There's a performance chart on the product box but it doesn’t directly apply to our device. It is based on benchmarking results obtained with a completely different product, so they only serve to emphasize the difference in performance between the interfaces. As you probably know, the peak theoretical bandwidth of USB 3.0 is 5 Gbps whereas Thunderbolt is, also theoretically, twice as fast. But we were going to benchmark the actual performance of the specific external disk. The MiniStation Thunderbolt HD-PA1.0TU3 requiring no external power or drivers or anything else, our job seemed to be easy enough. However, we encountered certain difficulties, even though through no fault of the disk itself.
Our Intel DZ77RE-75K mainboard was equipped with both USB 3.0 and Thunderbolt. The latter was implemented via an Intel L3310L controller, but the engineering sample of the controller on our mainboard did not work properly. It could see that something was connected to it, but couldn’t identify the connected device.
That’s why we had to reassemble our testbed using another mainboard with USB 3.0 and Thunderbolt. It was a Gigabyte GA-Z77X-UP5 TH. Besides the mainboard, our testbed included a non-overclocked Intel Core i5-3570K and two DDR3 SDRAM memory modules (Corsair Vengeance CMZ16GX3M4X1866C9R, 1333 MHz, 9-9-9-24-1T). The testbed ran Microsoft Windows Ultimate SP1 64-bit (Microsoft Windows, Version 6.1, Build 7601: Service Pack 1) and we also installed Intel Chipset Software Installation Utility 22.214.171.1240. The data-transfer rate was benchmarked with the help of CrystalDiskMark 3.0.1c.
It's a well-known fact that USB 2.0 cannot deliver in practice its theoretical 480 Mbps and we also know that the speed of a USB 3.0 connection is far from the specified 5 Gbps. Therefore, we suspected Thunderbolt to be slower than its promised 10 Gbps, too. However, we didn't really expect the disk to perform so slowly and, moreover, at the same speed irrespective of the interface.
So, the diagram above shows the actual performance of the Buffalo MiniStation Thunderbolt HD-PA1.0TU3. That’s enough to evaluate this product but, although our task seemed to be complete, we still wanted to achieve something better. The hard disk installed in the MiniStation obviously limited its performance. By the way, when connected via USB 3.0, the device is identified by the OS as “HD-PATU3” but the specific hard disk model (“ST1000LM024/HN-M101MBB”) is indicated in case of a Thunderbolt connection. The hard disk has dual marking because it is both Seagate Momentus ST1000LM024 and Samsung SpinPoint M8 HN-M101MBB. It is a 2.5-inch HDD with Serial ATA 3 Gbit/s interface, 1000GB capacity, a spindle rotation speed of 5200 RPM, and an 8MB buffer.
The MiniStation Thunderbolt HD-PATU3 series is not meant to be dismantled by the end-user. There are no screws or fasteners on the surface of this portable disk, yet we found it easy to take our sample apart. The PCB with connectors and the hard disk installed above it are protected with a casing on all sides and with a plastic plate from below. The whole arrangement is fastened with screws and locks to the top panel whereas the bottom metallic part of the case is held by a piece of dual-sided scotch tape. You can just slightly heat the bottom part up to detach it from the top part and then easily dismantle the whole device. So, we replaced the default hard disk with a Kingston SSD Now V+ Series (SNVP325-S2, 128 GB, SATA 3 Gbit/s) which we had previously used in our mainboard tests. Then we reran our performance benchmarks.
Of course, replacing the default 1TB hard disk with a 128GB SSD reduces the external disk’s capacity but the speed gets much higher. Now we can easily see that the Thunderbolt connection is faster than USB 3.0, even though not twice faster as we might expect from their specs. Besides, Thunderbolt is only ahead in terms of reading. It is slightly slower than USB 3.0 when it comes to writing.
The Serial ATA interface is implemented in the Buffalo disk via a SATA 3.0 controller although the default hard disk is SATA 2.0. We could see some performance benefits from replacing one SATA 2.0 disk with another because the results were not limited by the interface. But what if we use an even faster disk? We checked this out by installing a Crucial m4 SSD (CT256M4SSD2, 256 GB, SATA 3.0) and rerunning our tests. Here are the results:
That’s the performance level we can be satisfied with. With the fastest disk inside, the MiniStation makes the difference between the interfaces more conspicuous. Thunderbolt ensures the highest speed of reading, although writing is still faster with USB 3.0. The same data can be represented in a different way for better readability. The speed of reading comes first:
If you connect the MiniStation via USB 2.0, there is no need to replace the internal HDD with an SSD because the performance is going to be as low as before. In this case, it is the interface that acts as a bottleneck. The speed gets much higher if the MiniStation with default HDD is connected via USB 3.0 or Thunderbolt, but the latter ensure the same performance, being limited by the speed of the HDD itself. So, it is only after you replace the default HDD with a fast SSD that you can appreciate the benefits of the new interface. The faster your disk, the higher the data-transfer rate.
Now let’s see what we have at writing:
Everything we’ve said above about USB 2.0 and the default HDD remains true. The performance grows up with the replacement of the HDD with the slower SSD and then grows up some more with the faster SSD. The biggest surprise is that the Thunderbolt connection provides a lower speed of writing than USB 3.0. We can’t be sure if this is normal because we don’t have other peripherals to compare the Buffalo MiniStation Thunderbolt HD-PA1.0TU3 with. This product’s Thunderbolt and USB 3.0 interfaces are implemented via separate controllers, so the difference in write speed may be peculiar to its design only.