Western Digital has been quiet on the performance front the past several months as they have placed an emphasis on their GreenPower family of products that we recently reviewed.  However, they have been busily working on a new line-up of Caviar SE16 drives that feature their new 320GB per-platter technology.  This type of areal density places WD in direct competition with Samsung's F1 lineup featuring 334GB per-platter sizes with similar thermal, acoustic, and power envelope specifications.
 
It does not come as a surprise that the first drive from WD to utilize the 320GB per-platter technology is the Caviar SE16 320GB WD3200AAKS.  What is surprising to us is that WD is not changing the product designation on drives that feature this new technology.  It's not really that surprising as a manufacturer will want to clear out inventory of previous product before introducing new product. We would like to see WD following Seagate, Hitachi, or Samsung by changing model numbers when there is a major switch in product technology.
 
Ordering a WD3200AAKS could land you the new drive or the older design with two 160GB platters.  Since only the part number changes in this case, the one you want is WD3200AAKS-00B3A0. For all intent purposes this means you need to find a dealer that carries OEM drives and will guarantee this particular part number is available.  It's either that or take your chances with a retail package.  WD started shipping these drives in mass at the end of January so odds are that retail kits will contain mixed stock at best.
 
We have had numerous requests to test this drive and fortunately our review samples recently arrived from WD.  However, what has not arrived yet are competing drives from Samsung (F1), Seagate, and Hitachi but those will arrive shortly.  In the meantime, we thought it would be prudent to post some early test results with this drive and provide a short synopsis of our experiences to date with Western Digital's latest and greatest.  Oh yeah, before anyone asks, WD is mum as to which drive will receive the 320GB platters next although they briefly had the specs up for a 640GB drive on their website.  Also, no new updates on the Raptor product family.
 
HDTune -
 
 
 
 
HDTach -
 
 
The average transfer rate of 87MB/s~91MB/s is exceptional in this drive class and exceeds the 73MB/s~75MB/s capabilities of the Raptor 150GB drive.  However, for reasons we are still investigating, the random access time of 16.3ms is poor compared to current desktop drives such as the Samsung HD501J that feature a class high 14.0ms random access time.  Although the performance of the drive in actual applications is not hindered greatly, it is perplexing to us why this drive has such high random access times.
 
Let's take a quick look at a few benchmarks and see how this drive compares to the Samsung SpinPoint T166 HD501LJ.  
PCMark Vantage and the wrap up....
POST A COMMENT

26 Comments

View All Comments

  • DN - Monday, February 25, 2008 - link

    Owls, do you get upset that easily? May I suggest some KY and an adult magazine? LOL. Reply
  • Owls - Monday, February 25, 2008 - link

    "For all intent purposes"

    Do you guys even proofread?
    Reply
  • Samus - Saturday, February 23, 2008 - link

    "Testing 4 of these in raid 0 and 5 in raid 5 would give me an orgasm. "


    No they wont, because multiply that horrible access time by however many drives you plan to add. Your sustained transfer will be good, but it'll take 30 seconds just to bring up a large directory of files.

    And forget extracting rar's from and to it.

    RAID arrays are good for a collection of drives with low access time and average at best sustained transfer rates. ie, Raptor's. The low access time makes up for the fact your multiplying it, and the sustained transfer rate's scale with the number of drives you add.
    Reply
  • Justin Case - Saturday, February 23, 2008 - link

    Oh boy.

    Apparently you don't understand how access times work, and you don't even understand how file systems work. I suppose the fact that you use the expression "RAID arrays" should be a clue.

    You don't need to "multiply" anything; seek commands are sent to all drives at the same time. You will have to wait for the _slowest_ one, but that's still only the (maximum) seek time for _one_ drive.

    And reading a directory means reading only a couple of meta-files (typically placed at the middle of the drive, in modern file systems, so head travel times are never more than 50%). The system doesn't scan the whole disk looking for files; it only needs to read the file table. For large (multi-sector), contiguous file tables, that means arrays can actually read them _faster_ than single drives.

    Finally, most operating systems cache large chunks of the file table, so chances are you don't have to wait at _all_. The time you need to wait when displaying large directories in Windows Explorer, for example, is the time it takes to sort the data (by name, date, size, whatever) and prepare to display it (generate the icons, build thumbnails, etc.). Actual directory access (ex., for a program trying to save a file) is virtually instant.
    Reply
  • Christobevii3 - Saturday, February 23, 2008 - link

    I don't know if he's arguing that they would be slow in raid or that i can't orgasm :( Reply
  • Justin Case - Saturday, February 23, 2008 - link

    Well, I doubt that one would vibrate enough for that, but with 5 I'm sure you'd get there, eventually. ;) Reply

Log in

Don't have an account? Sign up now