+ Reply to Thread
Page 9 of 31
FirstFirst ... 7 8 9 10 11 19 ... LastLast
Results 121 to 135 of 462

Thread: Partition alignment importance under Windows XP (32-bit and 64-bit)..why it helps with stuttering and increases drive working life.

  1. #121

    Quote Originally Posted by darthpingoo View Post
    I've just one question :
    Why does you recommend to align the partition to 32K (= 64 sectors) as the SSD original partition is already aligned to 32K :??:
    Drives shipping today may have their partition aligned at 64 sectors but I was under the impression that drives shipped previously had partitions that started at sector 63. However, the definitive answer to that question is for someone from OCZ to respond to.

    Furthermore, even if the original partition is aligned at sector 64, the guides show you how to achieve alignment if you deleted the original partition and want to install Windows XP on the SSD.

    Quote Originally Posted by darthpingoo View Post
    It's not really clear on which value and why. It seems that the JMicron JM602 have 8 channels and 2 or 4K NAND Page Size. So 8x4K seems OK. But I don't know if it's really the reason of the 32K.
    See the following documents for details on why partition alignment helps.

    http://blogs.msdn.com/jimmymay/attachment/8999719.ashx

    http://download.microsoft.com/downlo...2180F/COR-T558

    Wendy's testing in the Apex thread seems to indicate that there might be some performance differences dependent on the chosen offset even when dealing with single drive setups. That is definitely something worth to look into further.

  2. #122
    Moderator HDCHOPPER's Avatar Users Country Flag
    Join Date
    Jan 2009
    Location
    Phoenix
    Posts
    4,422
    Mobo: Foxconn MARS P-35
    BIOS: Quatum Force G-28
    CPU: Intel E-8400
    RAM: OCZ Reaper HPC 1066
    Vid: ATI4850 crossfire
    PSU: Quattro 1000W
    HDD: OCZ Vertex 30 gig
    OS: win XP sp2

    I must have an early ocz solid ssd because mine is native 63 alingment ...also says 32 gig not 30
    I do not work for OCZ...READ STICKY'S FIRST ! ... OCZ Drives best in IDE mode for compatibility but single member raid is some times the best ... ..Read through the wiki section at the top of the forumtrouble with flashing:http://www.ocztechnologyforum.com/fo...ad.php?t=64121
    http://www.ocztechnologyforum.com/fo...ad.php?t=53832
    SSD Tweaking and Diagnostics Tools: http://www.ocztechnologyforum.com/fo...ad.php?t=51522

  3. #123

    I've been cursing myself over and over for this instllation to work...

    In my last attempt I did EXACLY what you did in your last reinstallof XP SP2 (VLK edition) - i have the same one. I ran ran the /wipe=HEAD cmd in MBRWiz and CLEAN on the disk before i created a partition using the Vista Recovery Disk, with align=32 as option.

    I ran the installation CD, Quick NTFS format - still I get "Error loading operating system"

    Though, if I run a complete NTFS format, (which obviously changes the align value to default(?)) the installation continues..

  4. #124
    OCZ Master Users Country Flag
    Join Date
    Sep 2008
    Posts
    2,327
    Mobo: OCZ DIY 17" Intel
    CPU: P8600
    RAM: 4Gb DDR3 OCZ 10066
    Vid: ATI Mobility Radeon™ HD 3800 Series
    HDD: Vertex 30Gb (waiting)
    OS: XP 32, Windows 7 x64

    Quote Originally Posted by mogie View Post
    I ran the installation CD, Quick NTFS format - still I get "Error loading operating system"
    Make, Model, and SN of SSD?

    Ditto for motherboard, bios version, etc.

    Are you using legit Microsoft sw?

  5. #125

    Asus p5Q-E @ BIOS v.1703

    OCZSSD2-1C64G 2.5"
    SN009270808004707

    BIOS settings:
    LBA Mode Supported
    Block Mode Not Supported
    PIO Mode: 4
    Async DMA MultiWord DMA-2
    Ultra DMA: Ultra DMA-5
    SMART mointorin: Supported

    All values are set to Auto, except 32Bit Data Transfer with is Enabled.

    Storage config is set to SATA as IDE. (not AHCI/RAID which will need drivers when installing XP).

    using legit original xp disk. (allways has)

  6. #126

    Quote Originally Posted by mogie View Post
    I've been cursing myself over and over for this instllation to work...

    In my last attempt I did EXACLY what you did in your last reinstallof XP SP2 (VLK edition) - i have the same one. I ran ran the /wipe=HEAD cmd in MBRWiz and CLEAN on the disk before i created a partition using the Vista Recovery Disk, with align=32 as option.

    I ran the installation CD, Quick NTFS format - still I get "Error loading operating system"

    Though, if I run a complete NTFS format, (which obviously changes the align value to default(?)) the installation continues..
    As long as you do not delete the partition and recreate it from within the Windows XP installer, your partition should be aligned even though you did a complete NTFS format (formatting does not change the alignment).

    The only reason we are recommending a quick NTFS format, is that people were reporting the opposite of your problem i.e. if they did a full NTFS format, the system would not reboot while if they did a quick NTFS format everything worked out fine.

    To check whether your partition is aligned or not when re-installation is complete, just run diskpar with the "-i" parameter.

  7. #127
    OCZ Master Users Country Flag
    Join Date
    Sep 2008
    Posts
    2,327
    Mobo: OCZ DIY 17" Intel
    CPU: P8600
    RAM: 4Gb DDR3 OCZ 10066
    Vid: ATI Mobility Radeon™ HD 3800 Series
    HDD: Vertex 30Gb (waiting)
    OS: XP 32, Windows 7 x64

    If you still have that error, go see threads in Forum about NTLDR missing, etc. that are basically the same problem.

    http://www.ocztechnologyforum.com/fo...ad.php?t=49500

    Almost all of these problems are solved with one of 3 things:

    A) new bios for MB

    B) removing everything else from the SATA on the SB except the SSD

    C) new cables
    Last edited by D111; 01-27-2009 at 01:30 PM.

  8. #128

    admin testing

    hello...I just wanted to share my experience with 2 ssd drives and alignment. I have one 60gb solid series ssd and one 30gb ssd. I went through a bunch of different alignments using everything from 64 to 2048. Everytime I formatted with 4096. I was only using diskpar.exe to align them. I also only ran them only on xp sp3 as a seperate/additional drive. I ran atto and the best results were 128 and higher. I noticed an increase from 150 read to 170 read going from 64 to 128+ and higher on BOTH. Example "diskpar -s 1", y, y, 128,....etc

    I did see the read speed go from topping 150 (64) to reaching 171(128+) on the 30gb.

    For myself I would not use 64 for any os alignment as it will slow your read speed, IMO. The write stayed the same and improved a little with 128 and higher. Now this is only with atto as a benchmark and only xp sp3 (using a single drive).

    I did not see much difference with 256, 1024, 2048 alignment. I guess some people are saying that vista uses 1024. Havnt tested that yet.

    happy testing

  9. #129

    Thanks for the info.

    Unfortunately, I will not have access to my Core drives for a couple of weeks so I cannot test other alignments and generate comparative benchmark results using my file copy tests.

    The reason I recommended alignment at 64 sectors, was because it resulted in a huge performance increase (300%) in regards to the file copy performance on my Core V2 drive (compared to the default misaligned XP partition). The Microsoft technical document ( http://blogs.msdn.com/jimmymay/attachment/8999719.ashx ) that explained why alignment improved performance did not indicate that other alignments could result in even better performance (for non-RAID disks). Therefore, I did not think of benchmarking other alignments against the 64 sector offset.

    When Wendy did her extensive tests on the Apex drive, a brief analysis showed that an offset of 128 sectors resulted in best performance for Apex if one used ATTO Disk Benchmark and PCMark Vantage for benchmarking.

    http://www.ocztechnologyforum.com/fo...1&postcount=40

    However, please note that the PCMark Vantage results do not vary that much regardless of alignment, compared to the dispersion of the ATTO results. I definitely put more trust in PCMark Vantage than ATTO as an indicator of real world performance.

  10. #130



    This output looks okay..doesn't it?

    Ran ofcourse a NTFS complete format before installation, at it seems to work out.. my Q now is just why?

  11. #131

    uploaded a video

    I cant believe the ability of winrar with video......anway....

    here I quickly changed settings and showed a simple increase from 64 to 128 using a solid series 30gb ssd....

    http://rapidshare.com/files/19036872...etter.rar.html

    as u will see I ran this with the camstudio running...
    I am able to get 171+ reads without anything running....just lettin ya know....

    hope this helps..

  12. #132

    Quote Originally Posted by mogie View Post

    This output looks okay..doesn't it?

    Ran ofcourse a NTFS complete format before installation, at it seems to work out.. my Q now is just why?
    It looks fine. Your partition is aligned at 64 sectors (32k) as it should.

    I cannot tell you why you had the boot problem as it never occurred on my system despite doing many, many, many re-installs (and rebuilds using Drive Snapshot).

  13. #133

    Quote Originally Posted by showtime33 View Post
    I cant believe the ability of winrar with video......anway....

    here I quickly changed settings and showed a simple increase from 64 to 128 using a solid series 30gb ssd....

    http://rapidshare.com/files/19036872...etter.rar.html

    as u will see I ran this with the camstudio running...
    I am able to get 171+ reads without anything running....just lettin ya know....

    hope this helps..
    Very nice video! We might need to update our recommendations to align partitions at 128 sectors (64kb).

  14. #134

    Partition alignment in Linux.

    There's plenty of advice for Windows, but none for Linux. So here goes:

    Step 1) Set up your partition manager to use 56 sectors/track. For instance:
    Code:
    fdisk -S 56 /dev/sda
    or
    cfdisk -s 56 /dev/sda
    or
    sfdisk -S 56 *other options here - if you're using sfdisk you presumably know wtf you're doing*
    or
    parted .... well, don't use parted.
    Step 2) Set up your partitions like you normally would.

    Step 3) There is no step three, you're done.

    I know linux works with this, but I'd appreciate it if someone partition a drive in such a fashion with systemrescuecd and install Windows on those partitions and let the rest of us know if it works. It would also allow Windows users to have multiple partitions without having to painstakingly align each one.

    If you're interested, read the following. Otherwise skip the rest of my post, it's just info you don't need to know but might want to know anyway.

    First thing to keep in mind: Cylinders, Heads, and Sectors do not really exist. Seriously, they don't. Take a screw driver and take apart your SSD. If you find a hard drive platter in there, let me know. However, your BIOS still speaks in terms of Cylinder+Head+Sector, so your hard drive has to support it. (even for normal platter hard drives, CHS is legacy - your hard drive will re-interpret CHS in terms that it understands.)

    What we're doing is this: Each sector is 512 bytes. Each track, by convention, consists of 63 sectors. So each track is 512*63=32256 which is not a multiple of 4096. Partitions must begin (and end) at track boundaries. Since we're defining each track as being 56 sectors, each track is 512*56=28672=4096*7 bytes, which means all partitions will begin and end at a 4096 byte alignment whether we like it or not. Not only will our sda1 partition begin and end at a 4096 byte alignment, but so will our sda2 swap partition and our sda3 / partition, and no amount of hanky-panky with making our partitions weird sizes will impact any of that.

    Proof of concept:

    First the old fashioned way.
    Code:
    root@sysresccd /root % fdisk /dev/sda
    
    The number of cylinders for this disk is set to 2610.
    There is nothing wrong with that, but this is larger than 1024,
    and could in certain setups cause problems with:
    1) software that runs at boot time (e.g., old versions of LILO)
    2) booting and partitioning software from other OSs
       (e.g., DOS FDISK, OS/2 FDISK)
    
    Command (m for help): n
    Command action
       e   extended
       p   primary partition (1-4)
    p
    Partition number (1-4): 1
    First cylinder (1-2610, default 1): 
    Using default value 1
    Last cylinder, +cylinders or +size{K,M,G} (1-2610, default 2610): +100M
    
    Command (m for help): a
    Partition number (1-4): 1
    
    Command (m for help): n
    Command action
       e   extended
       p   primary partition (1-4)
    p
    Partition number (1-4): 2
    First cylinder (15-2610, default 15): 
    Using default value 15
    Last cylinder, +cylinders or +size{K,M,G} (15-2610, default 2610): +1G
    
    Command (m for help): t
    Partition number (1-4): 2 
    Hex code (type L to list codes): 82
    Changed system type of partition 2 to 82 (Linux swap / Solaris)
    
    Command (m for help): n
    Command action
       e   extended
       p   primary partition (1-4)
    p
    Partition number (1-4): 3
    First cylinder (147-2610, default 147): 
    Using default value 147
    Last cylinder, +cylinders or +size{K,M,G} (147-2610, default 2610): 
    Using default value 2610
    
    Command (m for help): p
    
    Disk /dev/sda: 21.4 GB, 21474836480 bytes
    255 heads, 63 sectors/track, 2610 cylinders
    Units = cylinders of 16065 * 512 = 8225280 bytes
    Disk identifier: 0x000da626
    
       Device Boot      Start         End      Blocks   Id  System
    /dev/sda1   *           1          14      112423+  83  Linux
    /dev/sda2              15         146     1060290   82  Linux swap / Solaris
    /dev/sda3             147        2610    19792080   83  Linux
    
    Command (m for help): w
    The partition table has been altered!
    
    Calling ioctl() to re-read partition table.
    Syncing disks.
    root@sysresccd /root % fdisk -l -u /dev/sda
    
    Disk /dev/sda: 21.4 GB, 21474836480 bytes
    255 heads, 63 sectors/track, 2610 cylinders, total 41943040 sectors
    Units = sectors of 1 * 512 = 512 bytes
    Disk identifier: 0x000da626
    
       Device Boot      Start         End      Blocks   Id  System
    /dev/sda1   *          63      224909      112423+  83  Linux
    /dev/sda2          224910     2345489     1060290   82  Linux swap / Solaris
    /dev/sda3         2345490    41929649    19792080   83  Linux
    root@sysresccd /root %
    The first partition begins at 63*512=32256=4096*7.875, which is not an even multiple of 4096.
    The second partition begins at 224910*512=115153920=4096*28113.75, which is not an even multiple of 4096.
    The third partition begins at 2345490*512=1200890880*293186.25, which is not an even multiple of 4096.

    All three partitions will have poor performance. It is possible that you will get a partition that begins on a 4096 aligned boundary, but it's a 1/8 chance.

    Second, with 56 sectors/track:
    Code:
    root@sysresccd /root % fdisk -S 56 /dev/sda 
    
    The number of cylinders for this disk is set to 2937.
    There is nothing wrong with that, but this is larger than 1024,
    and could in certain setups cause problems with:
    1) software that runs at boot time (e.g., old versions of LILO)
    2) booting and partitioning software from other OSs
       (e.g., DOS FDISK, OS/2 FDISK)
    
    Command (m for help): n
    Command action
       e   extended
       p   primary partition (1-4)
    p
    Partition number (1-4): 1
    First cylinder (1-2937, default 1): 
    Using default value 1
    Last cylinder, +cylinders or +size{K,M,G} (1-2937, default 2937): +100M
    
    Command (m for help): a
    Partition number (1-4): 1
    
    Command (m for help): n
    Command action
       e   extended
       p   primary partition (1-4)
    p
    Partition number (1-4): 2
    First cylinder (16-2937, default 16): 
    Using default value 16
    Last cylinder, +cylinders or +size{K,M,G} (16-2937, default 2937): +1G
    
    Command (m for help): t
    Partition number (1-4): 2
    Hex code (type L to list codes): 82
    Changed system type of partition 2 to 82 (Linux swap / Solaris)
    
    Command (m for help): n
    Command action
       e   extended
       p   primary partition (1-4)
    p
    Partition number (1-4): 3
    First cylinder (164-2937, default 164): 
    Using default value 164
    Last cylinder, +cylinders or +size{K,M,G} (164-2937, default 2937): 
    Using default value 2937
    
    Command (m for help): p
    
    Disk /dev/sda: 21.4 GB, 21474836480 bytes
    255 heads, 56 sectors/track, 2937 cylinders
    Units = cylinders of 14280 * 512 = 7311360 bytes
    Disk identifier: 0x000da626
    
       Device Boot      Start         End      Blocks   Id  System
    /dev/sda1   *           1          15      107072   83  Linux
    /dev/sda2              16         163     1056720   82  Linux swap / Solaris
    /dev/sda3             164        2937    19806360   83  Linux
    
    Command (m for help): w
    The partition table has been altered!
    
    Calling ioctl() to re-read partition table.
    Syncing disks.
    root@sysresccd /root % fdisk -l -u /dev/sda
    
    Disk /dev/sda: 21.4 GB, 21474836480 bytes
    255 heads, 56 sectors/track, 2937 cylinders, total 41943040 sectors
    Units = sectors of 1 * 512 = 512 bytes
    Disk identifier: 0x000da626
    
       Device Boot      Start         End      Blocks   Id  System
    /dev/sda1   *          56      214199      107072   83  Linux
    /dev/sda2          214200     2327639     1056720   82  Linux swap / Solaris
    /dev/sda3         2327640    41940359    19806360   83  Linux
    root@sysresccd /root %
    You'll note that the commands I typed in are exactly the same, and the partition layout is similar, although the sizes are slightly tweaked. (within a few megabytes of each other, but the total disk size is the same)

    The important part is the starting points of each partition:

    The first partition begins at 56*512=28672=4096*7, which is an even multiple of 4096.
    The second partition begins at 214200*512=109670400=4096*26775, which is an even multiple of 4096.
    The third partition begins at 2327640*512=1191751680=4096*290955, which is an even multiple of 4096.

    If you don't want to take it from me, take it from Theodore Ts'o:
    Quote Originally Posted by [url]http://article.gmane.org/gmane.linux.utilities.util-linux-ng/1947[/url]
    For SSD's and HDD's that use a 4k internal sector size, being 4k aligned makes a big difference because it avoids read-modify-write cycles. We can achieve this easily if we simply use a CHS geometry of 56 sectors/track instead of 63 sectors. So, I would propose that we change the default geometry used by the partitioning tools in util-linux-ng, gparted, etc. so the default sectors is 56; furthermore, to catch those partitioning tools that use the HDIO_GETGEO ioctl, that we change the fantasy geometry generated in drivers/scsi/scsicam.c:scsicam_bios_param() and drivers/ata/libata-scsi.c to also use a 255/56 head/sector geometry.
    I'll have benchmarks, ummm, some other time. YMMV. I am not responsible for any damages etc etc.

    edit: there's a new thread at http://www.ocztechnologyforum.com/fo...ad.php?t=54379 with lots of information.
    Last edited by Pigeon; 04-10-2009 at 03:55 PM. Reason: additional clarification.

  15. #135
    Moderator HDCHOPPER's Avatar Users Country Flag
    Join Date
    Jan 2009
    Location
    Phoenix
    Posts
    4,422
    Mobo: Foxconn MARS P-35
    BIOS: Quatum Force G-28
    CPU: Intel E-8400
    RAM: OCZ Reaper HPC 1066
    Vid: ATI4850 crossfire
    PSU: Quattro 1000W
    HDD: OCZ Vertex 30 gig
    OS: win XP sp2

    Quote Originally Posted by therookie View Post
    Very nice video! We might need to update our recommendations to align partitions at 128 sectors (64kb).

    well if it's better gonna have to do it
    I do not work for OCZ...READ STICKY'S FIRST ! ... OCZ Drives best in IDE mode for compatibility but single member raid is some times the best ... ..Read through the wiki section at the top of the forumtrouble with flashing:http://www.ocztechnologyforum.com/fo...ad.php?t=64121
    http://www.ocztechnologyforum.com/fo...ad.php?t=53832
    SSD Tweaking and Diagnostics Tools: http://www.ocztechnologyforum.com/fo...ad.php?t=51522

+ Reply to Thread
Page 9 of 31
FirstFirst ... 7 8 9 10 11 19 ... LastLast

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts