Random Read/Write Speed

The four corners of SSD performance are as follows: random read, random write, sequential read and sequential write speed. Random accesses are generally small in size, while sequential accesses tend to be larger and thus we have the four Iometer tests we use in all of our reviews.

Our first test writes 4KB in a completely random pattern over an 8GB space of the drive to simulate the sort of random access that you'd see on an OS drive (even this is more stressful than a normal desktop user would see). I perform three concurrent IOs and run the test for 3 minutes. The results reported are in average MB/s over the entire time. We use both standard pseudo randomly generated data for each write as well as fully random data to show you both the maximum and minimum performance offered by SandForce based drives in these tests. The average performance of SF drives will likely be somewhere in between the two values for each drive you see in the graphs. For an understanding of why this matters, read our original SandForce article.

Iometer - 4KB Random Write, 8GB LBA Space, QD=3

Random write performance looks extremely good on the Agility 3, even with incompressible data (at least at low queue depths). Like the original Agility, it's impossible to tell the performance difference between it and the Vertex 3 here.

Many of you have asked for random write performance at higher queue depths. What I have below is our 4KB random write test performed at a queue depth of 32 instead of 3. While the vast majority of desktop usage models experience queue depths of 0 - 5, higher depths are possible in heavy I/O (and multi-user) workloads:

Iometer - 4KB Random Write, 8GB LBA Space, QD=32

Even as we ramp up queue depth in Iometer, the Agility 3 sticks to the performance of the Vertex 3. It's only with incompressible data that we see the first hint of a performance deficit, but even that isn't much.

Iometer - 4KB Random Read, QD=3

Random read performance is unfortunately limited to 120GB Vertex 3 levels. It's unclear to me whether this is an asynchronous NAND issue or an artificial firmware cap.

Sequential Read/Write Speed

To measure sequential performance I ran a 1 minute long 128KB sequential test over the entire span of the drive at a queue depth of 1. The results reported are in average MB/s over the entire test length.

Iometer - 128KB Sequential Read

Sequential read performance is lower than the Vertex 3. The 240GB Agility 3 performs more like a 120GB Vertex 2 than its 240GB sibling.

Iometer - 128KB Sequential Write

Sequential write speed is competitive but generally not better than the Vertex 3.

Introduction AnandTech Storage Bench 2011 - Heavy Workload
POST A COMMENT

57 Comments

View All Comments

  • swaaye - Tuesday, May 24, 2011 - link

    Well I guess we're in for a few year wait yet before SSDs get cheaper. Yikes. :) Come on brilliant process engineers!!! Reply
  • jb510 - Tuesday, May 24, 2011 - link

    If I write 100 MB of compressible data to a SF drive does it use 100MB of space? Ie is it stored compressed or is that just for when it's going through the controller.

    If I compress or encrypt my entire drive on the fly (several utilities exist to do this) how does that affect performance?
    Reply
  • tecsi - Tuesday, May 24, 2011 - link

    We need to include the $135 60GB and $230 120GB Agility 3s to review. The $500 240GB is clearly a great performing product. But expensive. We need to understand what performance hits, other than capacity, we would take by going with the lower cost devices.

    Presumably the 60GB and 120GB Agility 3s will ship in much higher volume, so it important to include them in the performance charts.
    Reply
  • JasonInofuentes - Tuesday, May 24, 2011 - link

    It is logical to expect that the difference in speeds between the 240 GB Agility 3 and the 120 GB Agility 3 will mirror that of the respective Vertex 3 drives, since the structural changes are the same (two die per device vs one die per device). In real world usage, this is likely to be noticeable in some instances but not groundbreaking. Reply
  • seapeople - Tuesday, May 24, 2011 - link

    You really must not have been very confident in your AnandTech 2011 Storage Bench if you didn't think you'd have to debut it by MAY.

    Seriously...
    Reply
  • MilwaukeeMike - Wednesday, May 25, 2011 - link

    That block of text describing the Storage Bench has been pasted into a bunch of reviews. Read some others and you'll see it. I don't know when it first showed up, but it was a few months ago, maybe even last year. Reply
  • ssd123 - Tuesday, May 24, 2011 - link

    Too bad that OCZ is more interested in time-to-market than making reliable products.

    Check out their forums; there is no end of problems with their drives. I've personally experienced it with their Vertex 2 that would blow away my boot partition every 2 weeks or so. The replacement V2 drive they provided had the same problem.

    The solutions they propose to unlucky customers range from updating BIOS, changing registry settings and updating drivers. Often, none of these fixes end up making a difference.

    I love the performance of the OCZ drives, when they are working.

    Unfortunately they do not deliver on the basic requirements of storage devices: they need to read and write data reliably.

    I'm no Intel fanboy but I have to admit their SSDs are rock solid: they install like a regular drive and just work. No registry tuning, no driver issues, no bios problems.

    OCZ has a great forums and a very engaged online community that's more than willing to help; it's a shame that they are so busy...
    Reply
  • sanguy - Tuesday, May 31, 2011 - link

    We're in the same boat - we want to use Vertex 3 (and prior Vertex 2) for one of our applications that really benefits from SSD performance but the reliability and support OCZ provides is dismal.

    We're using Intel G2 and G3 drives and while not as fast as the Vertex 2/3 they have been 100% reliable with ZERO issues with several hundred of them deployed.

    So I love it when I see people bashing Intel - but I'd gladly give up a bit of performance for my storage to be non-volatile ;)

    SG
    Reply
  • tecsi - Wednesday, May 25, 2011 - link

    We need to see performance numbers of not only $500 240GB Agility 3, but also $135 60GB and $230 120GB.

    These are clearly economically options for boot/app drives in desktops.
    Reply
  • tecsi - Wednesday, May 25, 2011 - link

    We need to know how the $135 60GB and $230 120GB Agility 3s fare vis-a-vis their larger, but much more expensive, $500 240GB brother.

    Both the 60GB and 120GB should provide ample capacity for a boot/app SSD.

    This would really complete the picture.
    Reply

Log in

Don't have an account? Sign up now