External drives: USB vs. FW800; 5400 vs. 7200 rpm. Benchmark data

Discussion in 'MacBook Pro' started by diablo2112, Apr 28, 2010.

  1. diablo2112 macrumors 6502

    Joined:
    Apr 16, 2010
    #1
    Folks,

    I recently picked up a couple of external drives, and decided to benchmark them to determine whether the interface and/or drive type made much of a difference. For those that don't want to read the details, here's the bottom line:

    If you're going to splurge on a FW800 interface, it's well worth fitting this with a 7200 rpm drive to maximize performance. Uncached sequential writes over FW800 were twice as fast on the 7200 rpm drive compared to the 5400 rpm. FW800 is a marked improvement over USB 2.0 as well.

    Full results are below.

    For background, when shopping for drives, I was interested in using the FW800 interface on the MBPs, for the simple reason it's rated almost twice as fast as USB 2.0. Some of the drives I was shopping for included 7200 rpm drives. My first thought was this was silly, since the interface limited the throughput to far-below the limits a 5400 rpm drive could produce, so why bother upgrading to 7200 rpm?

    Well, it turns out it does make a difference. I've got both a FW800 enclosure (G Drive Mini) and a USB 2.0 interface (Nexstar TX) as well as a 320GB 7200 rpm drive (Hitachi) and a 640GB 5400 rpm drive (Western Digital).

    So, I benchmarked both drives using both interfaces. Some interesting results! Turns out, the 7200 rpm drive does in fact dramatically improve performance in the FW800 interface. For sequential operations, Firewire has a dramatic improvement over USB 2.0; for random read/writes, drive speed seems to be a more important factor. And for large files, the combination of Firewire and 7200 rpm gives a pretty impressive throughput of almost 75MB/s. Full results are below. Enjoy.

    Note that the drive and interface are noted in the title bar for each drive.
     

    Attached Files:

  2. therealseebs macrumors 65816

    therealseebs

    Joined:
    Apr 14, 2010
    #2
    There's a second component you should look at when comparing drive interfaces:

    CPU load.

    One of the biggest advantages of FireWire over USB is that (in theory, anyway...) it uses less CPU time for the same amount of data transferred.
     
  3. ramone macrumors newbie

    Joined:
    May 7, 2010
    #3
    This is an interesting test and just what I was looking for. I was trying to compare FW800 + 5400rpm vs USB2 + 7200rpm.

    Thanks.
     
  4. Dale Cooper macrumors regular

    Dale Cooper

    Joined:
    Sep 20, 2005
    #4
    me too, thanks a lot!


    I'm looking for an external to keep my Aperture Library (and raw video for use with iMovie) - does anyone how an idea of how big difference hd speed and interface will make in actual use? I'm by no means a pro, so I'm just wondering if spending almost twice as much on a 7200 fw compared to a 5400 usb drive drive is really worth it. (need a portable drive, so the alternatives in the fw 7200 category is limited...).
     
  5. trondah macrumors 6502

    Joined:
    Dec 1, 2008
    #5
    The difference between 7200RPM and 5400RPM on FW800 should not be that big, something is off with this test.... What type of disks have you used?
     
  6. johnalan macrumors regular

    Joined:
    Jul 15, 2009
    Location:
    Dublin, Ireland
  7. Opstech macrumors 6502a

    Opstech

    Joined:
    Jan 5, 2010
    Location:
    California
    #7
    I use the FW800 Iomega 7200 500g drive for my movie transfers. It's super fast. Look it up
     
  8. Vyruz Reaper macrumors regular

    Joined:
    Dec 23, 2009
    #8
    +1

    can anyone confirm this. I thought that both the FIREWIRE capped the 7200RPM drives at the SAME exact performance of the 5400RPM(which firewire800 lets it work at max) so the difference should be very little
     
  9. diablo2112 thread starter macrumors 6502

    Joined:
    Apr 16, 2010
    #9
    5400 drive is the Western Digital Blue; 7200 Drive is the Hitachi.

    I didn't think there'd be much difference, either, but hey, nothing beats data and testing compared to theory.
     
  10. trondah macrumors 6502

    Joined:
    Dec 1, 2008
    #10
    Yeah well, I've seen plenty of tests showing 2.5" 5400 RPM drives to be just as fast and in some cases even faster than some 7200 RPM drives.

    Your test only shows the difference between your two drives, add some more into it if you want to actually have some useful data.
     
  11. plinden macrumors 68040

    plinden

    Joined:
    Apr 8, 2004
    #11
    What's the real world transfer speed with the different drives/connections? Copy a large (several GB) file and see how long it takes (and take a look at the Disk Activity window in Activity Monitor).

    I didn't even bother looking at your Geekbench scores. Synthetic benchmarks are meaningless.

    I can transfer at a steady 28MB/s with USB2.0 and a steady 39MB/s with FW400 on a NewerTech Ministack 2 (my Ministack 3 with FW800 is at home)
     
  12. diablo2112 thread starter macrumors 6502

    Joined:
    Apr 16, 2010
    #12
    Do your own tests if you don't like mine. Honestly, I don't care if you think these results are valuable or not. I posted them as a public service here, I documented my methods and both the drives and interfaces used, and I had 2 objectives to answer: does either the drive speed or the interface type make a difference in transfer speeds using an external enclosure.

    My initial assumption was interface did make a difference (I would have assumed FW800 was faster than USB 2.0), and that drive speed would NOT make a difference (as the interface was much slower than the drive). I found - through a documented, controlled test - that drive speed did make difference in addition to interface type. This was surprising to me.

    I frankly don't care if you tie a string and 2 tin cans together to test interface speeds. Knock yourself out.
     
  13. diablo2112 thread starter macrumors 6502

    Joined:
    Apr 16, 2010
    #13
    Gee. Maybe you should have looked at the benchmarks. My 5400 rpm tests showed an average sequential write speed of 25 MB/s with USB 2.0 and 39 MB/s with FW800. Coincidence? Yeah. Meaningless results. You were right. And, I do wonder what difference FW400 vs. 800 introduces...
     
  14. fehhkk macrumors 6502a

    fehhkk

    Joined:
    Jun 11, 2009
    Location:
    Chicago, IL
    #14
    It was a matter of time before someone said "XBench is useless" lol ...

    Might be useless looking at the number itself, but what matters, is the relative increase in the numbers.
     
  15. m85476585 macrumors 65816

    Joined:
    Feb 26, 2008
    #15
    I have two 1TB Samsung 7200 RMP drives, which can reach ~35MB/sec on USB, ~77MB/sec on FW800, and >120MB/sec on eSATA. Drive speed depends on data density, so it's possible for a high density 5400RPM drive to still max out FW800. In fact, according to reviews, the 1TB Samsung ecogreen (5400RPM) drive can reach 85MB/sec. As expected, the 1.5TB model is slightly faster and can reach ~100MB/sec.

    Of course, hard drives use relatively little power compared to the rest of the system, so I don't see any reason to buy a 25% slower drive to save a watt or two when my desktop PC is using ~200 watts.
     
  16. trondah macrumors 6502

    Joined:
    Dec 1, 2008
    #16
    That's just wrong, and I'm not even going to elaborate...
     
  17. vacuumtube macrumors newbie

    Joined:
    Feb 13, 2010
    #17
    I find these results interesting and useful. Thanks OP!
     
  18. Dwalls90 macrumors 601

    Dwalls90

    Joined:
    Feb 5, 2009
    #18
    I think it's always been understood that FW800>>>USB 2.0 ... but nice to see more proof I suppose. Hell, even FW400=>=USB 2.0. And FW doesn't hog CPU usage.
     
  19. Penn Jennings macrumors 6502

    Joined:
    Apr 22, 2010
    Location:
    Michigan
    #19

    I don't about these benchmarks but my FW 800 WD 2TB drive result is:


    Disk Test 44.32

    Sequential
    Write 60.82 MB [4K]
    Write 62.18 MB [256K]
    Read 16.14 MB [4K]
    Read 72.21 MB [256K]
     

Share This Page