by Rob Williams on June 21, 2010 in Storage
As the market begins to fill up with USB 3.0 storage devices, it might seem like we’re nearing a USB 2.0 end. But that will still take some time, and as Corsair helps prove with its Survivor GTR, there’s still life left in the now-aging standard. Not only is this drive ultra-fast, but it’s also ultra-durable – giving you the best of both worlds.
For most of our performance-type content, we hold nothing back when explaining our methodologies and beliefs. But as this is simply an external storage review, we don’t feel there’s quite as important a need to do that. If you’ve read our other performance content, you already know how seriously we take our testing practises, as it’s obvious that coming up with an accurate end score for any benchmark is very important. In the case of flash drives, we repeat all tests at least twice to verify that our results are accurate.
For our real-world transfer tests, the source files are stored on Intel’s X25-M G1 solid-state disk, which avails us a top-end read speed of around 250MB/s. Unless the USB device we’re testing with is able to write in excess of that, there should be no bottleneck.
To start things off, we’re using Iometer, a popular storage benchmarking application that’s as effective as it is customizable. It’s for both of those reasons that we choose to use it, and also thanks to the fact that it’s capable of outputting the results to both MB/s and IOPS (in/out operations per second). The latter is the value we focus on, as it’s become a standard for measuring performance in enterprise/IT environments.
Admittedly, running this test on most USB flash drives, especially 2.0 models, is not entirely important given the typical manner they’re used, but it’s our goal to see where one excels over another when dealing with such an intensive test. IOPS performance would be very important if you were to install an OS on a flash drive, as long as the bandwidth throughput is also good.
It’s become clear that the chips used in USB 3.0 flash drives far surpasses the performance of those typically used in 2.0 drives, but that’s likely a surprise to no one. It’s interesting though, because even in 2.0 mode, Super Talent’s high-end drive still delivered rather impressive IOPS performance… the 2.0 devices don’t even come close.
For Corsair’s GTR, the IOPS performance is actually less than that of Kingston’s mainstream DataTraveler drive, which is a bit strange as that drive is only rated for 10 MB/s read and 5 MB/s write speeds. As the test intro stated though, higher IOPS performance isn’t going to be that desired if the overall bandwidth doesn’t match up, which in the case of Kingston’s drive, it doesn’t.
HD Tune Pro 3.5
One of the simplest methods for testing storage is with HD Tune, and as it’s able to give reliable and repeatable results, we like using it in our testing. Although the program offers a good range of testing methods, we use the basic test that gives us read speeds and also access latencies.
As we saw from Super Talent’s fast SuperCrypt drive in USB 2.0 mode, the top-end limit we’re seeing is 33.3 MB/s read, and Corsair’s drive verifies that as it came out with the exact same result. As it has lower grade chips, the minimum performance was lacking just a wee bit, but the difference is almost not even worth mentioning.
For latency, Corsair’s drive improved a bit on Kingston’s, but didn’t come close to Super Talent’s… again, to be expected given the high-end chips used in that drive.
One of the more popular storage benchmarks currently is Futuremark’s PCMark Vantage. Even though this is a suite designed to benchmark your entire machine, it’s HDD suite test is quite robust and is good at delivering scores that scale well with the storage device you are benchmarking. Almost all of the storage companies we deal with regularly recommend using it, so we do.
I debated with myself over including a result here, because as you can see, Corsair’s drive wound up with a bizarre score. But it is what it is, so I am including it. After running PCMark’s test on the drive a total of eight times, across two different PC’s for verification, the end score would always be “N/A”. However, the sub-tests did have actual performance numbers, so for some unknown reason the program was simply unable to come up with an end result.
I’m unsure why this is the case, but I’ve had it happen in the past with certain SSD’s as well. I am including the result here only for interest’s sake, as it’s the only USB drive I’ve tested that wasn’t able to deliver an end score.