Trim Support – RAID 0 – 2 x 840 Evo TB

News Forums Main Trim Support – RAID 0 – 2 x 840 Evo TB

This topic contains 3 replies, has 2 voices, and was last updated by  Oskar 1 month ago.

Viewing 4 posts - 1 through 4 (of 4 total)
  • Posted in: Main
  • December 9, 2016 at 7:59 pm #39339

    Hello;

    The attached photo shows some serious performance degradation for my Evo 840 TB x2 RAID-0 set in my Mid-2012 MBP – 10.10.5. I have 3 volumes on this stripe; system volume takes up 220GBs with 40GBs free space (18%); data volume takes up 1.75 TBs with 180GB free space (~10%); rescue volume takes up 30GBs with 16GBs free space.

    In both snapshots; the topmost benchmark is from a week after manual trim; the middle ones right after manual trim; the bottom ones right before manual trim. Initially I was ticked off by the dreaded SSD “stuttering” during day to day computing. I checked out the state of the SSDs through your benchmark and saw the write performance of both drives to be horrible. When manual trim cleared the issue I was happy. Then a week later; this is where I’m at. Isn’t this too much performance degradation for a week? Also I don’t really understand why there would be such striking difference between the drives themselves.

    What is going on here, do you think? In the forum elsewhere, you recommended that one shouldn’t even need to use manual trim if trim is enabled and running. And in your FAQ you do say that trim is supported for striped RAID sets. System Information and Disk Sensei are both telling me that trim is enabled and working. What have I stumbled on to?

    I don’t know if Sensei’s health monitor is working properly since it doesn’t seem to report any difference between the drives when I switch from one to the other. For what it’s worth; health is reported at 100%.

    I hope you will be able to help me with this.

    Best.

    December 9, 2016 at 8:08 pm #39340

    Great – the forum doesn’t seem to have uploaded the picture I had chosen. So here are the values, then:

    SSD1:

    Before Trim Read: 1.04 GB/s
    Before Trim Write: 45 MB/s (!!)
    Before Trim Rewrite: 59.6 MB/s (!!)

    After Trim Read: 1.06 GB/s
    After Trim Write: 1 GB/s
    After Trim Rewrite: 446.9 MB/s

    1 Week After Trim Read: 849.2 MB/s
    1 Week After Trim Write: 636.4 MB/s (? – already?)
    1 Week After Trim Rewrite: 413.5 MB/s

    SSD2:

    Before Trim Read: 864.9 MB/s
    Before Trim Write: 208.3 MB/s
    Before Trim Rewrite: 292.9 MB/s

    After Trim Read: 1.05 GB/s
    After Trim Write: 1.02 GB/s
    After Trim Rewrite: 453.8 MB/s

    1 Week After Trim Read: 214.1 MB/s (???)
    1 Week After Trim Write: 194.4 MB/s (???)
    1 Week After Trim Rewrite: 56 MB/s (???)

    So SSD 2 has seen performance reduced down to 20% of its top shape across the board already in one week!!! SSD 1 has seen less severe degradation comparatively; but that doesn’t seem normal either!

    December 9, 2016 at 8:09 pm #39341

    This is just a post to get the forum to alert me to responses – forgot to check the box before…

    December 16, 2016 at 7:23 am #39353

    Hi
    Raid setups are known to degrade a lot in speed once the drives start filling up with data. However, you should try making a couple benchmarks and averaging them (Disk Sensei will show this automatically for you) to rule out any temporary low readings as the speeds can fluctuate a lot. Hoping to improve RAID support further in upcoming versions of Disk Sensei – as you mention, features like Health is not really clear about which SSD is selected.

Viewing 4 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic.