The Notebook Review forums were hosted by TechTarget, who shut down them down on January 31, 2022. This static read-only archive was pulled by NBR forum users between January 20 and January 31, 2022, in an effort to make sure that the valuable technical information that had been posted on the forums is preserved. For current discussions, many NBR forum users moved over to NotebookTalk.net after the shutdown.
Problems? See this thread at archive.org.

    Momentus XT 750GB Raid-0 performance degradation

    Discussion in 'Hardware Components and Aftermarket Upgrades' started by sangemaru, Apr 27, 2013.

  1. sangemaru

    sangemaru Notebook Deity

    Reputations:
    758
    Messages:
    1,551
    Likes Received:
    328
    Trophy Points:
    101
    Hi guys.
    I'm hoping someone might have some wisdom to shed on my current troubles.
    The machine in my sig is using a pair of Momentus XT 750GB's in RAID-0, 128KB stripes, 4k aligned, Windows 8 64-bit Enterprise.

    When I first installed windows and benched this RAID, i would get 300MB/s seq reads, 280MB/s seq writes, ~80MB/s 512K reads and writes, and about 8MB/s 4K and 4K QD32 reads and writes in CrystalDiskMark.
    The system's been loading slower these days and, being scared of a broken nand cache i benched the drives again.
    I'm now getting 112MB/s reads, 150MB/s writes seq, 20MB/s 512K and 1.5MB/s 4K and 4K QD32.

    This is TERRIBLE performance degradation. I have one partition of around 1.4TB, out of which around 700GB are free.

    Anyone have any ideas of why this is happening?
     
  2. trvelbug

    trvelbug Notebook Prophet

    Reputations:
    929
    Messages:
    4,007
    Likes Received:
    40
    Trophy Points:
    116
    When I had the original 500 gb momentus, I remember reading that Seagate did not cerify them for raid operations.
    I don't know if that holds true for the newer drives but it could be something you might want to check.

    Sent from a Galaxy far, far away
     
  3. sangemaru

    sangemaru Notebook Deity

    Reputations:
    758
    Messages:
    1,551
    Likes Received:
    328
    Trophy Points:
    101
    Yea, they were never certified, but people on the net have been reporting good performance and scaling overall with them.
    Mine had excellent performance as well, until about a month ago. Since then Windows has been really acting up. It won't show my power settings, it won't show my services running, and many other issues.
    Win8 fails so bad....
    I'm pretty sure it's a software thing, but I'd rather not have to reinstall the whole OS, since I have nothing to backup on.
     
  4. TANWare

    TANWare Just This Side of Senile, I think. Super Moderator

    Reputations:
    2,548
    Messages:
    9,585
    Likes Received:
    4,997
    Trophy Points:
    431
    The problem lies in you are now benching both read and writes from about the center of the magnetic platters. Data transfer here is slower than from that of the outer tracks. As you accumulate data on those outer tracks most benchmark programs will read those as being slower as well.

    This is where proper partitioning of the drive works best. In this way you can assign partitions by the speed you want for programs or data etc. Also get a good defragging program like perfect disk. Just remember without doing something now the raid will continue to get slower and slower............
     
  5. Marksman30k

    Marksman30k Notebook Deity

    Reputations:
    2,080
    Messages:
    1,068
    Likes Received:
    180
    Trophy Points:
    81
    The 300mb/s (basically 150mb/s) read is partly accelerated by the cache so it sounds right, the 280mb/s is much more consistent with how the HDD really performs sans cache. 140mb\s is really good for a 7200rpm drive with low density platters. However, as you fill up the drives, the speeds tend to drop to around 50-60mb/s (especially the low density platter models), I'm worried as to how low your read speeds are but your write speeds are quite consistent with how a filled up 2.5inch drive performs.
     
  6. namaiki

    namaiki "basically rocks" Super Moderator

    Reputations:
    3,905
    Messages:
    6,116
    Likes Received:
    89
    Trophy Points:
    216
    Did you enable or disable the write-back cache in the Intel RST settings?
     
  7. sangemaru

    sangemaru Notebook Deity

    Reputations:
    758
    Messages:
    1,551
    Likes Received:
    328
    Trophy Points:
    101
    I'd get that, but isn't benching supposed to use multiple positions on the disk to check for speeds? Also, at 50% usage I shouldn't be in the middle of the disk yet, more like around 1/3rd of the platter. But I'll take it into advisement.


    I agree the write speeds are more or less to be expected (although maybe I'd have hoped for them in the 180-200MB/s range), but the read is really disappointing.

    Write-back cache enabled.

    On an off-note, installed fancy cache to use some of that extra ram, and I can really feel the performance improvement in gaming. It's back to the performance the drive had back on fresh install, at least mostly, but I'm sacrificing 4GB of RAM for that. The drive speeds are still low though :(
     
  8. jclausius

    jclausius Notebook Virtuoso

    Reputations:
    6,160
    Messages:
    3,265
    Likes Received:
    2,573
    Trophy Points:
    231
    Let us know how it goes. Had a fellow laptop user use the Gen-1 XTs in RAID-5. The problem is the drives would spin down the platters and still server file data from data on the SSDs. This kept forcing the Intel controller to degrade the RAID volume. In the end he got rid of the RAID-5 volume due to so many problems. OTOH, I've been using them in RAID-1 for 2+ years w/ only one major issue brought about by suspending.
     
  9. TANWare

    TANWare Just This Side of Senile, I think. Super Moderator

    Reputations:
    2,548
    Messages:
    9,585
    Likes Received:
    4,997
    Trophy Points:
    431
    It counts on what you are using to benchmark the disk. A program like Crystal Disk Mark will use nthe available sectors. If all the outer sectors are filled it use slower inner sectors.............