Once in a while, out of need or out of desperation to use a discount coupon, I end up buying something, often technology related. This time, it was the latter case, so I went and purchased something I knew would almost always be of use – a USB flash drive.
This time around, I decided that 128Gb capacity was the “sweet” spot. Eyeing the cheapest options available, along with the discounts, led to the Sandisk Ultra Flair which I could get delivered at AU$30 a piece. While not insanely cheap, it was still cheap enough to get me to purchase it.
As with most SanDisk products, it is packed on a retail card. The drive itself occupies a small space on the card and is available in a number of colours. This particular unit advertises “up to 15X faster” performance than USB 2.0 drives, but the fine print makes it clear that the assumption is that the USB 2.0 drive only writes at 4MB/s. In other words, they’re inferring that it writes at up to 64MB/s. It also claims a 5 year warranty and the inclusion of a RescuePRO Deluxe software 1-year download offer. The drive is Made in China.
Splitting open the card reveals the key for the software offer as well as a warranty information statement. Of interest is that the Australian limited warranty is provided by SanDisk International of Dublin, Ireland – which potentially seems to be a tax-evading strategy.
The unit comes packaged in a moulded clear plastic tray with a thin plastic overlay which is sealed over it to “keep it fresh”. Opening this was rather difficult as the film didn’t want to peel, so instead, it was easier just to puncture it around the drive and extract it like a “pill”.
The drive’s USB port is non standard, instead being made of a sheet of folded brushed metal. Normal perforations in the shell are not present or show only as indents, as the drive is based around a system-in-package design residing inside just the USB connector shell area. The additional metal on the front, however, should serve as additional heatsinking which should allow the drive to maintain its performance over long read/write operations. The majority of the plastic portion of the drive serves no great functional purpose except to provide a lanyard attachment point and to avoid it being lost.
The design of the drive also has a subtle curve to it.
From the front, the USB 3.0 connector module can be seen, along with the system in package underneath which is soldered to the connector module.
Attaching it to a PC reveals the drive has a VID of 0781 and PID of 5591. The drive has a total capacity as provided of 122,980,499,456 bytes (114GiB), with 122,968,211,456 bytes free.
This is due to the inclusion of SanDisk Secure Access software and user manuals. The drive contains an MBR partitioning scheme, with the first partition starting at sector 32.
Performance testing of the drive was conducted on my “new” workstation running Windows 10 with its on-board AMD USB 3.0 controller based on the X370 chipset. Testing was performed with the drive utilising its supplied format, which was restored after destructive tests through restoring an image of the partition table.
Like some other drives, in a fresh unwritten state, the drive does not reach its full sequential read performance and only provides an average of 59.9MB/s. Once fully written, the drive achieves an average of 156MB/s read.
Writes repeatedly achieve a peak of about 37-38MB/s peaks in the early part of the drive, falling to an average of about 15MB/s through the remainder of the drive. This suggests there could be about 3-4GiB of pSLC cache to accelerate writes, with transfer to slower TLC memory occurring during drive idle periods under normal circumstances. This would improve the perceived performance under low write-duty-cycle applications.
Unfortunately, this is a far cry from the expectation of up to 60MB/s based on the package information. We’ll see if this magical figure can be attained through other benchmarks.
Random access performance within HDTune seems to show that read performance at 4KB to reach 5.9MB/s and at 64KB to reach 59.8MB/s. Write performance is less impressive at 0.736MB/s and 9.4MB/s respectively. There were also cases of long delays during write, suggesting that the background activity of the controller may interfere with “real-time” writes as it manages the cache and wear levelling activities.
CrystalDiskMark reports similar figures with a test file of length 1GiB. Sequential read is reported as 164.8MB/s, write as 37.72MB/s. Small block 4kB accesses reached 8.3MB/s read and 0.63MB/s write.
Testing the drive with ATTO was performed twice due to some inconsistencies which could be due to the behaviour of the internal cache. On large block accesses, the sequential write topped out at about 27MB/s and sequential read at about 166MB/s with spikes to 173MB/s. At accesses of 128KB and above, it seems the full performance can be attained. Unfortunately, there was no “magic” write-speed result here either.
Testing with H2testW did not find any data integrity errors, with the measured write speeds slightly higher than that reported by HDTune Pro and the read speed slightly lower, but otherwise in accordance with expectations.
The SanDisk Ultra Flair 128Gb USB 3.0 Flash Drive is both cheaper and marginally faster in write speed than the SanDisk Ultra Dual Drive m3.0 I purchased around four months ago. It seems to have employed some of the same strategies as desktop SSDs using TLC memory, by employing a portion of the memory as a higher-speed (possibly pSLC) cache area so as to increase performance on some benchmarks which only make limited amounts of writes and to improve user experience in low-write duty environments.
Unfortunately, it is still only “marginally” faster, with a true write performance hovering about 15MiB/s which falls short of even saturating a USB 2.0 bus. Unfortunately, the read performance does seem marginally slower, at 156MiB/s. Compared to the package which claims “up to 15X faster” than a USB 2.0 drive (of 4MB/s), it doesn’t seem this drive is able to meet the claim on my equipment.
While it is cheap, it is disappointing that USB 3.0 class devices are still exhibiting such poor write speeds, especially as SSD prices are also falling and they offer much improved speeds. It seems there has not been much progress in this segment, at least amongst SanDisk products. This may be, in part, due to the obsession with small size “system in package” designs. This also does result in the plug being noticeably warm upon unplugging.
That being said, the drive did not exhibit any problems during testing and is a bit more stylish than most low-cost drive options on the market. It is, we can say, “adequate” but not in any way special.