Thursday, May 5, 2011
Posted by Chris Sacksteder in "Digital Home Hardware & Accessories" @ 09:00 AM
Side Note on Add-In Adapters
Many new computers have USB 3.0 built in; soon all will. Until then, a number of inexpensive add-in PCI Express adapters are available. For laptops, there is at least one ExpressCard adapter at about $35; to me, that would be more awkward and expensive than the value. But for a desktop or tower system a PCI Express card is reasonable.
I obtained a StarTech PEXUSB3S2, about $30, then saw a Syba SD-PEX20047 on MeritLine for $15, so I have two to compare. They are about the same, except the StarTech included a half-height bracket, which would be important if your computer took only half-height cards. I was surprised both had an internal L4 molex power connector. The StarTech instructions said to simply connect this to your computer's internal power supply, but mine has none. The Syba instructions said this is optional for devices needing more power (and implied a Y-connect was supplied; it was not). On starting, Windows 7 did not find a driver for the Syba card, but one was easily installed from the provided mini-CD . . . after (reluctantly) reading the printed instructions to find it. I tried the benchmark programs described below on both adapters, and did not find any differences, so only the results for the Syba card are shown.
If you go this route, buy an extension cable so you don't drive yourself nuts reaching around to the back of computer to find the ports. I saw a 3ft "Type A male/female" one for $2 online ($6 shipping). Get a blue one so you can keep it separate from your USB 2.0 cables.
Speed Tests -- Let the Race Begin!
To get a good idea of the relative performance of USB 3.0 vs. 2.0, and of the Verbatim disk, I used two disk benchmark programs, devised some file copy tests and did several disk image backups. A summary of these methods is:
- ATTO Disk BenchMark. The default options were used (particularly Direct I/O).
- CrystalDiskMark (version 3.0.1). Reports sequential and random reads and writes, and takes a long time to run, so it must be good (:-)). The default settings of 5 runs for each test and a test size of 1000MB were used.
- Small File Copy: to test copying many small files, a subset of photos was copied to and from the target disk with the ROBOCOPY command. The data consisted of 6,341 files in 218 folders for a total of 6.42 GB. This replicates a common task I do often to mirror media files between systems and make backups to store elsewhere. To test reading and writing at the same time, which is something USB 3.0 can do but 2.0 cannot, the same files were written to a second location on the target disk while a second process was launched to read the first copy from the target disk to a different drive on the test computer. I’m not sure how often you will do something like this in real life, but since USB 3.0 is full duplex I thought it would be an interesting test. These tests were repeated 3 times and the average times calculated, then converted to MB/s. The system’s file cache of 2GB is much less than the amount of data copied, and copying was to and from a different location than the previous run, so caching effects should have been minimal.
- Large File Copy: here two HD recordings, one 3.90GB and one 7.94 GB were copied to and from the target disk. This is also something real-world and a common thing to do, such as when my little brother was on CNN recently and I wanted to cut the segment out of the 2 hour WTV recording made on my media center (hey, I have to brag about it . . . plus I really do wish I had had a USB 3.0 device when I moved the file!)
- Image backup: Macrium Reflect Free Edition was used to make a backup image of the test computer’s C: drive. This is something I do often. On this test system the C: drive is pretty large because there are other things besides the OS, and the resulting image file was about 63GB.
All reported tests were done on a modest Dell Inspiron 518 with a Pentium E2220 with 2 cores at 2.4GHz and 3GB of memory running Windows 7 Ultimate (32bit). The ATTO and CrystalDiskMark were also run on the Verbatim disk drive connected to a high-end workstation, to see if the small memory, low-end CPU, and 32-bit-ness of the test system made any difference; no difference was seen.
Here at the Thoughts Media Product Testing Center Annex #42, we don’t have many USB 3.0 devices to compare with the Verbatim drive. I like to use old internal SATA disks in a docking station for backups, and one day Meritline had a special on a MyGica USB 3.0 SATA docking station for $20 which was too good to pass up. I put a Western Digital 10,000RPM 80 GB SATA drive in that to use for testing. Besides that, I borrowed a Super-Talent USB 3.0 Express DUO 16GB flash drive (about $30), and also compared some USB 2.0 devices, in particular a trusty old Pocketec DataStor SATA USB 2.0 180GB disk that at one time was one of the fastest portable disks available.
So these comparison don’t tell us how the Verbatim disk drive fairs against direct competitors, but they are still informative in helping you decide if a USB 3.0 portable disk drive is worth investing in, and, in general, what to expect with USB 3.0.