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.

    half the performance in raid 1 array....

    Discussion in 'Alienware 18 and M18x' started by ARGH, Aug 30, 2011.

  1. ARGH

    ARGH Notebook Deity

    Reputations:
    391
    Messages:
    1,883
    Likes Received:
    24
    Trophy Points:
    56
    Using hdtune and crystaldisk as benchmkarks on my system i get half the speed if my disks are in raid 1. I made them non-raid disks and they doubled in speed. This is using alienwares raid drivers and all. Is this why raid 1 array is not offered in thei configurator page? I thought raid 1 would run as fast as single drive.....
     
  2. Flashpoint

    Flashpoint Notebook Evangelist

    Reputations:
    192
    Messages:
    419
    Likes Received:
    0
    Trophy Points:
    30
    I would more or less expect raid-1 to perform at Speed of a single drive. Assume 500MB/Sec to make it easier.

    Read performance should be roughly 2 * 500MB/s (excluding sata 3/6 issues), while Write Performance should in theory be 1 * 500MB/s - Overhead of mirroring to second drive.

    So I would expect around lets say 460-480ish on writes worse case, and 900+ on reads.

    Also: Check the speeds in Atto Download ATTO Disk Benchmark v2.34 | techPowerUp

    I have noticed discrepancies in hdtune/crystal.
     
  3. DR650SE

    DR650SE The Whiskey Barracuda

    Reputations:
    7,383
    Messages:
    8,222
    Likes Received:
    182
    Trophy Points:
    231
    In RAID 1 your system writes the same info to each disk. So it writes it twice, which explains the decrease in performance by 50%. Though it is like creating a real time backup so if one disk fails, you have an exact duplicate.
     
  4. ARGH

    ARGH Notebook Deity

    Reputations:
    391
    Messages:
    1,883
    Likes Received:
    24
    Trophy Points:
    56
    raid 1 is not stripe. it should not affect disk performance.
     
  5. Flashpoint

    Flashpoint Notebook Evangelist

    Reputations:
    192
    Messages:
    419
    Likes Received:
    0
    Trophy Points:
    30
    Raid 1 is mirror, thus it will affect performance.
     
  6. ARGH

    ARGH Notebook Deity

    Reputations:
    391
    Messages:
    1,883
    Likes Received:
    24
    Trophy Points:
    56
    no it really should not. i had a desktop raid 1 array with 1TB samsung drives and evga x58 sli board running them in onboard intel raid. their performance was not cut in half like the m18x is displaying. something is fishy. i am wondering if the dell intel drivers are at fault..
     
  7. Flashpoint

    Flashpoint Notebook Evangelist

    Reputations:
    192
    Messages:
    419
    Likes Received:
    0
    Trophy Points:
    30
    You might be running into the sata 3/6 flip flop.

    Raid 1 performance: (Identical Disks 500mb/s)

    Writes: 1 * 500mb/s - overhead.
    Reads: N * 500mb/s (where N = number of disks).

    If you're not seeing something similar to this definitely monitor the sata 3/6 of the devices.
     
  8. ARGH

    ARGH Notebook Deity

    Reputations:
    391
    Messages:
    1,883
    Likes Received:
    24
    Trophy Points:
    56
    how would i monitor this?
     
  9. Xeneize

    Xeneize Notebook Deity

    Reputations:
    824
    Messages:
    1,263
    Likes Received:
    9
    Trophy Points:
    56
    Raid 1 most definitely affects writes performance depending on the controller(shouldn't be half though) I actually agree that your mirrored drive may be running at sata2 speed, check with IRS tool
     
  10. ARGH

    ARGH Notebook Deity

    Reputations:
    391
    Messages:
    1,883
    Likes Received:
    24
    Trophy Points:
    56
    i should have clarified that i am using WD 750gb scorpio drives so they are sata 2.

    intel raid storage says both drives are at 3.0BG/s speeds so that should be sata 2.
     
  11. Flashpoint

    Flashpoint Notebook Evangelist

    Reputations:
    192
    Messages:
    419
    Likes Received:
    0
    Trophy Points:
    30
    OK that's not so bad then, that eliminates the weird sata 3/6 issues between the two ports.

    Do you have a crystal disk before and after or the numbers at least?

    Edit: Another thought, one of the sata ports might have a hardware issue. Have you tested both drives individually outside of a raid volume (however you want to call it). Just wondering if the two ports have identical performance. Port 1 was dead/had issues on mine.
     
  12. ARGH

    ARGH Notebook Deity

    Reputations:
    391
    Messages:
    1,883
    Likes Received:
    24
    Trophy Points:
    56
    Yes with raid crystaldisk benched 45 read and 87 write. In non-raid it was 97 read and write.
     
  13. Flashpoint

    Flashpoint Notebook Evangelist

    Reputations:
    192
    Messages:
    419
    Likes Received:
    0
    Trophy Points:
    30
    The write performance definitely makes sense due to overhead the read doesn't make any sense whatsoever. I would expect roughly 160ish since it will read from both drives (similar to what a stripe would do).

    Is this on each drive? or did you only bench one? ie see my previous edit.
     
  14. DR650SE

    DR650SE The Whiskey Barracuda

    Reputations:
    7,383
    Messages:
    8,222
    Likes Received:
    182
    Trophy Points:
    231
    The tests prove that it does affect disk performance on this platform. Keep in mind its software raid and not hardware raid.
     
  15. ARGH

    ARGH Notebook Deity

    Reputations:
    391
    Messages:
    1,883
    Likes Received:
    24
    Trophy Points:
    56
    doesn't matter if it's software raid. i had the same raid in the desktop evga x58 sli motherboard and it did not kill performance like this.
     
  16. Da_G

    Da_G Notebook Consultant

    Reputations:
    240
    Messages:
    279
    Likes Received:
    10
    Trophy Points:
    31
    Mmm, yeah, Software vs. Hardware RAID doesn't take a big performance hit unless you are doing RAID-5, where XOR parity calculations are done on every bit of data written to the drive. Other RAID levels should be very close in performance.

    I speak from experience, having done comparisons of a 4x SSD RAID on my desktop machine some years back, and comparing ICH9 vs. an Areca RAID controller with a hardware parity engine. In some rare cases the ICH9 software RAID actually managed to outpace the hardware RAID, most likely due to software RAID having access to the full complement of system RAM for use as a cache, while the hardware RAID has a dedicated DIMM for cache operations, limiting it's size.
     
  17. Flashpoint

    Flashpoint Notebook Evangelist

    Reputations:
    192
    Messages:
    419
    Likes Received:
    0
    Trophy Points:
    30
    Here's what I would do
    (1) Test both disk(s) in crystal disk to rule out drive differences (internally if the same skip (2))
    (2) If you have access to a usb 3.0 enclosure test both drives in that (if both are the same it will be a sata port/mb issue).
    (3) Check raid drivers, you said you have intel RST, you using an older, newest, somewhere in between version of the raid driver?
    (4) Bios update, if running on an older bios (not sure if you're running A03 etc).