No announcement yet.

Toshiba Q Series Pro 256GB Three-Drive SSD RAID Report

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Toshiba Q Series Pro 256GB Three-Drive SSD RAID Report

    Please feel free to comment about our story entitled "Toshiba Q Series Pro 256GB Three-Drive SSD RAID Report"

  • #2
    Re: Toshiba Q Series Pro 256GB Three-Drive SSD RAID Report

    Great review Jon, thanks!

    The Toshiba Q Pro is one of the few sleeper SSDs (as I call them) out there, that are overlooked by most people. Another sleeper is the Seagate 600 Pro that you mentioned. The main focus is on the one hit (benchmark) wonders like the Samsung 840 Pro. Whatever SSD wins in AS SSD is "the best of all", which of course is false, but a simple result that is easy to create on any PC becomes the ultimate test.

    I love seeing the varied results depending upon the particular test, very enlightening for those that pay attention. The truth is no SSD surpasses all others in every test, as you have clearly shown in this and other reviews. But the obsession with owning "the best" becomes a filter, we see only what we want to see.

    I'll ask for your comment on the following, although that may not be part of TweakTown's policy for their writers.

    A recent trend among some SSD enthusiasts has been to move away from using a RAID 0 array of SSDs as the OS drive, instead going back to single SSD OS drive. The main motivation for that is the recognition that 4K random read speeds do not increase in RAID 0 arrays, as sequential and high queue depth read and write speeds do. At best 4K random read speeds remain static, or usually decrease by 5% - 10%, and more so with RAID volumes of more than two SSDs. At least that is what is seen in... AS SSD, as I'm sure you know. Greater 4K random read speeds are believed to be better for booting and running an OS, which is why they have a high value in this scenario. I agree with this to some degree, but also know it is not that simple. I know that single SSDs with good (30MB/s or more) 4K read speeds, but are faster or slower by ~20%, boot an OS in the same amount of time.

    You are still an advocate of RAID 0 volumes as OS drives, as you state at the end of this review. What would you say to those that disagree, and feel that the scenario I described above is correct and the better way to go?

    Comment


    • #3
      Re: Toshiba Q Series Pro 256GB Three-Drive SSD RAID Report

      Hi Parsec, been a long time my friend. I will disagree with the assertion that 4k read is the most important metric. The operation system drive/s are writing small files over 50% of the time, so when you see that 80/20 its really not a true representation of whats going on. Here is another thing to think about. Take for example the Q Series Pro it has very low 4k read performance maybe 25MB/s yet it can literally read faster than an EVO that has a 4k read performance of 46MB/s. Than when we mix in sequential performance (loading games etc.) which is not an insignificant part of what an os drive is doing although no where near the degree that random performance is RAID is the clear winner. Take any trace based OS simulation like PCMark 8, 7, Vantage and RAID is the clear winner, in fact RAID clearly out-performs single drives in every single measurement except for (1) 4k qd1 read which is not as important as some tend to believe. Writing is more important, and drives that have superior write performance tend to be the fastest all around performers. RST RAID with Write caching can deliver up to 5 or 6 times the 4kQD 1 write performance of a single drive where it actually matters most to an os volume 4k qd1.
      Jon Coulter Storage Editor TweakTown

      Comment


      • #4
        Re: Toshiba Q Series Pro 256GB Three-Drive SSD RAID Report

        Thanks for your reply Jon, and good to see you back "in print" again (for a while now.)

        I've become skeptical of the simple, 4K QD 1 read speed performance, as the sole measure of how a SSD will perform as an OS drive. Reduced boot time being the main verification of that notion.

        I've used several well known SSDs as OS drives, Samsung 840 Pro and 840 Evo, SanDisk EX II, and a Plextor M6e, both in its PCIe slot adapter card, and in an M.2 slot (ASRock Z97 Extreme 6 board, in the so called "Ultra M.2" port. I been using UEFI booting Windows 8 installations, with and without the Windows 8 Fast Start feature. The 4K random read (QD 1) speed (AS SSD) of these SSD varies from ~34MB/s to ~40MB/s for the capacity of these SSDs I compared (250 - 256GB). That is a 20% difference between the lowest and highest 4K speed. I would note the boot time from the POST beep to the Windows desktop. IRST versions used ranged from 12.9 to 13.1.

        My boot time is virtually and subjectively identical with any of these SSDs in any scenario (~three seconds) I mentioned above. TweakTown has shown in various reviews that all the SSDs I used have significant differences in some aspects of performance, such as the PCMark 8 Consistency test (single SSD examples in the OCZ ARC 100 Series review), particularly in the Steady State 5 and Recovery phase. But all these SSDs with fresh OS installations, 90% free space, and up to 20% greater 4K QD 1 read speed, all boot an OS in the same amount of time.

        Your example of the Toshiba Q Series Pro is similar to my simple testing, given its low 4K QD 1 performance. There's more to it than one simple metric. Booting an OS occurs at low single digit queue depths (QD = 2-4), and any writing (Windows logs) that occurs will slow down the process, so writing faster will reduce boot time. But the RAID initialization extends the boot time, so any subjective gain is lost for the user. The performance available after booting with a RAID 0 array is why we want to use them.

        BTW, your multi-SSD RAID reviews inspired me to get a three SSD 240GB SanDisk Extreme II RAID 0 array (as well as a great price of $130 each) recently. Plus I had one to start with.

        Using AS SSD, I'm not losing any 4K QD 1 read speed compared to a single 240GB SanDisk Extreme II. I'm using the Z97 board mentioned above, and IRST 13.1 vs 11.2, so some differences are expected, but I'm happy with the result. 64K stripe size:

        Click image for larger version

Name:	as-ssd SD EXII x3 RAID 0 Z97 Win 8.1 MBs 8.20.2014.png
Views:	2
Size:	30.8 KB
ID:	754581

        Jon's result in AS SSD for three SanDisk Extreme II's in RAID 0: SanDisk Extreme II 240GB RAID 0 SSD Report - Benchmarks - AS SSD

        Now if I can get two more 480GB SanDisk Extreme IIs, to match the one I have now, that would be interesting. I do like both the Intel 730 and the Toshiba Q Pro RAID reviews you did, I can get the Toshiba's locally for $140 each. No wonder Intel is pushing the 730s for use in RAID 0.

        You must have gotten over 200,000 4K-64Thrd read IOPS, since both of my 4K-64Thrd IOPS were ~160,000.

        You also have Windows write caching disabled in your testing? Or write cache buffer flushing?

        You need not reply, just sayin'...

        Comment


        • #5
          Re: Toshiba Q Series Pro 256GB Three-Drive SSD RAID Report

          buffer flushing disabled, write caching enabled
          Jon Coulter Storage Editor TweakTown

          Comment

          Working...
          X