Super Talent’s USB 3.0 RAIDDrive

Is there any one PC component more overdue for an upgrade than the lowly USB port? The USB 2.0 spec was released a decade ago. Although its 480Mbps peak data rate might have been impressive in the era that preceded Windows XP, it’s laughably inadequate for today’s external storage devices. Thankfully, SuperSpeed USB 3.0 is upon us. The new spec promises a peak data rate of 5Gbps, which is about a tenfold increase over its predecessor.

Despite a lack of support for USB 3.0 in the latest core-logic chipsets from AMD and Intel, motherboard makers have eagerly hopped on the bandwagon. Most new high-end and mid-range mobos sport NEC’s two-port SuperSpeed controller, and we’ve seen a few laptops pop up with USB 3.0 connectivity, as well. The market doesn’t appear to be waiting for core-logic chipset support, and Intel is rumored to be working on a stand-alone SuperSpeed chip of its own.

With blue USB 3.0 ports storming the market, one problem remains: what are you going to plug into one? At the moment, the options are pretty limited. One of the few USB 3.0 devices actually available for purchase is Super Talent’s USB 3.0 RAIDDrive.

As its name implies, the RAIDDrive has a USB 3.0 interface and an internal RAID array. The array is a striped affair made up of two internal solid-state drives powered by JMicron JM612 flash controllers—the same chips Western Digital uses in its 2.5″ SiliconEdge Blue SSD. Each of the RAIDDrive’s internal SSDs is connected to an auxiliary RAID controller, which is in turn linked to a SATA-to-USB bridge that interfaces with the host system. The array is essentially hidden from the host, with the RAIDDrive presenting itself to the OS as a normal flash drive.

We’re looking at the 32GB RAIDDrive today, but 64 and 128GB capacities are also available. Super Talent says that those higher-capacity models will be a little quicker than the 32GB variant because of the flash chips involved. Lower-capacity SSDs are typically slower because they don’t have enough flash chips to fill a controller’s available memory channels. The 32GB RAIDDrive has 16 flash modules, which exactly matches the number of memory channels offered by the two JMicron flash controllers (each JM612 has eight channels). However, the 32GB drive’s 2GB flash chips don’t support interleaving, which the higher-density modules in the 64 and 128GB drives can use to speed transfer rates.

Like all too many PC accessories, the RAIDDrive is wrapped in glossy plastic. Sure, the finish looks sleek and stylish when it’s freshly polished and posing for pictures. However, thumb drives tend to get handled quite a lot, and doing so will leave the RAIDDrive a mess of greasy fingerprints and smudges.

Some notebook makers have already begun phasing out glossy plastics in favor of matte and textured finishes that wear much better in the real world, even if they’re not as shiny on the shelf. Super Talent would do well to consider a similar shift for the RAIDDrive, particularly given its $255 asking price. That’s a lot to pay for a flash drive that loses all its aesthetic appeal with frequent handling.

With 16 memory modules, two flash controllers, a RAID chip, and a SATA-to-USB bridge, the RAIDDrive packs quite a bit more silicon than the average flash drive. As one might expect, it’s also much larger, measuring 3.7″ x 1.3″ x 0.6″ (95 x 34 x 15 mm). The RAIDDrive is still easy to slip into a pocket and carry around all day, perhaps because it’s much lighter than its generous proportions might suggest. My Corsair Survivor flash drive is noticeably heavier than the Super Talent, despite occupying substantially less volume.

Of course, I’d trade the RAIDDrive’s featherweight plastic body for the Survivor’s chunky aluminum shell in a heartbeat. The Survivor also has a couple of handy mounting holes if you want to secure the drive to a lanyard or a keychain—no such luck with the RAIDDrive.

While not particularly imposing in one’s pocket, the RAIDDrive’s portly proportions can get in the way of adjacent USB ports. It’s a little difficult to see in the picture above, but the drive is obscuring access to all three of the adjacent USB ports. You can just squeeze a skinny thumb drive into the top-left port, but not without angling the RAIDDrive so it sits slightly crooked.

Desktop motherboards typically put their USB 3.0 ports right on top of each other, so the RAIDDrive is likely to block a second SuperSpeed port at the very least. If you want to measure your own system for clearance, the RAIDDrive’s body extends about 11 mm to the left and right of its USB plug and roughly 6 mm above and below it. When the drive is plugged in, there’s a 5.5-mm gap between the face of the USB port and the RAIDDrive’s body.

Performance

Obviously, Super Talent wasn’t setting out to build the world’s smallest thumb drive here. This baby’s all about performance, and we’ve run the drive through a quick handful of tests to see how it fares. We used the same system setup as in our recent SSD reviews: a Core i7-750 CPU, Radeon HD 4850 graphics card, Caviar Black 1TB hard drive, 4GB of RAM, and Gigabyte’s GA-P55A-UD7 motherboard. The motherboard uses the same NEC D720200F1 USB 3.0 controller that’s been making the rounds with other motherboard makers. We used NEC’s 1.0.18.0 USB 3.0 drivers and had the system running Windows 7 x64.

To get a sense of how much extra performance USB 3.0 provides, we tested the RAIDDrive in one of the board’s SuperSpeed ports and also in an old-school USB 2.0 port. Super Talent provides its own USB 3.0 mass storage drivers with the RAIDDrive, so we tested SuperSpeed performance with those, as well.

Before running a few benchmarks, we filled the drive to capacity and deleted its contents to ensure that all of its flash pages were in a used state. Next, we deleted the drive’s only partition and ran some synthetic transfer rate tests. Kudos to Super Talent for not polluting the RAIDDrive with hidden partitions, virtual optical drives, or other irritants.

Super Talent’s drivers give the RAIDDrive a healthy boost in HD Tach’s burst speed test. However, they don’t influence the results of the sustained transfer rate tests. Not that the RAIDDrive needs much help on that front. With sustained reads, it’s nearly five times faster when plugged into a SuperSpeed port. The drive’s write speeds are much slower with both interfaces, but you still get more than double the performance with USB 3.0.

Next, we formatted the RAIDDrive and conducted some real-world file copy tests. For these tests, I copied a folder to and from the drive containing three video files totaling 2.7GB, 1.5GB worth of MP3s, and 2.3GB of TR-related images, spreadsheets, and documents.

Although the RAIDDrive’s read speed is much lower here than it was in HD Tach, the move from USB 2.0 to 3.0 still yields more than a 2.5X increase in read performance. Writes are quicker, too, but by closer to 2.3X.

Once again, Super Talent’s drivers don’t have much of an impact on performance. There’s certainly no need to avoid the drivers, but they’re probably only necessary if you’re suffering from sluggish performance without them.

While swapping the RAIDDrive between USB ports for our transfer rate tests, I couldn’t help but notice that the drive gets a little toasty when in use. The 20 chips inside generate a fair amount of heat, and there’s nowhere else for it to go. Fortunately, although the RAIDDrive runs hotter than any other flash drive I’ve used, it doesn’t get so warm that it becomes uncomfortable to hold.

Conclusions

The RAIDDrive may not reach SuperSpeed USB 3.0’s theoretical peak transfer rates, but it’s substantially faster than anything limited to USB 2.0 speeds. This also happens to be one of the very first USB 3.0 flash drives on the market, which probably explains the relatively high $255 asking price for the 32GB model. Premium USB 2.0 flash drives with similar capacities cost well under $100, so you’re paying quite a lot for the RAIDDrive’s faster interface. You’re also getting more than double the transfer rates of USB 2.0 at its best, which goes a long way towards justifying the additional cost.

Unfortunately, the rest of the drive doesn’t. Between the easily smudged finish and bulky proportions, the RAIDDrive doesn’t feel as refined as a $250 flash drive should. Replacing the body with one that doesn’t pick up fingerprints or obscure adjacent USB ports would be a huge improvement, although the latter may be impossible given the sheer number of chips involved in the internal RAID array.

I can’t say that I’m particularly impressed with Super Talent’s two-year warranty coverage for the drive, either. The company’s own USB 2.0 flash drives are covered by a limited lifetime warranty, making the two-year deal look stingy for what should be a flagship product.

Overall, those issues leave me torn. The RAIDDrive is without a doubt the fastest USB storage solution we’ve ever tested, and it’s is very tempting on that basis alone. I just wish Super Talent had put this fancy USB 3.0 RAID array inside a nicer enclosure.

Comments closed
    • bwcbiz
    • 9 years ago

    If you have to deal with a bulky device anyway, a 2.5″ SSD coupled with an eSATA or USB 3.0 enclosure would run you under $200 for a 64 GB setup. And the performance should at least be comparable.

    I think too many companies promoting USB 3.0 have forgotten that there are other fast interfaces out there. Either that or this product is targeted toward the ignorant early-adopter with too much money on their hands. They should sell it bundled with a Monster Cable USB 3.0 extension cable.

    • d0g_p00p
    • 9 years ago

    Looks like peterc is a Super Talent shill

      • _Sigma
      • 9 years ago

      But that is usb 2.0
      See #9

        • Trymor
        • 9 years ago

        oops, can’t believe I forgot that…lol.

    • UberGerbil
    • 9 years ago

    “6 memory modules, two flash controllers, a RAID chip, and a SATA-to-USB bridge”… and still an easy-to-lose cap.

    • 5150
    • 9 years ago

    So when are flash drives going to start implementing some sort of TRIM or GC? LOL, but seriously.

    • Freon
    • 9 years ago

    “Internal RAID” naming seems to be a marketing gimmick. This seems practically synonymous with an increase in bus width when it is internal to a single device.

    Great numbers, though. It’s fast.

    • Ryu Connor
    • 9 years ago

    Random read and write performance numbers would be cool in the future.

    • Rakhmaninov3
    • 9 years ago

    Looks cool, but way too steep a price at this point.

    • Shinare
    • 9 years ago

    r[

      • Meadows
      • 9 years ago

      No, absolutely not. Never have.

      • wibeasley
      • 9 years ago

      I kinda like that –seeing a fold as a doubling, instead of an additional x1 layer. But even for people accustomed to binary formats, it would be hard to do that math in your head. Maybe it would work in situations where Log2 is easy to interpret.

      • peterc
      • 9 years ago

      The theoretical max of USB 2.0 is 480Mbps (60MBps) and the theoretical max of USB 3.0 is “10 fold”, 4.8Gbps (600MBps). But that is theoretical max speed with all the header info, not pure data.

      On a non-theoretical, “data-level”, the fastest USB 2.0 flash drives hit 32MB/s READ, so for a 10 “fold” improvement, we should be expecting 320MB/s which as you can see from the video link, this drive is really close to hitting.

      §[<http://www.youtube.com/gosupertalent#p/u/2/i5AyVdIMxqI<]§ Issues that affected speed of this test: 1. 32GB capacity. Larger capcity drives go faster (faster chips) 2. P55 based motherboards give lower USB 3.0 speeds. X58 boards are faster for this. 3. Fast Video cards can steal bandwidth from the PCI bus and give sub optimum USB 3.0 scores 4. Slower processor i7 - 720 is good but i7 - 975 is use in video 5. More RAM. Video was shot with 6GB of overclocked RAM Check out this document to learn more about performance expectations and this drive §[<http://www.supertalent.com/datasheets/USB%203.0%20RAIDDrive%20-%20Performance%20Expectations.pdf<]§ Cheers

      • bwcbiz
      • 9 years ago

      No, in this case ten-fold just means multiplied by 10. The expression pre-dates computers, so in this case too much binary thinking led you astray. 🙂

    • dpaus
    • 9 years ago

    Wouldn’t a simple USB 3.0 extension cord solve 50% of the problems with this device?

    • mwaschkowski
    • 9 years ago

    Really do need the random read/write to be part of the review for a high end component like this.

    I bought one recently to put my virtual os on (virtualbox with linux) and it worked quite well. I disagree with the review saying it didn’t get uncomfortably hot – mine was *really* hot when removing after being plugged in for a few minutes. Other than that the review was spot on.

    See here for the random read/write info:

    §[<http://www.legitreviews.com/article/1277/1/<]§ I've decided to return mine and get the enyo: §[<http://www.ocztechnology.com/products/solid-state-drives/usb-3-0-/ocz-enyo-usb-3-0-portable-solid-state-drive.html<]§ §[<http://www.ocia.net/reviews/oczenyo128/page6.shtml<]§ Edit: enyo is just as fast or faster and the 128GB is same price as the raiddrive @64GB! Its wider, but not as thick, and includes a cable - I was really worried about leaving my raiddrive plugged in on my laptop for most of the day and that it might snap off! I have my laptop on a laptop stand and if something fell on it, bam!, there would go the raiddrive, so a cable is way better for me. Shout out to the TR guys - your sites and articles are great!

    • Farting Bob
    • 9 years ago

    hmm, the real world tests, especially write tests are a little lower than i would like if im paying $255 for 32GB. Only twice as fast, costing nearly 3 times as much and being huge?

    USB 3 will be great in a few years when the chipsets support it natively and the devices can really show of its potential but until then its not worth much of a premium IMO.

      • shank15217
      • 9 years ago

      What do you mean by chipset support it natively? How would that make a difference? AMDs consumer chipsets have more than enough pcie 2.0 lanes to support a usb3 controller, not so bright on the Intel side where all north bridge pcie lanes are still 1.0 (lynnfield) but they have gone around that with a bridge chip.

    • brute
    • 9 years ago

    you dont have the OLD SCHOOL space legomen? only the star wars?

    • shank15217
    • 9 years ago

    How about random small file write and read performance? The super talent driver seems to use main memory as a second buffer and pumps up the pre-fetching (hence the faster read speed).

      • luipugs
      • 9 years ago

      i believe the faster read speed is because of the raid drivers, since that is the main purpose of a raid.

      also, all(?) flash drives have near instantaneous random access times, so i don’t think we would see much of a difference in performance between this and other flash drives even with usb 3.0 and raid drivers.

        • shank15217
        • 9 years ago

        flash drives vary wildly in random small read/write scenarios. The raid isn’t visible to the computer, its internal to the flash drive, so it wouldn’t be visible to the custom driver however the custom driver is optimized in various ways to improve this drive’s speed in certain scenarios.

      • OneArmedScissor
      • 9 years ago

      Eh…were you planning on running the Facebook database off of one of these?

        • shank15217
        • 9 years ago

        Copying a bunch small files onto an external storage isn’t a new idea.

          • Freon
          • 9 years ago

          Are you starting up 20 indepedent copy actions at once?

Pin It on Pinterest

Share This