AnandTech Storage Bench - Light

Our Light storage test has relatively more sequential accesses and lower queue depths than The Destroyer or the Heavy test, and it's by far the shortest test overall. It's based largely on applications that aren't highly dependent on storage performance, so this is a test more of application launch times and file load times. This test can be seen as the sum of all the little delays in daily usage, but with the idle times trimmed to 25ms it takes less than half an hour to run. Details of the Light test can be found here. As with the ATSB Heavy test, this test is run with the drive both freshly erased and empty, and after filling the drive with sequential writes.

ATSB - Light (Data Rate)

On the Light test, the Corsair Neutron NX500's average data rate is slightly slower than the other two Phison E7 drives, and more substantially behind the other MLC NVMe SSDs. Of the three Phison E7 drives, the NX500 fares the best when the drive is full.

ATSB - Light (Average Latency)ATSB - Light (99th Percentile Latency)

The average latency rankings are almost identical to the average data rate rankings, except that the WD Black has jumped ahead of the Phison E7 drives. For 99th percentile latency, the NX500 performs better than the Zotac SONIX but is only faster than the Patriot Hellfire or WD Black when the test is run on a full drive.

ATSB - Light (Average Read Latency)ATSB - Light (Average Write Latency)

The differences in average read and write latency between the Phison E7 drives are pretty much negligible, and their read latencies are pretty close to the competition. The average write latencies are clearly higher than almost all the competing NVMe SSDs.

ATSB - Light (99th Percentile Read Latency)ATSB - Light (99th Percentile Write Latency)

The best NVMe SSDs provide 99th percentile read latencies that are half of what the Phison E7 drives provide, when the test is run on an empty drive. When the drive is full, the 99th percentile read latency of even the 3D TLC-based drives worsens to the level of the Phison E7 drives, leaving only a few MLC-based drives with any significant advantage. On the write side, the three Phison E7 drives perform similarly, and the top NVMe SSDs offer 99th percentile write latencies that are barely more than a tenth as long as the NX500's.

ATSB - Light (Power)

The NX500 is again in last place for energy efficiency, but the OCZ RD400A and Zotac SONIX are very close, and only the drives with 3D NAND are substantially more efficient.

AnandTech Storage Bench - Heavy Random Performance
Comments Locked

45 Comments

View All Comments

  • shabby - Wednesday, August 16, 2017 - link

    According to toms review the 500gb one still does 2gb/s read/write at qd1, nothing touches it.
  • Ratman6161 - Wednesday, August 16, 2017 - link

    Yes, according to Tom's and others, the 500 GB 960 EVO is still the drive to have in its size class - which is why I bought one. On the other hand the 256 GB 960 EVO doesn't stand up against the competition like the 500 and up drives do. So, when comparing these drives with an eye to actually purchasing one, its a great idea to find reviews of the actual size you are looking at and comparing against other like sized drives.
  • mapesdhs - Thursday, August 17, 2017 - link

    Trouble with the EVO is when its cache is full it can't take the heavy writes, and its random & steady state performance are not that great. They're good enough (I bought a 250GB for my brother), but the Pro is way better, and the 950 Pro is better aswell.
  • CrazyElf - Wednesday, August 16, 2017 - link

    This doesn't justify the premium over SATA SSDs.

    Unlike say, the Samsung SSD 960 series or the Intel SSD 750, which do have something to offer, especially when it comes to sequential performance, these drives really struggle to justify the premium.

    Yeah it seems increasingly like Intel (SSD 750 dominates sustained writes on 4k) and Samsung (dominates most other benchmarks).

    I mean, products like Optane are expensive, but at least they have some premium (ex: the good 4k performance).
  • versesuvius - Wednesday, August 16, 2017 - link

    Corsair expects to be paid higher for whatever it puts together no matter what. The same is true with Cooler Master and Logitech. For each single one of their products there is an equivalent product on the market that is higher in quality and performance and considerably lower in price.
  • valinor89 - Wednesday, August 16, 2017 - link

    I don't know about Cooler Master, but logitech sells plenty of cheap stuff and some products have very good prices for what they offer. The G502 was quite the revolution in perf/price and I find my G610 to have very good quality compared to Corsais Oferings.
  • Bullwinkle J Moose - Wednesday, August 16, 2017 - link

    With PCIe lane counts coming up, I hope to see Samsung and Intel start using them

    With a full 16 lanes being used, benchmarking can finally change along with the SSD processors/firmware

    Intel may take the lead with xpoint initially, but with the ability to run several simultaneous tests on an SSD max iop / mixed mode, copy/paste, 100GB uncompressed read speed, 100GB uncompressed write as well as torture tests as the onboard processors finally catch up to the lane counts, I think Samsung may yet have a few surprises for xpoint

    4 lanes ain't gonna cut it for my 16 core cpu

    Get with the times!
  • Ratman6161 - Wednesday, August 16, 2017 - link

    No matter how many PCIe lanes your ThreadRipper CPU has available (assume that's what you are talking about since you say 16 core) the spec for NVMe is still x4
  • Billy Tallis - Wednesday, August 16, 2017 - link

    NVMe has nothing to say about PCIe lane counts. You're probably thinking about the M.2 connector, but that is hardly the only way to connect a PCIe SSD.
  • hlm - Wednesday, August 16, 2017 - link

    Yes, it is when NVMe is used through U.2 and M.2 that you get four-lane PCIe. When used through SATA Express, NVMe is stuck with two-lane PCIe.

Log in

Don't have an account? Sign up now