nMP Disk Speed Test

Discussion in 'Mac Pro' started by rasputin666, Feb 2, 2014.

  1. rasputin666 macrumors member

    rasputin666

    Joined:
    Mar 1, 2009
    #1
    Just received my nMP this weekend. i had a feeling my thunderbolt drive seem slow. here's the picture breakdown.

    1. machd is my internal 256GB in my nMP.

    2. iTunes is my external Qx2 4TB raid box connected eSata to a Lacie eSata hub then thunderbolt to my nMP---blah

    3. iTunes2 is my external Seagate 3TB Desktop drive attached via a Seagate Thunderbolt adapter (STAE129)---disaster

    now the adapter came with a USB stick but i think i installed that driver correctly, i believe it was to make sure the adapter sees large drives, and my nMP sees the 3TB. Any idea what the speed issue could be?
     

    Attached Files:

  2. Jacksonc macrumors 6502

    Jacksonc

    Joined:
    Dec 1, 2013
    Location:
    Jony's house
  3. analog guy macrumors 6502

    Joined:
    Mar 6, 2009
    #3
    TB sure isn't the problem. i have a 2-drive (SSD) RAID0 in a sonnet echo express iii-d enclosure getting speeds not too far off of your internal drive's performance. (see a thread i posted in the peripherals sub-forum for specifics).

    i suspect the issue is the qx2, lacie or seagate stuff. the owc enclosure i purchased for testing was horrid with the same drives in RAID0 -- worse than usb3. enclosures and controllers do matter.
     
  4. rasputin666 thread starter macrumors member

    rasputin666

    Joined:
    Mar 1, 2009
    #4
    that's what i was expecting, the Qx2 having the worse results, especially esata hub to TB to nMP but the TB is the one with the lowest scores and direct to the nMP. tomorrow i will disconnect all drives and just test the TB drive. then i'll connect another drive to that TB adapter and check the speed to rule out the drive itself.

    to the other reply, how to i check the RAM speed and how would impact the TB external read/write?
     
  5. rasputin666 thread starter macrumors member

    rasputin666

    Joined:
    Mar 1, 2009
    #5
    more tests--all tests consistent today at about 115/120 read/write. at least i am not at 44.8 write and 104 read on my TB drive!!!!!

    1. re-ran all tests, no reboot, on all day, 115/120 range 3 different tests
    2. changed TB cable and port, no sped difference from #1
    3. disconnected ALL other drives, no change from #1
    4. took that 3TB drive out of thunderbolt adapter and put that drive in USB 3 adapter---116 Write and 112 Read
    5. Took the drive that was in the USB 3 and put it into the thunderbolt adapter--99.6 Write and 97.1 Read
    6. for reference, took 3TB drive and put it in a USB2 adapter--43.3 Write and 42.2 Read

    conclusions? not sure! except my USB 3 and Thunderbolt are basically same exact speeds Read/Write at about 115/120 range. Both Thunderbolt and USB 3.0 are about 3 TIMES faster than USB 2.0 for me. Could my Seagate 3TB drives have 5400 RPM drives that are causing these numbers to be low? I know drive mechanics impact speed and SSD externals will be faster. I'm just thought I'd be more overwhelmed with USB 3 and Thunderbolt speeds since i've never seen them in action before.

    ----------

    Sorry, somewhat of a noob...how do you run a RAM disk speed test?

    ----------

    Forgot to mention one more test i ran. My iPhoto file is 15.89 GB. i dragged that into my TB drive in question and it took 4 minutes and 11 seconds to copy it TO the Thunderbolt drive from my internal nMP hard drive [256GB Stock]....i then copied it FROM my Thunderbolt drive to my internal drive and it took 2 Minutes and 35 seconds.

    I thought i was good with numbers but how do i equate those times and the Read/Write results above? Any correlation?
     
  6. VirtualRain macrumors 603

    VirtualRain

    Joined:
    Aug 1, 2008
    Location:
    Vancouver, BC
    #6
    A new 3TB Drive (with nothing on it) will do about 140MB/s... and performance will go down as the drive fills up (outer tracks are faster than inner tracks) so 115-120MB/s sounds about right to me if there's some data on there.

    16GB over 4min is an average of 64MB/s... 2:35 is 106MB/s which is reflective of your external's performance. I don't think there's anything particularly noteworthy here other than old fashioned hard drives are slow. ;)
     
  7. Jacksonc macrumors 6502

    Jacksonc

    Joined:
    Dec 1, 2013
    Location:
    Jony's house
    #7
    1. Download RAM disk creator at http://bogner.sh/wp-content/uploads/2012/12/RAMDiskCreator.app_.zip

    2. Open it and and make a 6gb RAM disk for the test.

    3. In Blackmagic Disk Speed Test, select the RAM disk and set the stress to 5gb.

    That's all. Remember not to store anything important on a RAM disk, because you'll loose all data stored on it when you eject, but it's nice for storing temporary files you don't need to keep.
     
  8. Anim macrumors 6502a

    Anim

    Joined:
    Dec 16, 2011
    Location:
    Macclesfield, UK
    #8
  9. Jacksonc macrumors 6502

    Jacksonc

    Joined:
    Dec 1, 2013
    Location:
    Jony's house
    #9
    To do a RAM disk speed test, you need a RAM disk, don't you? I can't think of a better way.
     
  10. Anim macrumors 6502a

    Anim

    Joined:
    Dec 16, 2011
    Location:
    Macclesfield, UK
    #10
    Well yes, but how is that relevent to the speed of a thunderbolt drive? Why do you want the OP to run a ram disk benchmark? How will that help him solve his TB speeds?
     
  11. rasputin666 thread starter macrumors member

    rasputin666

    Joined:
    Mar 1, 2009
    #11
    for fun, i did the ram disk test. 6GB stressed at 5GB
     

    Attached Files:

  12. Jacksonc macrumors 6502

    Jacksonc

    Joined:
    Dec 1, 2013
    Location:
    Jony's house
    #12
    Thanks! That is a lot faster than my imac!
     
  13. hfg, Feb 3, 2014
    Last edited: Feb 3, 2014

    hfg macrumors 68040

    hfg

    Joined:
    Dec 1, 2006
    Location:
    Cedar Rapids, IA. USA
    #13
    Your Qx2 interfaced with Thunderbolt seems consistent with what I am seeing from my Qx2 with 3TB disks on eSATA to a oMP 5,1. Those speeds seem reasonable for rotating hard disks in RAID-5 (4 disk).

    Your 3TB Seagate on Seagate Thunderbolt adapter seems a bit slow perhaps ... but again the Thunderbolt interface isn't going to make the disk transfer data any faster than it normally does. :)

    I also tested another Qx2 with 1TB disks using a Seagate GoFlex portable Thunderbolt Adapter with a SATA-to-eSATA cable on my iMac. Drives here were empty, while the original eSATA test unit was about 1/2 full.

    My Qx2 with 4 ea. 3TB / 1TB disks in RAID-5:
    === eSATA === === === Thunderbolt Seagate GoFlex Adapter ===

    . (small differences due to reading variances while test running)
     

    Attached Files:

  14. rasputin666 thread starter macrumors member

    rasputin666

    Joined:
    Mar 1, 2009
    #14
    The point a few have made about drive capacity vs data was missed by me until now. The Qx2 4TB raid box is eSata hub to TB has 2.5TB used of the 4TB and the 3TB on the thunderbolt adapter is at 1.9TB used of the 3TB, so their 'fullness' may also be a contributing factor. For some reason, that never dawned on me that the data placement on the platters impacts speed. Didn't ponder that before :)
     
  15. rasputin666 thread starter macrumors member

    rasputin666

    Joined:
    Mar 1, 2009
    #15
    For historical purposes i'll add that the Qx2 has 4 1TB Samsung drives Concatenated as 1 large drive. Don't really know the impact of the setup has on the speed tests.
     
  16. VirtualRain macrumors 603

    VirtualRain

    Joined:
    Aug 1, 2008
    Location:
    Vancouver, BC
    #16
    That's called a spanning array and is an odd choice since it offers the worst of both performance and risk... It only offers the performance of a single disk with the same risk as running RAID0 (all data is lost if one drive fails). I would suggest reconfiguring it for JBOD or RAID0 if it's not too painful to juggle the contents around.
     
  17. rasputin666 thread starter macrumors member

    rasputin666

    Joined:
    Mar 1, 2009
    #17
    i made that in 2008 when i knew a lot less about this stuff :)....actually your suggestion is not that difficult. the Qx2 is cloned daily to a 4TB USB3 drive, i can update the clone, blow away the Qx2, set the RAID0 up and reverse the mirror with CCC. i never revisited the way i set that up. sounds like a great little project when i am snowed in during the next 2 storms in 4 days here in the northeast! Thanks for the assist, can't wait to see if there is an impact on performance.
     
  18. VirtualRain macrumors 603

    VirtualRain

    Joined:
    Aug 1, 2008
    Location:
    Vancouver, BC
    #18
    Cool. Assuming your enclosure doesn't have any unforeseen bottlenecks, your performance should increase about four-fold! :)
     
  19. rasputin666 thread starter macrumors member

    rasputin666

    Joined:
    Mar 1, 2009
    #19
    193 Write and 242 Read, so Read was about the same, regardless of RAID and Write went from 140 to 193 when i changed from Span to RAID0
     
  20. VirtualRain macrumors 603

    VirtualRain

    Joined:
    Aug 1, 2008
    Location:
    Vancouver, BC
    #20
    That's not great at all for 4 disks in RAID0... I would complain to OWC... I would think it should perform much better than that.
     
  21. rasputin666 thread starter macrumors member

    rasputin666

    Joined:
    Mar 1, 2009
    #21
    well, it's through a Lacie eSata hub then TB to the nMP and it's 4 years old [and so are the 1TB drives inside it], so i missed that window a bit :)....I'm not thrilled with my TB speeds at all, even from my Seagate drive in my Seagate TB Adapter [STAE129]. I'm going to test a few more things and then I might have to call Apple and see what's going on. Right now that is at 63 Write and 91 Read, so something is not right.
     
  22. hfg, Feb 5, 2014
    Last edited: Feb 5, 2014

    hfg macrumors 68040

    hfg

    Joined:
    Dec 1, 2006
    Location:
    Cedar Rapids, IA. USA
    #22
    That does seem slow.

    See my post #13 above where I show 205/233 MBs in RAID-5 with an older OWC Qx2. It is probably the same vintage as yours and the disks are WD10000LSRTL 4-year old WD Black 1TB with 32MB cache. I connected the Qx2 eSATA port to a Seagate GoFlex portable Thunderbolt adapter attached to a Mac Mini for that test. The drives in the RAID-5 were empty and freshly formatted. You are running full 4-disk RAID-0 which should be somewhat faster than my 4-disk RAID-5.


    -howard
     
  23. rasputin666 thread starter macrumors member

    rasputin666

    Joined:
    Mar 1, 2009
    #23
    i have that thunderbolt adapter arriving Friday for my goflex 1TB portable and i plan on increasing my tests. right now, my TB adapter for the 3TB desktop drive tops at about 90-100 [just got 103 Write/104 Read], so something is wrong. i really hope my nMP doesn't have an issue that causes me to replace it. i have even rearranged my cables so that it's on different thunderbolt buses and i have tried daisy chain to the eSata hub, no difference. Shouldn't be possible because of the eSata hub i the middle, but my Qx2 is the fastest right now. the TB drive directly connected to my nMP is over 100 MB/s LESS! Costanza's not happy

    ----------

    Just remembered this-i had 2 Hitachi 2TB internal sata drives that i connected to the thunderbolt adapter bare. formatted them, portioned them and then ran the test, empty. 115/117...i gotta think TB should be faster than that. empty 7200 RPM drives, directly connected to nMP via Thunderbolt. #@#@
     
  24. Anim macrumors 6502a

    Anim

    Joined:
    Dec 16, 2011
    Location:
    Macclesfield, UK
    #24
  25. k-hawinkler macrumors member

    k-hawinkler

    Joined:
    Sep 14, 2011
    #25

Share This Page