Seagate is launching the industry's first 6Gb/s7200rpm 2TB hard drive today. The 2TB Barracuda XT contains a four platter design sporting 500GB each and rotating at 7,200 RPM. Seagate is including a new 64MB cache scheme, five-year warranty, maximum sustained transfer rate of ~140MB/s, and an estimated street price of $299. The drives should be available later this week in the retail channel.

The big news is full support for the SATA 6G interface along with auto-configure support for the older SATA 1.5 or 3Gb/s interfaces. Seagate is also launching a new version of their SeaTools software that will allow users to short stroke the drives for increased performance, at the cost of capacity.

Of course, one might be wondering where the SATA 6G controllers are right now. It turns out that Marvell is finally ready to start shipping their 88SE9123 controllers after several delays due to a variety of problems, most centering on dual controller designs planned for several motherboard updates in the next 60 days. We expect to see the first native SATA 6G implementation on a Southbridge from AMD early next year.

In the meantime, ASUS will be shipping their P7P55D Premium shortly with the Marvell 9123 chipset. This board features a PEX PLX8613 PCIe bridge chip that will convert four of the PCIe x1 lanes (250MB/s each) into two 500MB/s lanes. While still short of the maximum theoretical 600MB/s transfer speed of the SATA 6G specification, it will provide enough burst bandwidth for these first generation 6G hard drives. Expect to see Marvell 9123 equipped boards from Gigabyte in the near future.

We will be comparing the Barracuda XT 2TB drive to the latest WD Caviar Black 2TB shortly.

Gallery: Seagate XT


View All Comments

  • chartguy - Friday, October 16, 2009 - link

    The Barracuda 7200.7 was introduced back in 2003. It quoted a sustained transfer rate of 58 MB/sec. Six years later, they're only up to 138MB/sec with the Barracuda XT. That's nowhere near "Moore's Law" growth.

    Sadly, storage throughput has become the chokepoint of computing performance. Given the massive increases in areal density, it's hard to understand why transfer rates haven't gone up more, but that's the reality of it.
  • sgoyeche - Tuesday, October 20, 2009 - link

    Speaking of Moore's law, why not do what AMD has done in going from 1 processor on a die to 4 processors on a die? In the case of mechanical hard drives, why not build striping into the internal workings of a single hard drive? Have each platter of a multiple platter hard drive perform as if it were multiple single-platter hard drives, but contained in a single 3.5 inch hard drive case? This would no doubt require a new standard (SATA IV / SATA 4 perhaps), new controller and new data connections etc, but could in theory quadruple the speed of 4 platter hard drive... Is this not viable, or is it simply the musing of one who has had too many cups of coffee today? Reply
  • sgoyeche - Tuesday, October 20, 2009 - link

    Further to my previous post, a quick google seach yielded a patent "US Patent 7324301 - Striping data simultaneously across multiple platter surfaces" found at"> that describes striping within a single hard drive. It will be interesting if it is commercialized.

  • Von Matrices - Tuesday, September 22, 2009 - link

    Are there going to be any reviews of any mainstream (i.e. < $150) mechanical hard drives in the future? It's great to know about the best SSD's and top-of-the-line mechanical HDD's, but I would like to see a some reviews of drives targeted toward the lowest price per GB. The reason I say this is that there are so many 1TB drives on the market (5, 4, 3, and 2 platter models) and no other site I know of has any reviews of the 2 platter 1TB models. While the 2 platter drives would be expected to be the fastest of all the 1TB models, a comparison would be great. Reply
  • Akkuma - Tuesday, September 22, 2009 - link

    Agreed. I just got myself two Samsung Spinpoint F3 1TBs and there is next to no news, other than they were going to be released, or reviews about them. Reply
  • v12v12 - Monday, September 28, 2009 - link

    What about I'm glad larger drives are coming out, b/c I have so much media/data coming in everyday from video editing/archiving I do; my concern is defrag times. These drives take extremely long times to defrag, putting lots of stress on the drive while it's essentially not useful till it's finished.
    ___Sure I love how mechanical drives are really pushing maximal capacities, BUT the latency innovations have not caught up with the price of just larger platter densities. It takes forever to do anything with these drives regarding scanning and general maintenance. A full virus scan on a mechanical 2TB drive would be insanely long. I can't wait for solid state drives to archive mechanical drives for good!
  • chizow - Monday, September 21, 2009 - link

    What's the point of increasing external transfer speeds to SATA 6Gb/s on a 7200RPM mechanical HDD limited to 140MB/s STR and maybe 200MB/s max burst? Its like putting a 200mph speedometer on a moped. SSDs need SATA 6Gb/s, 7200RPM drives may never exceed 300MB/s. Kinda surprised this wasn't mentioned in the blog post at all, its clearly a useless checkbox feature to help sell this drive. Reply
  • afkrotch - Monday, September 21, 2009 - link

    It's called cost. One interface for all drives, versus 3 interfaces for all drives.

    Features. New NCQ with QoS for audio/video streaming. Something that I'm sure would get a lot of use when it's slapped onto a 2 TB hdd. Better power management.

    I wouldn't consider this like putting a 200 mph speedometer on a moped. I see it as putting a roof and stereo on a moped. It might not be faster, but it's got other perks.

    Gaining higher performance isn't limited in simply adding more speed. If that were the case, the P4 would still reign supreme.
  • erple2 - Monday, September 21, 2009 - link

    While it is true that there isn't that much of a benefit to adding these 600 MB/s interfaces on a 7200 RPM drive, there is still a benefit. If you break down the time it takes to get from platter to the CPU, there are essentially 4 things that have to be done:

    1. move head to relevant point on platter
    2. read data off platter into data buffer
    3. transfer data from buffer to host SATA controller
    4. transfer data from host SATA controller to some memory buffer for reading by the CPU

    The slowest (by a significant chunk) is 1 and 2 (probably 1 is slowest). They've improved step 3 which does ultimately reduce the amount of time it takes to read the data. However, the first rule of optimization is to speed up the slowest component. In this case, it's moving the head to the relevant point on the platter, and reading the data off the platter. I think that improving that will net far more advantage than just improving step 3 (which is what SATA3 does).

    It's more like putting a high performance muffler/tailpipe on a small-displacement non-turbocharged car. Sure you have all the airflow you'd ever need, but in the grand scheme of things, it just doesn't do that much at all. You're _much_ better off putting a turbo charger on the engine to boost performance. Well, various arbitrary restrictions on what you may or may not do to a car for a particular race may not allow turbos, but that's a different story than what we have here.
  • chizow - Monday, September 21, 2009 - link


    It's more like putting a high performance muffler/tailpipe on a small-displacement non-turbocharged car.

    Ya its like putting a Nitrous banger on a lawn mower. Not the ride-on type, one of those self-propelled 15cc rabbit hoppers. ;D

Log in

Don't have an account? Sign up now