Sign in with
Sign up | Sign in

Results: Tom's Hardware Storage Bench v1.0

SanDisk A110 PCIe SSD: Armed With The New M.2 Edge Connector
By

Storage Bench v1.0 (Background Info)

Our Storage Bench incorporates all of the I/O from a trace recorded over two weeks. The process of replaying this sequence to capture performance gives us a bunch of numbers that aren't really intuitive at first glance. Most idle time gets expunged, leaving only the time that each benchmarked drive was actually busy working on host commands. So, by taking the ratio of that busy time and the the amount of data exchanged during the trace, we arrive at an average data rate (in MB/s) metric we can use to compare drives.

It's not quite a perfect system. The original trace captures the TRIM command in transit, but since the trace is played on a drive without a file system, TRIM wouldn't work even if it were sent during the trace replay (which, sadly, it isn't). Still, trace testing is a great way to capture periods of actual storage activity, a great companion to synthetic testing like Iometer.

Incompressible Data and Storage Bench v1.0

Also worth noting is the fact that our trace testing pushes incompressible data through the system's buffers to the drive getting benchmarked. So, when the trace replay plays back write activity, it's writing largely incompressible data. If we run our storage bench on a SandForce-based SSD, we can monitor the SMART attributes for a bit more insight.

Mushkin Chronos Deluxe 120 GB
SMART Attributes
RAW Value Increase
#242 Host Reads (in GB)
84 GB
#241 Host Writes (in GB)
142 GB
#233 Compressed NAND Writes (in GB)
149 GB

Host reads are greatly outstripped by host writes to be sure. That's all baked into the trace. But with SandForce's inline deduplication/compression, you'd expect that the amount of information written to flash would be less than the host writes (unless the data is mostly incompressible, of course). For every 1 GB the host asked to be written, Mushkin's drive is forced to write 1.05 GB.

If our trace replay was just writing easy-to-compress zeros out of the buffer, we'd see writes to NAND as a fraction of host writes. This puts the tested drives on a more equal footing, regardless of the controller's ability to compress data on the fly.

Average Data Rate

The Storage Bench trace generates more than 140 GB worth of writes during testing. Obviously, this tends to penalize drives smaller than 180 GB and reward those with more than 256 GB of capacity.

Despite a heady advantage in the sequential performance benchmarks, the little A110 doesn't quite secure a first-place finish (even if the difference between those top four finishers is negligible at best). The 256 GB Vector actually earns top honors, though that model is comparatively very expensive. SanDisk's PCIe-connected A110 is otherwise surrounded by Samsung's compelling SSDs.

Display all 19 comments.
This thread is closed for comments
  • 0 Hide
    Mike Friesen , September 5, 2013 10:03 PM
    Awesome new stuff. Can't wait to see if this drive actually uses the full potential of the M2, and if Samsung or OCZ can one-up them.
  • 5 Hide
    cryan , September 5, 2013 10:31 PM
    Quote:
    Awesome new stuff. Can't wait to see if this drive actually uses the full potential of the M2, and if Samsung or OCZ can one-up them.


    Samsung actually has some pretty awesome M.2 PCIe action going on. We're trying to get our hands on everything, so stay tuned.


    Regards,
    Christopher Ryan

  • 0 Hide
    radiovan , September 5, 2013 10:32 PM
    It will be nice to see vendors implement the NVMe connectors in the desktop mobo's, which in turn will redefine case design, as less storage space will be required for storage. I am aware that the initial intent is to direct these at the mobile market, but desktops can benefit as well.
  • 2 Hide
    cryan , September 5, 2013 11:08 PM
    Quote:
    It will be nice to see vendors implement the NVMe connectors in the desktop mobo's, which in turn will redefine case design, as less storage space will be required for storage. I am aware that the initial intent is to direct these at the mobile market, but desktops can benefit as well.


    You'll really see NVMe take off on the desktop with the move towards SATA Express. A SSD on SATA Express will leverage NVMe and two PCIe Gen 3 lanes. Though some motherboards will (and already do) have M.2 connectors, M.2 really makes more sense in mobile applications. M.2 will only get traction on the desktop insofar as it will begin to replace mSATA. Tons of mainboards, especially smaller form factor products embrace mSATA, and moving to M.2 is a natural transition. However, M.2 drives are hard to find right now, and we really won't see a plethora of options until next year.

    Regards,
    Christopher Ryan
  • 1 Hide
    nekromobo , September 6, 2013 12:50 AM
    I got M.2 toshiba ssd in my Sony Vaio Pro 13.. review that?

    and it should have samsung M.2 in some countries..
  • 2 Hide
    CaedenV , September 6, 2013 8:59 AM
    I may no longer have motivation to upgrade my system based on CPU specs, but with DDR4, M.2, new restive storage based SSDs, and better chipset features I will still have enough reason to upgrade in a year or two.
  • 0 Hide
    jimmysmitty , September 6, 2013 9:07 AM
    Quote:
    Quote:
    It will be nice to see vendors implement the NVMe connectors in the desktop mobo's, which in turn will redefine case design, as less storage space will be required for storage. I am aware that the initial intent is to direct these at the mobile market, but desktops can benefit as well.


    You'll really see NVMe take off on the desktop with the move towards SATA Express. A SSD on SATA Express will leverage NVMe and two PCIe Gen 3 lanes. Though some motherboards will (and already do) have M.2 connectors, M.2 really makes more sense in mobile applications. M.2 will only get traction on the desktop insofar as it will begin to replace mSATA. Tons of mainboards, especially smaller form factor products embrace mSATA, and moving to M.2 is a natural transition. However, M.2 drives are hard to find right now, and we really won't see a plethora of options until next year.

    Regards,
    Christopher Ryan


    That's what I was thinking. SATA Express is going to be fast enough for now as I have used PCIe SSDs before (OCZ Revo based drive) and compared to my 520 its hard to notice a difference, especially since there are other bottlenecks stopping it from being able to utilize that bandwidth.

    This will be great for ultra portable systems though and ITX systems.
  • 2 Hide
    cryan , September 6, 2013 10:52 AM
    Quote:
    I got M.2 toshiba ssd in my Sony Vaio Pro 13.. review that?

    and it should have samsung M.2 in some countries..


    Absolutely... just send it my way and consider it done.

    Regards,
    Christopher Ryan


  • 0 Hide
    m32 , September 6, 2013 12:11 PM
    Christopher Ryan, I'll take this off of your hands.

    Regards,
    m32
  • 0 Hide
    mikeangs2004 , September 6, 2013 7:19 PM
    will there be RAID or SLI/CFX for PCIe based SSD's?
    I don't think so b/c it's already way above 6G limit.
  • 0 Hide
    markhahn , September 7, 2013 11:04 AM
    the power numbers don't make a lot of sense. I suspect either the pcie interface is being kept in a high-power mode (plls running, etc) or else your molex is supplying power to other stuff on the board. (there must be a dc-dc regulator on the auxiliary card, since molex provides only 12,5.)
  • 0 Hide
    cryan , September 7, 2013 6:04 PM
    Quote:
    the power numbers don't make a lot of sense. I suspect either the pcie interface is being kept in a high-power mode (plls running, etc) or else your molex is supplying power to other stuff on the board. (there must be a dc-dc regulator on the auxiliary card, since molex provides only 12,5.)


    This is almost certainly true. I mean, we know it uses DC to DC to step the 5v down to 3.3v. But oddly, SanDisk rates the max write power consumption at 5.5w, and I could only get it max at 3.5w. Of course, it also possesses a deep slumber state, but that's contingent on having a L1.0 PCIe endpoint, and that hasn't even been ratified yet by PCI-SIG.

    So definitely take the power consumption results with a grain of salt. There are challenges to testing this drive in this way.


    Regards,
    Christopher Ryan
  • 0 Hide
    cryan , September 7, 2013 7:16 PM
    Quote:
    will there be RAID or SLI/CFX for PCIe based SSD's?
    I don't think so b/c it's already way above 6G limit.


    If you happened to have the ability to run two+ PCIe SSDs, M.2 or otherwise, you can always soft-raid them. There are reasons why you'd want to avoid such a setup, but it'd totally work. Intel's 910 PCIe SSD shows 2 or 4 drives to the OS and then they can be soft-raided from there, for instance.


    Regards,
    Christopher Ryan

  • 0 Hide
    abbadon_34 , September 8, 2013 4:31 PM
    what command line switches were use in robocopy?
  • 0 Hide
    mikeangs2004 , September 9, 2013 5:31 PM
    Is there TRIM/garbage collection?
  • 0 Hide
    Dax corrin , September 10, 2013 11:33 AM
    Awesome stuff, wish I could afford it.
  • 0 Hide
    Blaise170 , September 15, 2013 9:44 AM
    It seems like new progress is being made in the field of NAND every day. I can't wait until you can get high capacity drives for around the same price as HDDs, but I've heard that even as late as 2017 SSD storage will only have a 33% share.
  • 0 Hide
    ryyple , October 4, 2013 12:20 AM
    I'm a complete novice to all this stuff, yet I have a few questions that maybe someone can answer.

    Are these M2 devices still considered SATA drives?
    Using a PCIe (or other) interface for the M2 how could you clone these devices?
    Can these devices then be cloned using a standard HDD duplicator?
    Is there such a thing as a M2 "adapter" that plugs into the above duplicator?

    Thanks for your response.
  • 0 Hide
    tpidner17 , November 12, 2013 7:22 PM
    Christopher,

    On page four of this review, in the maximum 4kb speed chart, it says "in MB/s [higher is better]." However, shouldn't that refer to the fact that it is measuring max IOps, rather than write speed? It seems like a type to me, so I wanted to let you know.