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


View All Comments

  • lmcd - Friday, October 27, 2017 - link

    Funny, when SSDs came out we were promised they'd be orders of magnitude faster than mechanical HDDs. The first ones weren't. Reply
  • Drumsticks - Sunday, October 29, 2017 - link

    I think my primary point of criticism with your view is that, despite Intel not reaching their goals, it is measurably better than the competition at a better price per GB per performance.

    It might not have hit the endurance targets they wanted, but for 2-3x the cost of your opponent, they've achieved some 20x endurance, on par in the worst case, and 6-8 times better in many measurably important scenarios.

    On top of that, you tout the benefits of SLC and claim that SLC is undeniably faster than XPoint, because if SLC wasn't held back by the cheap and underpowered controllers of yesteryear, it could really fly. How do you know the same thing is not true of 3DXP? Perhaps better understanding of its use and a better controller, and maybe a second generation of the memory, will enable Intel to reach even higher performance and endurance heights. You can't claim that SLC is better, and is simply held back by 2014 era controllers, while not allowing that 3DXP could likely be held back by it's controller as well. We have no idea what the performance ceiling for 3DXP is, because we've seen all of one generation of products.

    To extend your analogy, I might be disappointed if I only received $450, but if everybody else is still only making $200, I'm going to go home happy.
  • Rektide - Friday, October 27, 2017 - link

    This shows an Intel 900P with a total write endurance of 8.7PB, and a 850 Pro with a total write endurance of 150 TB. But a stress test of a 850 Pro in fact survived 9.1 PB[1]! Meanwhile, if you reach 8.7PB on the 900P, Intel will forcibly move your drive into read-only mode.

    If you look at Ark, Intel describes it's Endurance Rating to mean "Endurance rating indicates the expected data storage cycles to be expected over the life of the device." When they say that, they mean "and not a megabyte more". Whereas when Samsung says it, they, at least once, meant "but this drive may go 60x more than it's rating". I really severely dislike this twist of the knife, this drastic change Intel and Intel alone is perpetrating against it's consumers.

    Does the P4800X also commit seppuku the megabyte it reaches it's Endurance Lifetime? I'm not sure if all Intel drives are so malicious, or is it consumer & enthusiast drives? How do I know which drives are programmed to self destruct on me, Intel?

    [1] http://www.guru3d.com/news-story/endurance-test-of...
  • ddriver - Friday, October 27, 2017 - link

    It is actually worse than locking in read-only mode - on the next boot cycle the drive gets bricked. So if you didn't manage to get that data in time, it is gone forever. Great feature. Reply
  • Spunjji - Saturday, October 28, 2017 - link

    I have seen this with Intel consumer SSDs. It's amazing - it doesn't even tell you that it's failing and that you have one (yes, one) chance to backup your data before it goes forever. Usually the drive just throws an error, so your average user reboots the system and bam, the drive's not even in the BIOS anymore.

    Their drive failure behaviour is criminal.
  • voicequal - Saturday, October 28, 2017 - link

    Agreed - bricking is a terrible failure mode for a consumer drive. For an enterprise drive, it *might* make sense, since its you want things to fail hard and fast so that backup systems can detect and take over. Reply
  • FwFred - Friday, October 27, 2017 - link

    Wow, need a block button to improve SnR. Reply
  • Meteor2 - Tuesday, October 31, 2017 - link

    Indeed -- though I simply skip any comments written by ddriver. Never any value in them (but a remarkably effective troll). Reply
  • Lolimaster - Saturday, October 28, 2017 - link

    The 1000X was supposed to be on latency and endurance, not much more, and maybe 10X random 4K performance. Reply
  • CajunArson - Friday, October 27, 2017 - link

    I thought you were a reliable AMD koolaid drinker?

    It's funny how you insult these products that you have never used but would never in a million years hurl their $7000 Radeon SSG under the bus... you know, the one that uses a consumer-grade GPU slapped together with a RAID-0 array of cheap consumer-grade NVME drives to supposedly do rendering jobs faster than a regular GPU.

Log in

Don't have an account? Sign up now