Sequential Read Performance

Our first test of sequential read performance uses short bursts of 128MB, issued as 128kB operations with no queuing. The test averages performance across eight bursts for a total of 1GB of data transferred from a drive containing 16GB of data. Between each burst the drive is given enough idle time to keep the overall duty cycle at 20%.

Burst 128kB Sequential Read (Queue Depth 1)

The HP EX920 breaks another record by delivering burst sequential read speeds of almost 2.5GB/s, when no other TLC-based SSD we've tested has yet hit 2GB/s on this test. The EX920 is even beating the MLC and 3D XPoint-based drives on this test.

Our test of sustained sequential reads uses queue depths from 1 to 32, with the performance and power scores computed as the average of QD1, QD2 and QD4. Each queue depth is tested for up to one minute or 32GB transferred, from a drive containing 64GB of data. This test is run twice: once with the drive prepared by sequentially writing the test data, and again after the random write test has mixed things up, causing fragmentation inside the SSD that isn't visible to the OS. These two scores represent the two extremes of how the drive would perform under real-world usage, where wear leveling and modifications to some existing data will create some internal fragmentation that degrades performance, but usually not to the extent shown here.

Sustained 128kB Sequential Read

On the longer sequential read test with higher queue depths, the Samsung 960 PRO catches up to the EX920 and Samsung's TLC-based competitors come close. However, this only applies to a clean drive and data that was written sequentially. When the data has been fragmented by random writes, the sequential read speed of the EX920 drops by a lot, leaving it merely average and much slower than the Samsung drives (or the Optane SSD that is immune to this form of fragmentation).

Sustained 128kB Sequential Read (Power Efficiency)
Power Efficiency in MB/s/W Average Power in W

With such high sequential read performance on a clean drive, it's no surprise to see the HP EX920 near the top of the efficiency chart, even though the drive is pulling more than 5W. On fragmented data the EX920 ends up worse off than any drive that isn't using a Silicon Motion controller, because the sequential read speed tanks while power consumption remains about the same.

The HP EX920 gets up to speed very quickly, reaching full sequential read speed at QD2 when several high-end NVMe SSDs require quite high queue depths to deliver their best performance. The EX920 is eventually surpassed by numerous competitors, but at low to mid queue depths it is definitely the fastest.

Sequential Write Performance

Our test of sequential write burst performance is structured identically to the sequential read burst performance test save for the direction of the data transfer. Each burst writes 128MB as 128kB operations issued at QD1, for a total of 1GB of data written to a drive containing 16GB of data.

Burst 128kB Sequential Write (Queue Depth 1)

The burst sequential write speed of the HP EX920 is great, but it is overshadowed a bit by the large lead Samsung's 970 EVO has on this test.

Our test of sustained sequential writes is structured identically to our sustained sequential read test, save for the direction of the data transfers. Queue depths range from 1 to 32 and each queue depth is tested for up to one minute or 32GB, followed by up to one minute of idle time for the drive to cool off and perform garbage collection. The test is confined to a 64GB span of the drive.

Sustained 128kB Sequential Write

On the longer sequential write test, the HP EX920 falls a bit further behind as the scores of the top-tier drives are a bit more spread out than for the burst test. The EX920 is still providing performance that is worthy of a high-end NVMe SSD, but it can't match Samsung or WD.

Sustained 128kB Sequential Write (Power Efficiency)
Power Efficiency in MB/s/W Average Power in W

The power efficiency of the HP EX920 on the sustained sequential write test is good, but several drives deliver higher performance while drawing less power, and the 970 EVO offers substantially higher performance with only slightly higher power draw. The most efficient drive is still the Toshiba XG5, which performs the same as the EX920 on this test but draws only 3W while the EX920 requires more than 5W.

Like almost all drives in this class, the EX920 shows very consistent performance across most of the test duration, with the only hint of SLC write caching effects showing up in the slight drop in performance after QD2. The Toshiba XG5 is the most notable exception, as it takes a bit longer to get up to speed.

Random Performance Mixed Read/Write Performance
Comments Locked

32 Comments

View All Comments

  • olderkid - Monday, July 9, 2018 - link

    When is the last time you bought a hot cake?
  • SanX - Monday, July 9, 2018 - link

    Who here remembers two decades ago history when DRAM prices dropped 5-6 times in a matter of few months and no one got bankrupt?

    Yes, the electronic industry is screwing people for decades. In the brain of salespeople the 10nm Apple A11 4.3 billion transistor chip can cost $25 but similar transistor count some Intel Xeon processor made even by ancient 20-30nm tech by their crazy logic can not cost less then $1000-2000
  • Adramtech - Tuesday, July 10, 2018 - link

    SanX, All these companies from 20 years ago are not in the game because there are not enough profits to go around....and there were many more from this time period that went out of business.
    Samsung
    NEC
    Hitachi
    Hyundai
    Toshiba
    LG Semicon
    TI
    Micron
    Mitsubishi
    Fujitsu
  • Adramtech - Tuesday, July 10, 2018 - link

    Extra Credit: what companies from this 1990's list make DRAM today?
  • Totally - Tuesday, July 10, 2018 - link

    I'll play

    Samsung
    TI <- technically still does since it is partnered with Micron
    Micron
  • Dr. Swag - Tuesday, July 10, 2018 - link

    Surprise surprise, cost/transistor went down from 20-30nm to 10nm, and also the xeon has a much bigger die size and so had lower yields.
  • Totally - Tuesday, July 10, 2018 - link

    You're speaking words that one doesn't understand.
  • SanX - Tuesday, July 10, 2018 - link

    Rotfl rotflnao Swag and Totally. You made my day...how poor anandtech readers degraded so miserably.... Oh, yea, sure, Intel is that dumb company which makes potato chips. It also as a complimentary business makes retarded design processors on older tech because wants more defects on the yield, low margin, and has no brain, no money and no advanced factories. Sure, it can not lower the production cost to $10-20 bucks like all others obviously do.

    Lololololol. Congrats, you two made the most stupid comment of the year.
  • Hectandan - Thursday, July 12, 2018 - link

    Xeons are likely to cost less than $1K.
    But production cost is nothing in this industry. R&D, demand, etc almost always drive the prices. Otherwise why are iPhones selling for $1K with a $25 chip?
  • FullmetalTitan - Thursday, July 12, 2018 - link

    Cause Samsung charged them $124 a piece for the OLED display and driver IC assembly for one. That is compared to ~$30 for the LCD displays used in the iPhone 9.

Log in

Don't have an account? Sign up now