Synthetic Benchmarks - ATTO and CrystalDiskMark

Benchmarks such as ATTO and CrystalDiskMark help provide a quick look at the performance of the direct-attached storage device. The results translate to the instantaneous performance numbers that consumers can expect for specific workloads, but do not account for changes in behavior when the unit is subject to long-term conditioning and/or thermal throttling. Yet another use of these synthetic benchmarks is the ability to gather information regarding support for specific storage device features that affect performance.

Both SilverStone and Yottamaster claim read and write speeds of around 2000 MBps, and these are backed up by the ATTO benchmarks provided below. ATTO benchmarking is restricted to a single configuration in terms of queue depth, and is only representative of a small sub-set of real-world workloads. It does allow the visualization of change in transfer rates as the I/O size changes, with optimal performance being reached around 512 KB for a queue depth of 4. The numbers for both the MS12 and HC2-C3 are similar, which is not a surprise in this section because the bridge chip as well as the internal SSD are the same. The effects of the thermal solution and other bridge chip firmware aspects are not evident in this quick benchmark.

ATTO Benchmarks
TOP: BOTTOM:

CrystalDiskMark. on the other hand, uses four different access traces for reads and writes over a configurable region size. Two of the traces are sequential accesses, while two are 4K random accesses. Internally, CrystalDiskMark uses the Microsoft DiskSpd storage testing tool. The 'Seq128K Q32T1' sequential traces use 128K block size with a queue depth of 32 from a single thread, while the '4K Q32T16' one does random 4K accesses with the same queue configuration, but from multiple threads. The 'Seq1M' traces use a 1MiB block size. The plain 'Rnd4K' one uses only a single queue and single thread . Comparing the '4K Q32T16' and '4K Q1T1' numbers can quickly tell us whether the storage device supports NCQ (native command queuing) / UASP (USB-attached SCSI protocol). If the numbers for the two access traces are in the same ballpark, NCQ / UASP is not supported. This assumes that the host port / drivers on the PC support UASP.

CrystalDiskMark Benchmarks
TOP: BOTTOM:

Both storage enclosures support UASP, as shown in the 'Rnd4K' numbers. The absolute numbers for both are essentially the same and within the expected run-to-run variations.

The 2021 AnandTech DAS Testbed and Test Suite AnandTech DAS Suite - Performance Consistency Benchmarking
POST A COMMENT

19 Comments

View All Comments

  • meacupla - Thursday, August 12, 2021 - link

    Oh, look at that! a newish NVMe to USB bridge chip from ASMedia

    Even though I know this chip blows the competition out of the water, I would have liked to see how this chip compares to its predecessors, ASM2362, RTL9210 and JMS583.

    Was there any stability or drop out issues with these units? I didn't see any mention in the article, so I assume it was smooth sailing, which is nice to see.
    Reply
  • Drkrieger01 - Thursday, August 12, 2021 - link

    I'd also be curious to know if they experienced any drop outs during testing. I saw this a fair bit with some JMicron controller based units, but I found flashing to another firmware helped tremendously.
    Nice to see some new hardware available for those still using Sneaker-Mail.
    Reply
  • ganeshts - Friday, August 13, 2021 - link

    No issues with these two storage bridges attached to the ECU06. I did experience dropouts last year with the Yottamaster C5 [ as documented here : https://www.anandtech.com/show/16133/usb-32-gen-2x... ]. Reply
  • name99 - Thursday, August 12, 2021 - link

    "USB 2.0 ports are guaranteed to deliver only 4.5W (900mA @ 5V). "

    No. USB2 ports guarantee 2.5W. It's USB3 ports (blue connector) that guarantee 4.5W.
    And lots of USB3 stuff will just fail (possibly randomly) when connected to USB2 because
    - 2.5W is enough to get the drive awake and to read, maybe even an occasional write, but not enough for a run of sustained writes.
    - I don't think power negotiation is part of baseline USB3 spec. (Or if it is, most of that hardware doesn't seem to implement it properly.)
    Reply
  • repoman27 - Sunday, August 15, 2021 - link

    Yep. USB 2.0 is up to 2.5 W (500 mA @ 5 V), while USB 3.2 specifies up to 4.5 W (900 mA @ 5 V) for single lane operation and up to 7.5 W (1500 mA @ 5 V) for dual-lane operation. Reply
  • DigitalFreak - Thursday, August 12, 2021 - link

    USB 3.2 Gen 2x2, a standard that had little reason to exist. Reply
  • DigitalFreak - Thursday, August 12, 2021 - link

    I guess at least it wasn't called USB 3.3 Gen 3. Reply
  • back2future - Monday, August 16, 2021 - link

    within time, seen from now, yes, but what to recommend?
    Waiting for Thunderbolt 4 or building on USB 4.x?
    Reply
  • dwillmore - Thursday, August 12, 2021 - link

    I read the testbed page--which I usually skip--because I wanted to know where the heck you found a USB3.2 gen 2x2 port. AiC of course.... On the way I learned about the new Type-E connector for internal connection to case USB-C ports. So, that was nice. I even found a USB 3.2 Gen1x2 host adapter--which is a cursed item if ever there was one. Reply
  • watersb - Friday, August 13, 2021 - link

    One of my favorite segments to review right now. My laptop purchased in 2021 is by far the most capable device I own. USB4 or whatever they are calling it this week for fast mobile backup storage is a key part of the workflow.

    Utterly stupefying how clumsy the market messaging has been. The only way to know what I'm are buying is to test it myself. Or get a review from a source that I can trust to go beyond "it looks pretty and we have a relationship with this brand". YouTube reviews have become more sophisticated, but the professionals here help a lot. Thanks!
    Reply

Log in

Don't have an account? Sign up now