divergent ssd speed test results

Discussion in 'Mac Pro' started by zephonic, Feb 4, 2016.

  1. zephonic macrumors 65816

    zephonic

    Joined:
    Feb 7, 2011
    Location:
    greater L.A. area
    #1
    This may not be very critical, but inquisitive minds want to know; I have an Samsung 850EVO (500GB) on an OWC Accelsior S PCIe card in my '09 quad.

    Trim Enabler's benchmark shows:

    trim enabler disk speed test.png


    BlackMagicDesign's Disk Speed Test shows:

    blackmagic disk speed test.png

    330MB/s or 524MB/s read speeds seems like an unusually large deviation, or are they measuring two different things? If not, who is right and who is wrong?
     
  2. Weaselboy Moderator

    Weaselboy

    Staff Member

    Joined:
    Jan 23, 2005
    Location:
    California
    #2
    I know BlackMagic uses incompressible data for its test, so that might make it read differently. Maybe try AJA for a third opinion.
     
  3. jimj740 macrumors regular

    jimj740

    Joined:
    Jul 25, 2013
    Location:
    Pining for the company that once gave us Clarus
    #3
    In general, it is very difficult to characterize device performance on OSX, since it is practically impossible to bypass the buffer cache. The blackmagic benchmark, in my personal experience, is more susceptible than most other benchmarks to caching by OS X.

    This does NOT mean that blackmagic is a bad benchmark - it's purpose is to demonstrate the performance of the system for different video rates, and the system includes both filesystem and the buffer cache.

    The root cause of your confusion is that you are trying to characterize device performance with a tool designed to characterize system performance.
     
  4. MrAverigeUser, Feb 5, 2016
    Last edited: Feb 11, 2016

    MrAverigeUser macrumors 6502a

    MrAverigeUser

    Joined:
    May 20, 2015
    Location:
    europe
    #4
    I had exactly the same problem 10 days ago.

    look here and also at following postings:

    http://forums.macrumors.com/threads/the-samsung-840-840-pro-ssd-thread.1508877/page-51#post-22470709


    I have a response of Oscar, the developer of TrimEnabler at Cindori about the irritating READ-Performance test results.
    This is indeed caused by a bug:

    https://www.cindori.org/forums/topic/bug-in-performance-test-of-trim-enabler-3-4-2-2/


    In my case (840EVO) , AJA and BLackmagic results are constantly nearly the same and performance as predicted and specified by Samsung. There is nothing wrong with the 850 Evo, it´s just a little bug of coconut battery, nothing else.
    But better install and enable TrimEnabler.
     
  5. zephonic, Feb 11, 2016
    Last edited: Feb 11, 2016

    zephonic thread starter macrumors 65816

    zephonic

    Joined:
    Feb 7, 2011
    Location:
    greater L.A. area
    #5
    Thanks for sharing that. But the link appears to be broken?
     
  6. MrAverigeUser macrumors 6502a

    MrAverigeUser

    Joined:
    May 20, 2015
    Location:
    europe
    #6

Share This Page