Closed Thread
Results 1 to 1 of 1

Thread: OCZ SSD...What you need to know in easy to understand format.

  1. #1
    SSD TIGER Tony's Avatar
    Join Date
    May 2003
    Posts
    9,176
    Mobo: various as always testing
    BIOS: various as they never work correctly
    CPU: usually AM3, but sometimes AM2+ and i7
    RAM: I test so much its hard to say
    Vid: 3870x2's...this is the one fixed variable
    PSU: PCPC 1200W/860W
    HDD: Vertex or Vertex EX
    OS: All M$ from XP forward

    OCZ SSD...What you need to know in easy to understand format.

    SSD is a reasonably new technology, although the memory used is not new what is new are the controllers used in the drives. Controller technology is moving at a furious pace with new parts available sometimes 2 months after the previous model.

    Now, there are parts I can not talk about here but I can give the basics of how SSD drives are working and what you need to know about their setup.

    The first and most important piece of info you need right from the start is SSD is NOT HDD. Spinning platter hard drives are much more forgiving on setup than SSD. SSD drives do need users to apply settings that allow them to work correctly.

    So what do you need to know part 1


    Nand IC's used in SSD's have a 4KB page file.

    Let us start with our JMicron based SSD's

    The JMicron controller MCU (602) used in Core series, Solid series and Apex (2x in raid 0) works in the following way.

    602 SSD controller is basically 8bit MCU basis. internally 16bit operation. (In / Out)
    The controller is actually 4 chips embedded in 1 package so 4 MCU's.
    Flash chip provides multi page read and write. (Actually 2page) with Interleave.
    So, 2Byte * 4way * 2 way * 2 page = 32page operation for the maximum performance solution.

    32page = 32 * 4KB = 128KB stripe.

    So keeping things simple:

    Core and Solid series drives use 1xJM602 controller, with the controller being 4 controllers in 1 package they are actually in effect running raid0 with a 128k stripe.

    Now Apex uses a separate raid controller IC on the PCB of the SSD with 2x602 controllers. So stripe size will be 2x128k so 256k stripe.

    So Core and Solid 128k
    Apex................. 256k


    What you need to know part 2


    When installing Microsoft Operating Systems the partitions on the drives do need to be aligned to suit SSD architecture.

    Default NTFS cluster size for drives 2MB and larger is 4KB.

    The default starting sector on Windows XP by default is sector 63. This has a profoundly bad effect on SSD's as this misalignment causes an increase in IO (input ouput) commands which may lead to controller pauses or stutters.

    To keep things simple, alignment values when divided by 8 must give an answer in whole numbers...integer.

    Vista and the upcoming Windows 7 do actually align partitions semi correctly. I say semi as tuning the alignment can unleash hidden speed.
    From testing installation partitions under vista alignment of 1024k and 2048k have been seen, while this alignment does work ok, it is far from being tuned for the controller used on the SSD.

    There are 2 pieces of software supplied by Microsoft that can be used to align partitions, Diskpar and Diskpart...Diskpart being an evolution of Diskpar

    Forum member theRookie has written an awesome guide explaining partition alignment here on the OCZ Forum...http://www.ocztechnologyforum.com/fo...ad.php?t=48309

    Thru extensive testing here on the forum we have concluded the following offsets work best with the following drives.

    Solid and Core series 64k or 128 sectors *** although 128k or 256 sectors may be better.
    Apex series, single drive 128k or 256 sectors ***although i do feel 256k / 512 sectors may be better.
    Apex series dual drive raid0 256k or 512 sectors, ***again 512k / 1024 sectors may be even better here


    ***Some additional thoughts

    The reason for the Apex alignment increase is that there are 2 controllers in each drive, I have applied notes expressing a further increase in offset may yield even better results. Reason for this is I feel matching internal stripe size with external offset looks to be key.

    Remember Solid and Core series have 128k internal stripe
    Remember Apex series has 256k internal stripe.

    So 2xApex drives would need 2x256k which is 512k or 1024 sectors.
    So 2xSolid drives would need 2x128K which is 256k or 512 sectors.

    Obviously matching raid controller stripe size is also most important. Results look good if you match stripe on raid to that of internal stripe on 1 drive, I do however feel matching 2x the internal stripe may be key here to allow the drives to have maximum throughput. The only issue with this line of thinking is many raid bios files...such as that for ICH9 only allow a maximum stripe of 128k or 256k, so in some cases you may not be able to apply the stripe that would allow the SSD raid array to work at full speed. Many hardware raid cards do allow stripes much larger than 128k, this may be key if using Apex drives where you could try 256 and 512k stripes or even 1024k stripes with 4 drives.

    Take note here, partition alignment with SSD is the most important factor you have to deal with when installing your new drive. Tuning the drive's alignment can add 30 to 50% more performance so failing to apply the correct alignment could ruin the experience you have with the drive.


    Microsoft have now released a PPTX explaining the importance of alignment, it is aimed at Windows 7 BUT it does make mention of Vista and XP.

    http://download.microsoft.com/downlo...T558_WH08.pptx

    I have highlighted a few choice excerpts.








    What you need to know part3


    Windows XP was in NO WAY written to work with SSD drives. It does not even make an effort to align the primary partition correctly after the master boot record (MBR). It also brought to the table a host of enhancements that Microsoft added to enhance the end user experience with regular HDD (spinning platter hard drives). The big issue here is most of these enhancements are performance killers when used with SSD drives.

    drive indexing
    prefetch
    defragment

    All NOT needed with SSD drives, there are many other optimizations which I outline here in OCZ's XP tweak guide. http://www.ocztechnologyforum.com/fo...ad.php?t=47244

    Windows Vista does make an effort to align primary partitions, so on the whole end user experience with these OS's and SSD is better...HOWEVER, Vista brough even further enhancements for regular HDD's which can slow SSD down even further.

    drive indexing
    superfetch
    prefetch
    defragment

    All NOT needed with SSD drives, again there are many other optimizations that can be made, OCZ again have a tweak guide that explains what to change and where.
    http://www.ocztechnologyforum.com/fo...ad.php?t=47212

    A short note for those wondering why prefetch and superfetch along with drive indexing are not needed.

    Normal HDD access time is in the region of 11 to 15ms. Normal SSD access time is in the region of 0.2ms. What this means is SSD has found and served data faster than regular HDD has even found it. Microsoft added these enhancements as data on HDD becomes extremely fragmented. So it places markers and indexes regular drives so it can speed up data recovery and enhance end user experience. It places regular use applications and files in special cache folders so that they can be accessed quicker. The issue is all this work it does in the background is slowing the SSD drive down due to additional writes and reads the controller is doing.

    So we need to remove HDD enhancements to allow SSD drives to have an easier time coping with the read/write requestes the OS makes of it.


    For those looking for a simpler way to apply SSD optimizations to XP or Vista forum member elpamyelhsa has created a SSD tweak application.

    You can grab the file HERE or HERE

    On the whole the main focus with all the tweaks that improve end user experience with SSD revolve around reducing the amount of random file writes the operating system makes to the SSD. With the small write cache found on the JM602 controller this makes a huge difference. However VERTEX SSD drives have a much larger inbuilt cache and a much better I/O SSD controller so do cope with the amount of random file writes the OS does make. I do stress though it is still worth the effort to follow our tweak guides even with Vertex drives as while being fast they will be slowed by the OS; so while the end user experience is good it is not as good as it could be.

    Regarding VERTEX, SSD controller information is still under NDA by OCZ so at time of publishing I am unable to expose more than I have.. I will however be testing VERTEX and will publish alignment guides etc over the coming days.

    Overall the OCZ SSD forum has a wealth of SSD related information. It is well worth a visit for the latest alignment and tweak optimization guides as well as viewing other end user experience and problem solving.

    Be sure to check the sticky section as well as regular end user threads for upto date info.
    Last edited by FZ1; 04-10-2009 at 05:41 PM. Reason: Added thread prefix

Closed Thread

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