Here’s a shout out to long-time TenForums member and Win10.Guru subscriber Toni Fasth. He was kind enough to comment on a recent story of mine: Interesting US$200 2.5″ Storage Experiment. In that story, I compared performance for three devices: a 1TiB NVMe drive in a Sabrent USB 3.1/Thunderbolt enclosure, a 2TiB HDD in a USB 3.0 enclosure, and a 5TiB HDD in a USB3.1 enclosure. He observed that I was not getting the kinds of results I should have been seeing from the NVMe drive. Further, he correctly asserted that this meant the drive for that device didn’t support the USB Attached SCSI Protocol (UASP). This reminded me that I’d written a story in July 2019 called Getting from BOT to UASP, in which I’d confessed my inability to get my production PC to run UASP drivers, neither for love nor money.
Guess which machine I ran my tests on? The production PC, of course. Now that I’ve recovered from my Homer Simpson moment (Doh!), I decided to re-run those tests on another machine that DOES support UASP (and Thunderbolt 3, in fact): one of my Lenovo X380 Yoga laptops. It showed more or less what Toni said it would, to no great surprise for either of us. Sigh.
Why Say UASP Outperforms BOT Significantly?
The results speak eloquently for themselves. The lead-in graphic for this story shows the results for the same drive in two use cases. Case 1 (to the left) shows CrystalDiskMark output via a USB-C connection on the Lenovo. Case 2 (to the right) shows the CrystalDiskMark output via a USB 3 connection on the production PC (which uses the bulk-only transport, aka BOT, through the USBstor.sys driver). Huge, huge difference that varies in the range of just under 2 to around 4 times as fast for UASP versus BOT. Comfortingly, that’s how it’s supposed to work.
What a Difference a (Fast) Interface Makes
Obviously, if you want to get the best performance out of your hardware, using the fastest highway to ferry data between a storage device and the PC is the right way for it to travel. The faster connection made only modest differences in the results for the two HDDs I also discussed in the preceding story. Thus, I’m happy to say that while Toni’s brilliant observation considerably speeds up the NMVe device, it doesn’t impact the HDD devices much one way or the other. That means my basic conclusion that the 5 TiB is a good buy for both cost and performance reasons still stands. I’m not sure if that means I’m good at this stuff, or just lucky, but I’ll take it!
Author: Ed Tittel
Ed Tittel is a 30-plus-year computer industry veteran. He’s a Princeton and multiple University of Texas graduate who’s worked in IT since 1981 when he started his first programming job. Over the past three decades he’s also worked as a manager, technical evangelist, consultant, trainer, and an expert witness. See his professional bio for all the details.