AnandTech Storage Bench - The Destroyer

The Destroyer is an extremely long test replicating the access patterns of very IO-intensive desktop usage. A detailed breakdown can be found in this article. Like real-world usage, the drives do get the occasional break that allows for some background garbage collection and flushing caches, but those idle times are limited to 25ms so that it doesn't take all week to run the test. These AnandTech Storage Bench (ATSB) tests do not involve running the actual applications that generated the workloads, so the scores are relatively insensitive to changes in CPU performance and RAM from our new testbed, but the jump to a newer version of Windows and the newer storage drivers can have an impact.

We quantify performance on this test by reporting the drive's average data throughput, the average latency of the I/O operations, and the total energy used by the drive over the course of the test.

ATSB - The Destroyer (Data Rate)

The average data rate from the Crucial P1 on The Destroyer is comparable to other entry-level NVMe drives like the Phison E8-based Kingston A1000 and the Intel 660p. The P1 also roughly matches the average data rate of the Crucial MX500 SATA SSD, while several high-end NVMe drives deliver more than twice the performance.

ATSB - The Destroyer (Average Latency)ATSB - The Destroyer (99th Percentile Latency)

While the average data rate from the Crucial P1 may have been similar to the MX500, the average latency is about twice that of the MX500, and slightly higher than the Intel 660p. The situation is better for the 99th percentile latency, where the Crucial P1 comes close to the MX500 and shows half the 99th percentile latency of the Intel 660p.

ATSB - The Destroyer (Average Read Latency)ATSB - The Destroyer (Average Write Latency)

The average read and write latencies for the Crucial P1 on The Destroyer are both slightly worse than what the Intel 660p provides, and the Crucial MX500 provides a much better average write latency.

ATSB - The Destroyer (99th Percentile Read Latency)ATSB - The Destroyer (99th Percentile Write Latency)

The 99th percentile read latency from the Crucial P1 is very good, competitive with several high-end NVMe SSDs. However, the 99th percentile write latency is quite poor compared to almost any other NVMe SSD or mainstream SATA SSD. This is a huge difference in behavior compared to the Intel 660p. The Crucial P1 is optimized much more for reliable read latency, at significant cost to worst-case write latency under heavy workloads.

ATSB - The Destroyer (Power)

The Crucial P1 requires significantly more energy to complete The Destroyer than the Intel 660p, despite the near-identical hardware and very similar overall performance numbers. The Crucial P1's relatively poor efficiency on this test isn't a serious issue given that the drive is intended for less demanding use cases, but combined with the high 99th percentile write latency this points to the P1 possibly experiencing higher write amplification on The Destroyer than the Intel 660p experiences.

Introduction AnandTech Storage Bench - Heavy
Comments Locked


View All Comments

  • DigitalFreak - Thursday, November 8, 2018 - link

    At this rate, by the time they get to H(ex)LC you'll only be able to write 1GB per day to your drive or risk having it fail.
  • PeachNCream - Thursday, November 8, 2018 - link

    Please don't give them any ideas! The last thing we need is NAND that generously handles a few dozen P/E cycles before dying. We've already gone from millions of P/E cycles to a few hundred in the last 15 years and data retention has dropped from over a decade to under six months. Sure you can get a lot more capacity for the price, but NAND needs to be replaced with something more durable sooner rather than later. (And no, I'm not advocating for Optane either, just something that lasts longer and has room for density improvements - don't care what that something is.)
  • MrCommunistGen - Thursday, November 8, 2018 - link

    I was expecting the extra DRAM to provide a more meaningful advantage over the Intel 660p... I guess it makes sense that Intel left it off to save on BOM.
  • Ratman6161 - Thursday, November 8, 2018 - link

    This could be a very good standard desktop drive if 1) the price is right and 2) you can accept that the 1 TB drive is really only good for up to 900 GB. You would just partition the drive such that there is 100 GB free (or make sure you always just keep that much space free) so you always have the maximum SLC cach available. For the price to be right, it has to be lower. Taking the prices from the article, the 1 TB P1 is only $8 cheaper than a 970 EVO. Now if they could get the price down to the same territory as the current MX 500 they might have something.
  • Billy Tallis - Thursday, November 8, 2018 - link

    Leaving 10% of the drive unpartitioned won't be enough to get the maximum size SLC cache, because 1GB of SLC cache requires 4GB of QLC to be used as SLC. However, 10% manual overprovisioning would definitely reduce the already small chances of overflowing the SLC cache.
  • mczak - Thursday, November 8, 2018 - link

    On that note, wouldn't it actually make sense to use a MLC cache instead of a SLC cache for these SSDs using QLC flash (and by MLC of course I mean using 2 bits per cell)? I'd assume you should still be able to get very decent write speeds with that, and it would effectively only need half as much flash for the same cache size.
  • Billy Tallis - Thursday, November 8, 2018 - link

    Cache size isn't really a big enough problem for a 2bpc MLC write cache to be worthwhile. Using SLC for the write cache has several advantages: highest performance/lowest latency, single-pass reads and writes (important for Crucial's power loss immunity features), and your SLC cache can use flash blocks that are too worn out to still reliably store multiple bits per cell. A slower write cache with twice the capacity would only make sense if consumer workloads regularly overflowed the existing write cache. Almost all of the instances where our benchmarks overflow SLC caches are a consequence of our tests giving the drive less idle time than real-world usage, rather than being tests representing use cases where the cache would be expected to overflow even in the real world.
  • idri - Thursday, November 8, 2018 - link

    Why don't you guys include the Samsung 970 PRO 1TB in your charts for comparison? It's one of the most sought after SSDs on the market for HEDT systems and for sure it would be useful to have your tests results for this one too. Thanks.
  • Billy Tallis - Thursday, November 8, 2018 - link

    A.) Samsung didn't send me a 970 PRO. B.) The 970 PRO is pretty far outside the range of what could be considered competition for an entry-level NVMe SSD. It's a drive you buy for bragging rights, not for real-world performance benefits. The Optane SSD is in that same category, and I don't think the graphs for this kind of review need to be cluttered up with too many of those.
  • PeachNCream - Thursday, November 8, 2018 - link

    Not to be obtuse, but by price the 970 PRO is well within the range of competition for the P1 given that the 1TB 970 retails for $228 on Amazon right now and the MSRP for the 1TB P1 $220. Buyers looking for a product will most certainly consider the $8 difference and factor that into their decision to move up from an entry-level product to a "bragging rights" option given the insignificant difference in cost. Your first point is valid. I would have stopped there since its reasonable to say, "Physically impossible, don't have one there pal."

Log in

Don't have an account? Sign up now