive are aligned according to 16065-sector logical cylinders, it appears that connecting the drive after booting resulted in a different logical geometry than resulted from having it present during boot - weird...

At any rate, to get back to the NORMAL case of 16065-sector logical cylinders the traditional tools give you all the control you need to align partition start points (or internal clusters in the case of FAT32 file systems) at any granularity you want: you just have to precede the partition with however many logical cylinders it takes to achieve it (using sizes that round up or down appropriately depending upon which your partitioner of choice does).

As for your last point, you obviously agree with Microsoft (which cavalierly changed a decades-old partition layout without worrying about the consequences to older systems). Since I multi-boot the newer systems with systems going back to Win98SE my view is different: I prefer to stick with a layout with which the older system (and third-party) software is compatible and to modify the registries of Vista and Win7 to use it as well.

And no, I'm not a Brit: I just adopted the phrase because I can't think of an Americanism that so precisely captures the idea which it embodies. In this case, traditional partition definitions ALLOWED quite a few tweaks that could be used to adjust where the internal file structures wound up, but since they were virtually never used a lot of software didn't bother allowing for them, resulting in potentially destructive confusion if they encountered them. By contrast, a 'bog-standard' layout is generally processable by anything, hence a safer approach.
__________________
Buy a Hoover and prove technology sucks.
MidnightCowboy is offline   Reply With Quote
Old 12. Feb 2011, 04:58 AM   #8 (permalink)
Member
 
Join Date: Feb 2011
Posts: 2
Default

Quote:
Originally Posted by MidnightCowboy View Post
by crank on 11. February 2011 - 5:44 (#66172)

I thought the whole cyl/head thing was irrelevant these days, and for some time?
It's only relevant because a lot of software still adheres to it, not because disks need to.

Even 10-year-old versions of Partition Magic could get destructively confused (though should not have) when they were running on a system whose BIOS defaulted to one logical geometry and encountered a disk partitioned using a different logical geometry (despite the fact that the disk was accessed using logical block numbers and didn't care about cylinders, heads, and tracks).

That's still true today, as evidenced by the fact that using Vista/Win7 Disk Management to operate upon a disk partitioned using logical-cylinder-aligned partitions can result in partition loss (as can using XP Disk Management to operate upon a disk partitioned using the new Vista/Win7 partitioning approach).

Quote:
I did my little test where I got first partitions starting on Sector 63 in System 7
I think I missed that detail earlier: I thought that was the result you got when you partitioned the Advanced Format disk using XP (and that's exactly the result that you SHOULD have gotten there).

If you got that result on Win7 when creating the first partition on an empty disk, that's unexpected (though an intelligent partitioner might have provided that result if there were already OTHER partitions present that were logical-cylinder-aligned).

Quote:
wouldn't the default for an advanced format drive in a System7 PC NOT be the XP version of a partition?
Yes - but the fact that the disk was an Advanced Format drive should be irrelevant.

Quote:
I just redid the test, using Paragon Partition Manager 11, the other was Easeus, and I have a starting partition of 128,520, which is 0 mod 8 putting it obliquely mathematically. That should be ok, no?
Yes: you got the alignment you were seeking to minimize unaligned write slowdowns. EASEUS apparently just uses a different rounding approach to size than Paragon PM and Partition Magic do.

Quote:
If the start sector is div by 8, then the number of sectors before it is div by 8 because of the +1 needed to correct for starting with 0.
Exactly.

Quote:
the alignment tool reports that both the 62 MB partition and the 1.819 TB partition are misaligned. [WD Align-Powered by Acronis, they have another free version that is 'powered' by Paragon, they are so similar must be derived from the same program.][And I must say, using my brand new 2 TB drive as a crash test dummy/chew toy somehow wasn't in the plans I had for it]

This won't look right here, but aligning made this:
+-----------------------------------------------------------------------------¦
¦Pos Idx Type/Name Size Boot Hide Start Sector Total Sectors DL Vol Label ¦
+--- --- --------- ---- ---- ---- -------------- -------------- -- -----------¦
¦ 1 1 07-NTFS 63M No No 63 128,457 I: New Volume ¦
¦ 2 2 07-NTFS 1.8T No No 128,520 3,906,895,545 J: New Volume ¦
+-----------------------------------------------------------------------------+

into this:

+--- --- --------- ---- ---- ---- -------------- -------------- -- -----------¦
¦ 1 1 07-NTFS 61M No No 2,048 124,928 ¦
¦ 2 2 07-NTFS 1.8T Yes No 126,976 3,906,895,872 New Volume ¦
+-----------------------------------------------------------------------------+

I should have deleted the 62 MB partition, so redoing it, and I still get the align tool reporting mis-aligned for the single partition, which is now:

+-----------------------------------------------------------------------------¦
¦Pos Idx Type/Name Size Boot Hide Start Sector Total Sectors DL Vol Label ¦
+--- --- --------- ---- ---- ---- -------------- -------------- -- -----------¦
¦ 1 2 07-NTFS 1.8T No No 128,520 3,906,895,545 J: New Volume ¦
+-----------------------------------------------------------------------------+

And after the chiropracty:

+-----------------------------------------------------------------------------¦
¦Pos Idx Type/Name Size Boot Hide Start Sector Total Sectors DL Vol Label ¦
+--- --- --------- ---- ---- ---- -------------- -------------- -- -----------¦
¦ 1 2 07-NTFS 1.8T Yes No 126,976 3,906,895,872 New Volume ¦
+-----------------------------------------------------------------------------+
What is happening is simple: the WD alignment tool considers any partition misaligned if it doesn't comply with the new Vista/Win7 partitioning model, which aligns all partition starts and ends on 1 MiB (2048 logical 512-byte sector) boundaries (I think it even allocates a full 1 MiB for the MBR and logical partition EPBR structures, thus preserving that alignment in ALL cases whereas the old-style algorithms 'stole' a logical 63-sector 'track' for those structures from the start of the immediately-following partition). There is, however, no reason whatsoever to do this for XP NTFS partitions, because all they need to optimize small-write-request performance on Advanced Format drives is to be aligned on 4 KB (8 logical 512-byte sector) boundaries.

If the tool could be used on older (non-Advanced-Format) disks as well (can it?) one could theorize that WD had a second objective in mind: converting existing old-style-partitioned disks to work with Vista and Win7 (which could otherwise screw them up by using their new-style partitioning algorithms on the old format - a really, really stupid oversight on Microsoft's part). But my impression had been that WD was touting the tool solely for use to optimize layout on Advanced Format disks for use with XP systems that by default partition them using old-style algorithms (though that's also dangerous usage, since in that environment there's an excellent chance that later partitioning activity using the old-style mechanisms will potentially screw up the new-style format that use of the tool imposed).

So the clusterf*** continues...

Quote:
I wanted to detail the above in case you thought it useful how the latest partitioning packages were treating advanced format drives in a Windows 7 environment and to get any feedback you might have.
Save for the fact that the WD tool is (at least I thought) specifically intended for use on Advanced Format drives, the behavior of other partitioning software has (and should have, unless it incorporated the kind of 'tweaking' of old-style layouts that I've described) nothing whatsoever to do with whether a drive is Advanced Format or not. I'd be somewhat surprised if any partitioning tool (save possibly on Linux, where they seem to have a somewhat better handle on partitioning compatibility than exists in the Windows world) pays any attention at all to the drive's sector size (for that matter, IIRC at least some Advanced Format drives report a sector size of 512 bytes for compatibility with existing software...).

Quote:
I thought GPT partitioning to be my primary worry for updating the reviews
GPT partitioning is finally becoming relevant, now that disks larger than 2 TB are starting to appear (though of course RAIDs have been able to exceed 2 TB in size for a long time). Those of us still multi-booting pre-Vista systems alongside our newer systems likely won't be embracing it any time soon, though (well, I guess 64-bit XP can access it as a data-only disk, and there's no real reason why 32-bit XP couldn't as well, since Server 2003 SP1 can): 2 TB per disk just isn't that confining. and won't be until MUCH larger disks become available.
- bill is offline   Reply With Quote
Old 12. Feb 2011, 07:36 AM   #9 (permalink)
Site Manager
 
MidnightCowboy's Avatar
 
Join Date: Aug 2008
Location: South American Banana Republic, third bunch from the left
Posts: 15,285
Default

Thanks for joining us here - bill, and welcome Sorry about your forced enrollment but I hope you can appreciate the reasons behind it.

This is an important area for discussion and here in the forum the posts are recorded chronologically and don't become split by other comments, making the whole subject much easier for everyone to follow. Plus of course it saves our valuable vertical space in the main site comments and prevents others from being pushed out prematurely.
__________________
Buy a Hoover and prove technology sucks.
MidnightCowboy is offline   Reply With Quote
Old 14. Feb 2011, 06:49 AM   #10 (permalink)
Member
 
Join Date: Feb 2011
Posts: 2
Default A 3 TB drive that doesn't depend upon GPT...

Just stumbled upon the continuation of a discussion that I had earlier been part of last summer at http://www.msfn.org/board/topic/1410...87#entry934987

It seems that Seagate has a 3 TB USB drive that mimics 4 KB sectors in order to fake out NT/2K/XP/Vista/Win7 systems into supporting more than 2 TiB drives using an MBR-style partitioning format.

NT has always supported (then-SCSI) large-sector drives as data drives (the first sector-size-sensitive software that I wrote for NT was over 15 years ago) and possibly even as boot drives, and my vague recollection is that one of the goals of large-sector ATA drives was to allow them to exceed 2 TiB in size using the MBR format (before the GPT format was created as a way to extend things much more generally). Now Seagate has apparently managed to leverage this support to make their larger-than-2-TiB USB drives usable in older 32-bit Windows environments (though quite possibly not as bootable drives: existing BIOS support, at least for ATA and/or USB drives, may be less flexible in this area than NT and its successors are).
- bill is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT +1. The time now is 05:36 AM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2020, vBulletin Solutions, Inc.