Performance, Pricing and Conclusion
Performance:
Since this is such a simple design and with such a simple bridge controller, it did not take much to find the ultimate capabilities of it. We tested on several different USB 3.0 controllers under Windows 8.1 Update (UASP support was native since Windows 8). All controllers (Intel, ASMedia) delivered nearly identical results.
Sequential read:
Very respectable. This was with an 850 EVO, which gave us the highest speed here (low latencies helped here). Most other SSDs tested (M550, MX100, SSD 730, Pro 2500) turned in an even 380 MB/sec. The OCZ Radeon turned in 361 MB/sec.
Sequential write:
341 MB/sec on a 500GB 850 EVO. 342 MB/sec was the highest we saw (with the 1TB MX100 and 512GB M550). The OCZ Radeon went 338 MB/sec. The Intel SSD 730 240GB was able to write at 286 MB/sec despite falling back to its lower power mode (the 730 Series is one of very few SSDs to optionally use the 12V supply when connected to a capable system such as a desktop).
Random access:
The simplest tool for USB connected device testing is ATTO. Here we tested using QD=2:
…and QD=4:
Note the improvement in performance at the higher queue depth. Anything higher than QD=4 saw no further improvement, suggesting that the JMicron bridge can only handle a queue that deep. This is not by any means a deal breaker, as the vast majority of use as an external device is not going to see workloads demand higher than four simultaneous operations.
That said, this SSD has significantly higher random access performance when directly connected via SATA, but this is still very good given that the data is passing over USB.
Pricing:
The FE2005 is available at Amazon for $19. They have also very recently updated this to a newer FE2006 model, which is $2 cheaper and employs a different closing method. The newer model removes the lock, but adds the option of connecting external power. While we had no issue powering the most hungry drives, external power would help in cases where the device was on a longer cable or connected to an unpowered hub.
Conclusion:
As far as external USB to SATA enclosures go, the Inateck FE2005 is about as simple as they come. It's only plastic, but more than adequate for most applications. The performance was very good considering its simplicity, and the price is low enough to pick one of these up even if only for the occasional OS cloning / upgrade process. I'm personally eyeing one of their new 3.5" docks, as desktop HDD's also see a benefit from queued commands, making those media copies and backups even faster than before.







How bad do the ATTO speeds
How bad do the ATTO speeds drop when using Win7 that does not have the UASP support?
Yes windows 7 has the largest
Yes windows 7 has the largest installed base, and that is not likely to change depending how cloudy win 10 gets. How about Mint, or other Linux distros, what is the support for USAP?
Most of the vendors have
Most of the vendors have added UASP support (at the driver level) for Windows 7, but those wanting the best compatibility should consider upgrading.
They’ll upgrade when they
They’ll upgrade when they realize you don’t have to even touch Metro to get the benefits Windows 8.1 offers… which is never because they are technologically blind, and only see the “trends”.
This seems slightly faster
This seems slightly faster than the previous Inateck you tested.
https://pcper.com/reviews/Storage/Inateck-USB-30-HDD-Enclosure-and-Bluetooth-Speaker-Review-quick-easy-and-inexpensive
I picked up the USAP version you mentioned in the article and have been happy with it so far. That desktop dock looks like a good replacement for the USB3 BlacX I have on the bench.
> it’s time we look at some
> it’s time we look at some ways to link external SATA devices using this faster protocol
Allyn,
Would USB 3.1 be a natural development for enhancing USB pathways to fast SATA devices?
See summary discussion of the PCIe 3.0 128b/130b “jumbo frame” here:
http://supremelaw.org/patents/SDC/SATA-IV.Presentation.pdf
I believe the current USB 3.1 implementations use a 128b/132b “jumbo frame”:
http://en.wikipedia.org/wiki/USB#3.1
“changing the encoding scheme to 128b/132b”
Keep up the good work.
/s/ Paul