How much impact does a SATA3 SSD make over stock SSD?

Discussion in 'iMac' started by g-boac, Aug 23, 2011.

  1. g-boac macrumors 6502

    Joined:
    Oct 7, 2007
    #1
    Question - I have a 2011 iMac with the 2TB + SSD (factory) option and absolutely love it. I am very satisfied with the speed and performance, and am very thrilled and grateful to have welcomed this iMac into the family. Been saving since fall 2007, and it's worth it.

    Had a question - has anybody here replaced their factory SATA2 SSD with an aftermarket SATA3 SSD (e.g., OWC Mercury Extreme 6G, Vertex 3, Intel 510), and in doing so, did you notice an appreciable real-world speed gain in boot time, shutdown time, and application start time?

    I know what the benchmarks are numerically, but wondering whether there is a true, perceivable, "real world feel" difference - whether a SATA3 drive drastically unlocks the true potential of the 2011 iMac that the factory SATA2 drive keeps bottled in. For reference - when I replaced my 7200RPM HDD in my 2010 MPB with an SSD, the difference was mind-blowing. Like I got a new machine. Asking here whether there is a similar (or at least, a noticeable) jump in performance between the stock Apple Toshiba SATA2 SSD and an aftermarket SATA3 SSD.

    I'm particularly interested if any folks have experience with the OWC Mercury Extreme Pro 6G. Had an OWC Mercury (3Gbps / SATA2) in my 2010 MPB, and was very satisfied with it.

    cheers, and thanks for any insights,
    Mark
     
  2. reputationZed macrumors 65816

    Joined:
    Mar 9, 2011
    Location:
    34°55′42″N 80°44′41″W (34.
    #2
    My understanding is that unless you have one of the very early SSD's the performance boost in going from a SATA2 drive to a SATA3 drive is pretty marginal. It may be noticeable in some situations but I'd guess more often than not you would not be able to tell the difference. Whatever the real world gain its not going to be anywhere near as mind blowing as moving from a HDD. If you were going to replace your existing SSD because you needed a bigger drive than it might make sense to go with the 6GB (depending on what deals are available at the time) but otherwise I doubt its worth the cost.
     
  3. speacock macrumors member

    Joined:
    Jul 26, 2011
    Location:
    UK
    #3
    Very little I would think

    I was lucky enough to be given an STEC MACH8 SSD, these are very high IOPS drives designed for server applications, a 50GB one costs about $1200 and even that can't saturate a SATA2 bus let alone a SATA3 one.
     
  4. g-boac thread starter macrumors 6502

    Joined:
    Oct 7, 2007
    #4
    Hi speacock - thanks for your note. Point taken. Question - can you clarify for me what a high IOPS drive means or implies? I did go to STEC's specification page, and saw that the throughput for both of these drives was 90MB/s read and write. . .in which case this would indeed be much slower than SATA2 (or 3) saturation speed.

    It seems that the selling point (and cost justification) for the STEC MACH8 series is reliability, durability (many read/write cycles), and environmental robustness, as opposed to speed.

    Thanks for your insight - as well as any other than anybody may be able to offer,

    Mark
     
  5. speacock macrumors member

    Joined:
    Jul 26, 2011
    Location:
    UK
    #5
    Hi Mark,

    You're absolutely right that they're optimised for IOPS rather than raw throughput (you'd have thought one would translate directly to the other, but it seems not to). I did some benchmarking with them a while back and had two in a RAID 0 stripe. They gave consistent read and write throughput at around 200MB/s, but the IOPS were up around the 20,000 (compared with the 5,000-8,000 that I understand a standard drive offers. SSDs are massively faster IOPS-wise than spinning disk for random reads and writes, but there are still differences between the various SSDs so I'm told).

    From what I understand (I have the luxury of being able to speak to some high-end storage specialists at work, though I rarely understand what they say anymore), this means that the raw sequential read/write isn't that spectacular, but under real-world loads they perform very well. I use mine as pagefile and scratch area for Photoshop and while for most things it makes little difference, there are some things where I've seen massive improvements. I haven't done much more testing since I first got the drive, but there was one test where the time for a particular Nik Software filter that I was using regularly dropped from 80s to 6s. It was such a change that I thought something had gone wrong.

    I guess if you put a couple of those OWC drives (I've heard the Crucial M4s are good too) in a stripe, then you could quite easily flood a SATA channel when doing long sequential reads and writes, but under normal operating conditions where there are a lot more random reads and writes than sequential ones, then the IOPS performance starts to play more of a part than raw throughput.

    I guess the long and the short of it is that even a low-end SSD is much, much quicker than spinning disk, you could probably get a quicker one, but I suspect the difference between a fast SSD and a very fast SSD is much smaller than the difference between an SSD and a spinning disk.

    Hope that helps.
     

Share This Page