Random Read Performance

Our first test of random read performance uses very short bursts of operations issued one at a time with no queuing. The drives are given enough idle time between bursts to yield an overall duty cycle of 20%, so thermal throttling is impossible. Each burst consists of a total of 32MB of 4kB random reads, from a 16GB span of the disk. The total data read is 1GB.

Burst 4kB Random Read (Queue Depth 1)

The Optane SSD 900P doesn't break the record for QD1 random reads, but only because we've also tested the 32GB Optane Memory M.2, which is about two microseconds faster on average for each 4kB read. The Optane SSD 900P is still about 7 times faster than any flash-based SSD.

Our sustained random read performance is similar to the random read test from our 2015 test suite: queue depths from 1 to 32 are tested, and the average performance and power efficiency across QD1, QD2 and QD4 are reported as the primary scores. Each queue depth is tested for one minute or 32GB of data transferred, whichever is shorter. After each queue depth is tested, the drive is given up to one minute to cool off so that the higher queue depths are unlikely to be affected by accumulated heat build-up. The individual read operations are again 4kB, and cover a 64GB span of the drive.

Sustained 4kB Random Read

When longer transfers and higher queue depths come into play, the Optane SSD 900P passes the Optane Memory M.2 and remains more than 6 times faster for random reads than any flash-based SSD.

Both Optane devices more or less level off at queue depths of 8 or higher. The Optane SSD 900P saturates at about 1800 MB/s while the Optane Memory tops out around 1300 MB/s. The Samsung 960 PRO 2TB hasn't caught up by QD32, and doesn't surpass the QD1 random read performance of the Optane SSD until the Samsung reaches a queue depth of about 8.

Random Write Performance

Our test of random write burst performance is structured similarly to the random read burst test, but each burst is only 4MB and the total test length is 128MB. The 4kB random write operations are distributed over a 16GB span of the drive, and the operations are issued one at a time with no queuing.

Burst 4kB Random Write (Queue Depth 1)

The burst random write performance of the Optane SSD 900P is slightly higher than the Intel SSD 750 1.2TB, and about 14% faster than Samsung's fastest.

As with the sustained random read test, our sustained 4kB random write test runs for up to one minute or 32GB per queue depth, covering a 64GB span of the drive and giving the drive up to 1 minute of idle time between queue depths to allow for write caches to be flushed and for the drive to cool down.

Sustained 4kB Random Write

With higher queue depths in play, the Optane SSD 900P scales up faster than the Intel SSD 750 1.2TB, leaving the Optane SSD with a 7-10% lead over the Samsung 960s and Intel 750.

Samsung's 960 PROs and the larger 960 EVO all trail slightly behind the Optane SSD's random write performance for queue depths 1 to 4, then the Samsung drives level off and leave the Optane SSD with a substantial performance advantage at high queue depths. The Intel 750 is slightly faster at QD1 and QD2, but saturates at an even lower performance level than the Samsung 960s.

AnandTech Storage Bench - Light Sequential Performance
POST A COMMENT

205 Comments

View All Comments

  • Nikijs - Monday, October 30, 2017 - link

    pls anand. just kill ddriver acc. he ruins all comment section. he's maybe "smart", but lives in another dimension, where he thinks he is only one who understands something. i bet he never ever achieved something worthy in his life. thats where all hidden anger comes from. Reply
  • daremighty - Tuesday, October 31, 2017 - link

    I don't agree that Anandtech's approach to measure random performance. QD1 random is directly reflect the latency of memory chip (NAND or 3D Xpoint). Between the queue, there should be some idle time and it didn't explain the real random performance of device. Under random workload, the device should handle multiple random requests - it means deeper QD is more natural to explain the random performance of QD. Probably, many device would require deep QD to saturate the random, but I think it is still valid metric. I think in random, random performance with deep (64 or 128?) QD is as much important as low QD (1/2/4?). Again, low QD is just shows the NAND performance, not SSD performance. Reply
  • rep_movsd - Tuesday, October 31, 2017 - link

    Seems like the great ddriver is an expert in all things, and his opinions of "Hypetane" are based on solid fact and "decades" of experience (of bashing intel I guess).
    All the people who buy Intel are idiots and those who praise Intel technology are shills...

    Meanwhile, Optane and similar technologies will eventually replace SSDs and ddriver will still be grumbling about how SLC would have been better if given a chance....

    Get with the times - no one is forcing anyone to buy anything Intel - and if you think anandtech fudges benchmarks, put your money where your mouth is and try doing a fraction of what they do...

    Don't pour cold water on others efforts just because you have some PTSD with Intel for whatever reason...

    Reply
  • "Bullwinkle J Moose" - Tuesday, October 31, 2017 - link

    "Get with the times"

    Thats a great comment!

    Seems like ddriver might have gotten a timeout several days ago and yet a few of you can't seem to get over him

    Just admit it, you loved his comments and want him back, or else you could get with the times and get over him

    He's gone, but look at the bright side.....
    I'M BACK!
    Reply
  • rep_movsd - Wednesday, November 1, 2017 - link

    Yes, I love him, like all trolls love other trolls Reply
  • mrhamdnloanoffer - Wednesday, November 1, 2017 - link


    Do you have bad credit or in need of urgent loan to solve a pressing need? We lend secured and unsecured loans to honest and reliable individuals and companies globally at 3% interest rate. Here is the solution to your financial problem, do you also need money for your, project, business, taxes, bills, and many others reason, contact us today for that loan you desire, we can arrange any loan to suit your budget at low interest rate.Contact us
    Reply
  • DocNo - Saturday, November 4, 2017 - link

    Intel's caching software for Optane sucks - super finicky, not easy to integrate with an existing system and most of all requires specific motherboard to work (which I happened to have, but not the right partition layout - I dunno, technical documentation for what they want is pretty nonexistent).

    Luckily I stumbled PrimoCache. Downloaded a trial and had it working with my m2 Optane in 5 minutes. Made a noticeable and rather dramatic difference, even when loading stuff off of my Samsung EVO SSD. For $20 it was worth the frustration of not having to figure out Intel's poorly documented and overly fussy software. And if you don't have an Intel supported board, this lets you use Optane for caching with any board with an m2 slot.

    These Optane drives would be awesome for servers - based on the experience with my desktop I now use PrimoCache on a couple of my servers and even with cheap SSDs the difference is amazing. With larger Optane drives? I should be even better. And at $120 for the server version it's by far the fastest way to add SSD caching to Windows Server. I'm extremely happy with it!
    Reply
  • weevilone - Sunday, November 5, 2017 - link

    That's interesting.. wish I had known about PrimoCache when I was tinkering with the little Optane stick. Intel's software was a huge mess, and Intel was less than helpful in working through it. When I finally went to remove it and throw in the towel, neither the software nor the UEFI could remove the stuff. I wound up having to reinstall Windows. Reply
  • Kwarkon - Monday, November 6, 2017 - link

    I'm quite curious what exact issues you had, especially with disabling Optane? Reply
  • mattlach - Saturday, December 30, 2017 - link

    Looking at that random 4k write performance, I'm thinking a pair of these would be absolutely fantastic as a mirrored SLOG/ZIL device for my massive ZFS pool. It's very tough to predict through.

    Question is how they would perform on the dual socket Westmere-EP Xeons powering my storage box, with only PCIe Gen2... Probably won't make a huge difference since the write speed peaks out at about 1.7GB/s and 4x PCIe Gen2 tops out at 2GB/s.

    I wouldn't mind a decent boost to sync write speeds, and this 900p seems like its tailor made for the job. No cache, so there is no need for battery/capacitor backup, and very high speed, low latency random writes...
    Reply

Log in

Don't have an account? Sign up now