|
|
| Introduction |
When Asus asked us if we'd be interested in a reviewing their new, high-end USB 3 implementation that offered better performance than anything available elsewhere, we were skeptical. Historically, USB performance has been a function of which controller (Intel, AMD, NEC, Texas Instruments, VIA, etc.) was used. Similarly, other types of products, like WiFi routers, have exceeded specified standard performance by using custom hardware. Companies like Linksys and D-Link for example, have offered enhanced operational modes, but only if you bought specific routers and wireless cards. But we digress.
Asus claimed that while you did need certain USB controllers to take maximum advantage of this new USB 3 implementation, they'd built a solution that could accelerate transfer speeds on any third-party USB 3.0 (and even 2.0) controller. For external SSD setups, you won't need a specific brand of SSD or manufacturer either. SSDs from any of the major vendors would work. Hard drives are also supported, but the inherently lower speed of mechanical drives makes it more difficult for them to take advantage of the improved throughput.
The new standard is called UASP -- USB Attached SCSI Protocol -- and we're here to tell you that it works. Current performance gains are modest, but that will improve as more high-end drives come to market. Here's how Asus offers improved throughput for all the USB controllers on its motherboards... |
| Thirteen Years of Change -- Same Protocol For Transfer |
When USB debuted in 1999, it offered maximum throughput of 12Mb/s. Today, USB 3.0 offers 4.8Gb/s. That's not surprising. What is surprising is that modern USB 3 controllers use the same Bulk-Only Transport (BOT) transfer protocol that first debuted in 1998. Before the advent of USB 3, relying on BOT made sense. Because modern hard drives were significantly faster than the USB 2 bus itself, it didn't really matter if the bus supported certain features -- the HDD was always going to be waiting on the host controller.
UASP -- USB Attached SCSI Protocol -- is designed to fix all these problems, and bring USB 3 fully into the 21st century. It does this by implementing queue functions, reducing command latency, and allowing the device to transfer commands and data independently from each other.
That's where Asus comes in. You don't need a specific type of SSD or HDD to take advantage of UASP (more on that in a moment), but you do need a specific USB 3 controller. In Asus' case, that means an Asmedia (abbreviation: ASM) controller designed to support UASP, and an external enclosure that's also capable of taking advantage of the newer protocol. Asus sent us two different enclosures, both of which use the ASM 1051E. The motherboard the company provided for our testing, the P8Z77-V, has an onboard ASM 1042 controller as well.
Here's the really nifty part. While you'll see best performance with a UASP solution, those of you with Intel (and possibly NEC) controllers aren't out of luck. Asus is also offering what it calls "Turbo," or "Optimized BOT" mode. This operating mode conforms to the BOT standard (no multiple simultaneous data transfers, for example) but streams commands without waiting for constant acknowledgement. This significantly reduces latency and improves performance. |
| Test Setup, CrystalDiskMark |
Test Setup: We tested the Asmedia 1042's UASP capability using Asus' P8Z77V motherboard, along with a Thermaltake USB 3 enclosure and a USB 3.0 Boost Cable solution. Our primary hard drive was a 1TB Western Digital Caviar Black (32MB cache).
The Intel and Asmedia USB 3 controllers are well-matched in Normal mode -- and both leap ahead once UASP/Turbo modes are enabled. The Asmedia controller has a slim edge in the Sequential Read and 512K tests and a huge advantage in the 4KQ32 read test. The latter advantage is thanks to the parallelism we showed earlier -- commands can be processed independently from data flow and data streamed much more effectively as a result. ![]() Intel's write performance drops off oddly here; Turbo Mode is slower than normal in certain cases. Real-world performance and other benchmarks don't show this discrepency, however, so we don't put too much weight on it here. Full UASP support gives the Asmedia controller better performance in the 4KQ32 test, but 4K performance drops off. Let's see how test results look in more representative scenarios.
|
| PCMark 7 |
PCMark 7's extensive storage benchmarks are designed to measure performance in a variety of real-world scenarios, including file copies, Windows Defender scanning, and application launches. We've broken the data down into two groups -- one graph for the Corsair Force 3 SSD (using both the Intel and Asmedia controllers) and a second graph for the Western Digital VelociRaptor and Caviar Green hard drives, tested on the Asmedia 1042 controller.
The SSD graph shows Intel ahead of Asmedia in normal mode by roughly 3%. When we turn Turbo mode on, Intel's performance doesn't budge. UASP, however, makes a modest difference -- the Force 3 solid state drive is eight percent faster when running in this mode.
The VelociRaptor only picks up 2% here, and that's small enough to be within the margin of error. The Caviar Green's score doesn't even budge. This lines up with what Asus told us -- you need an SSD to see much of a real-world advantage. |
| Real-World File Copy Performance |
We ran two sets of file copy tests to compare the various controllers and hard drives. Our initial data set -- a 2GB directory of mixed files -- was completely unaffected by any of the various settings. Once we swapped out our small files for a 26GB directory of encoded AVIs, we saw a moderate performance difference between normal BOT transfer and UASP/Turbo mode, at least in certain cases.
Activating UASP only improved our HDD - SSD copy speed by about four percent, but our intra-SSD copy time improved by nearly two minutes, or just over 15%. Since Asus sent over two compatible controllers (its own Boost Cable and a Thermaltake), we added a Corsair F120 SSD to the mix and did a USB-to-USB copy across the Asmedia controller. While UASP was enabled on both devices, performance was essentially identical. This is likely because the single XHCI controller in the Asmedia 1042 had its hands full moving data across the bus.
Since this test wasn't particularly illuminative, we dropped it for the HDD comparison. Here's the VelociRaptor and Caviar green breakout, again on the Asmedia 1042.
Again, the straight data copy shows only a small performance increase on the VelociRaptor. The caviar Green, again, is too slow to benefit from the better protocol's advantages. The intra-drive copy is where we see the SCSI protocol shine; even the mechanical VR drive improves its performance by roughly 10%. The Western Digital Caviar Green, in contrast, falls flat.
Did you notice that the WD VR drive is actually faster than the Force 3 in both the initial data copy and the intra-drive copy? So did we. It may be related to the asynchronous nature of the SSD's Flash; asynchronous dives transmit and receive data according to a timed pulse rather than on the rising and falling edges of their clock. Here's the Intel controller's before-and-after when using the Corsair SSD.
Using Turbo Mode improves the data copy time by about 10%, roughly the same boost we saw from the Western Digital VelociRaptor. Comparing the two controllers, we see that Intel has a slight advantage over Asmedia in Normal Mode, with the ASM 1042 eking out a small win over Intel in UASP vs. Turbo. |
| Conclusion | ||||
Performance Summary: UASP support is a great idea, particularly for someone building a desktop they plan to keep for a number of years. True, evaluated strictly in the short term, UASP might seem gimmicky. It only improves performance if you own a top-end mechanical external drive or intend to deploy external SSDs, and that's not something a lot of people are doing.
![]()
|