The USB 3.0 Boost software from Asus is designed to take advantage of the recently introduced UASP (USB attached SCSI protocol) to give a performance boost to supported USB 3.0 devices, but also includes performance optimization features for non-UASP devices. The main advantage of this new protocol is that it can handle multiple commands at once, rather than the single command limit with the older BOT protocol. It will recognize on the fly the protocols supported by the device, and enable whichever mode that will give the best transfer rates.
Currently, UASP is only supported on a select number of motherboards and devices and while Puget Systems sells compatible motherboards, we have not yet found a UASP external hard drive enclosure that meets our strigent standards of quality. Since this software is supposed to improve performance for any USB 3.0 device, we wanted to find out exactly how well it works - even on a the non-UASP devices in our current product line.
For the most current list of supported devices, we recommend checking Asus's USB 3.0 Boost website. Supported devices are listed at the bottom of the page.
For our testing, we wanted to find out exactly how much of a benefit (if any) the USB 3.0 boost software has on the USB 3.0 external hard drive enclosures we at Puget Systems currently sell - the Vantec NexStar-3 w/ Asmedia 1051 controller chip. For our testing platform, we used the Asus P8Z68-V Pro motherboard. To see how this software scales, we tested the USB 3.0 enclosure with three different hard drives that span a range of speeds. These hard drives are the Western Digital Scorpio Black 500GB, Intel 320 80GB SATA II SSD and the Intel 510 250GB SATA 6Gb/s SSD.
We ran benchmarks first with the software uninstalled, again with the software installed and a third time connected directly to the onboard SATA controller bypassing the enclosure. This should give us a good idea of how well the software does versus not having the software installed and how close it gets to the performance of connecting directly to the SATA controller on the motherboard. CrystalDiskMark was used to record the performance of each drive in each configuration.
Our first results for the sequential write speeds are very promising. The Intel 510 SSD is still nowhere near SATA 6GB/s speeds, but does get a nice little 16MB/s boost. The real surprise here is the Intel 320 SSD, which more than doubles in speed with a 152MB/s boost in performance. For a non-UASP device, that is an impressive increase in performance. The Western Digital drive does get a slight boost in performance, although it is actually still within our margin of error.
After the great improvements we saw in the sequential read, the sequential write results are a bit disappointing. The Intel 510 SSD actually gets a performance hit of 23MB/s, while the other two drives essentially had no change in performance.
Once again, we see a really nice boost in performance in the random read 512KB set of tests. The Intel 520 SSD gets a nice 57MB/s boost and the Intel 320 SSD gets a giant 149MB/s increase. That's over a 3x increase in write speeds! Interestingly, this puts the Intel 320 SSD at faster than standard SATA speeds, while the Intel 510 SSD is still nowhere near as fast as a direct SATA connection. The Western Digital drive again gets the least improvement, although the increase is now outside of our margin of error.
Unlike the sequential write, we recevied an increase in performance across the board in the random write 512KB tests. The only significant increase this time however is on the Intel 510 SSD just under a 20MB/s increase.
Switching over to how the USB 3.0 Boost software affects the IOPS, we see a mixed set of results with the 4KB random read tests with only one QD. The Intel 510 SSD and Western Digital Black both get a slight hit, while the Intel 320 SSD gets a slight boost.
On the 4KB random write tests with only one QD we again have mixed results. This time the Intel 320 SSD and Western Digital Black both get a slight boost, while the Intel 510 SSD gets a slight hit.
Turning the QD up to 32 shows us just how much of an advantage SATA has over the USB 3.0 interface in terms of IOPS. The Intel 320 SSD does get a 259 increase in IOPS, although it is still way below SATA speeds. Both the Intel 510 SSD and Western Digital Black drives get a slight performance hit, although it is very minimal.
For our last test, we see only minor changes in performance. The Intel 510 SSD and Western Digital Black drives are essentially unchanged, while the Intel 320 SSD gets a very slight performance boost.
While we overall saw an improvement in read and write speeds with the USB 3.0 Boost software, the amount of improvement heavily depended on the hard drive used. Because of this, only devices with the potential for very high transfer rates (external drive with a SSD for example) will show any noticeable increase in performance. For slower devices such as USB thumbdrives or external drives with a traditional hard drive, the benefits will be barely - if at all - noticeable.
Given the dramatic improvement we saw in some instances and how much better the software is supposed to perform with UASP supported device, UASP support is definitely high on our list of requirements for future USB 3.0 devices. We are looking forward to finding a UASP-enabled hard drive enclosure that meets our various standards, but for now this software does not provide the enough benefits for us to pre-load on systems that purchase an external hard drive.