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)

Starting off, the Phison E12 delivers a much higher average data rate on The Destroyer than their previous controllers, but it's not quite enough to match the top TLC-based drives on the market. Even the Plextor M9Pe with its aging Marvell 88SS1093 controller is a bit faster. On the other hand, the HP EX920 shows that Silicon Motion also has a lot of catching up to do with their next controller.

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

Average and 99th percentile latencies from the Phison E12 are among the best we've seen from a TLC-based SSD. Premium drives using 3D MLC or Intel's Optane SSDs can be a bit better, but with even 99th percentile latency approaching 1ms for TLC drives, there's not much room left for meaningful improvement on these scores.

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

The average read latency from the Phison E12 is similar to most other high-end SSDs, but the average write latency is better than any other TLC-based SSD.

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

The Phison E12 turns in great scores for both 99th percentile read and write latency. The write latency particularly stands out, and it is clear that the E12 behaves well under pressure.

ATSB - The Destroyer (Power)

The Phison E12 uses far less energy on The Destroyer than the previous-generation E7-based Patriot Hellfire, and is a bit more efficient than the slower but lower-power E8-based Kingston A1000. The WD Black still holds a significant lead in power efficiency over all the other NVMe drives, but the E12 is making some progress toward that goal.

Introduction AnandTech Storage Bench - Heavy


View All Comments

  • gglaw - Thursday, July 19, 2018 - link

    new cheap competition is exactly what the market needs. I've posted this before but the pursuit of performance for most of the readers here is pretty much wasted unless i've gotten so insensitive to speed changes I can't tell the difference anymore. I have more computers that I'm embarassed to admit to and more parts than I have time to finish building more to test. SSD's in general I wouldn't be surprised if I've tested more than the whole AT staff literally, and not just running a few hours of the same suites, hundreds of hours through my LAN room. I've had every major SSD chipset released for at least several years and within I'd say the last 2-3 years I can tell absolutely no difference between the cheapest 3D NAND SATA drive and the 970 EVO and ADATA 8200 which is a gem most people haven't even heard of that are among the fastest drives ever produced for consumer use. My slowest drives are the Micron 2TB dirt cheap $250 drives that go on sale every 2 weeks (cheapest per GB drive ever made), a couple Crucial MX 3D NAND drives, a few enterprise Samsung/HP pulls that are actually brand new sold through Newegg outlet, and some Samsung 850 non-pro's. Especially on the Destroyer tests the numbers look dramatically different, but this test is absolutely useless for at least 99% of the users on this site other than the few big organization admins who frequent these comments routinely.

    No one including myself through hundreds of hours through my LAN room have noticed/commented on any speed difference running Twitch/streaming, chat engines, browsers, while doing many hours of OW, WOW, HOTS, and Steam library games. This is about the intensity of use of the vast, vast majority of the public, likely moreso. Now other than my one flagship with about the fastest you can get of every piece, all I go for is cheap $/GB and decent warranty on my SSD's. My current flagship is the top Ryzen2, ADATA 8200 NVME drive (basically same as 970 EVO which is in my Coffee Lake rig), GTX 1080 - and sadly the only noticeable difference is playability on higher graphics settings due to the GTX 1080s on the higher machines. The 16 threaded CPU and turbo SSD's no one can tell is even in the box. And the budget machines running RX 580s, GTX 1070s/1060s, do everything identical to the flagships other than a couple graphics settings needing to be turned down. (All the rigs are also only going at 1440p since I never felt it was enough of a difference to justify the cost to go 4k).
  • romrunning - Thursday, July 19, 2018 - link

    The only point I think you've established is that you haven't run any loads that significantly use any one component - SSD, CPU, or GPU. Well, at least you saw the GPU difference in "higher settings". If you're just doing games, I can see why you might think the big performance products don't have much of an impact. However, depending on your use case, there can be huge performance/time savings in different products.

    A high-thread count CPU can make significant time differences in rendering. So if that is something you do a lot, especially if you do this for work, then you will notice a huge difference in the time saved on rendering.

    In virtual server hosts, you can tell a difference from an all-SSD array vs all-HDD array in the responsive of the virtual guests. In databases, you can tell big differences in SSD storage vs HDD storage (not considering memory), especially with disk I/O-intensive backups & restores.

    So these are just some use cases that can highlight the significant impact of better-performing components. But all cases are different in their usage scenarios. We can't just give up on seeking better performance & go only for low-cost just because our particular use-case doesn't really display the impact of higher-performing parts.
  • PeachNCream - Thursday, July 19, 2018 - link

    I think the point of the post you were responding to was to argue that there is little real world difference in most home computing scenarios between different SSDs (not SSDs vs HDDs as your response points out) despite the benchmarks showing sometimes dramatic performance advantages or disadvantages. The other accompanying parts of the post that wander into the CPU and GPU were just there to enhance the credibility of the claim by offering an argument of broad experience. Reply
  • romrunning - Thursday, July 19, 2018 - link

    I, too, wish for more performance MLC products; however, I think we might just transition to Optane drives & hope they go down faster in price. Reply
  • The_Assimilator - Thursday, July 19, 2018 - link

    Article title: "A Next-Gen NVMe SSD Controller"

    Article conclusion: " doesn't appear that the combination of the E12 controller and 64L TLC is at all future-proof."
  • shabby - Thursday, July 19, 2018 - link

    Need a catchy title for them clicks. Reply
  • Holliday75 - Thursday, July 19, 2018 - link

    It is Phison's next gen controller. Reply
  • tygrus - Thursday, July 19, 2018 - link

    28nm for the controller isn't new. I assume they could decrease power consumption if they used smaller fab node like 22nm. The FLASH chips themselves use/produce a lot of heat as well and 28nm might be cheaper than 22nm so compromises were made. Maybe they could bring out another version for higher performance AIC PCIe x8 or PCIe v4 signals next year (22nm fabed controller). Reply

Log in

Don't have an account? Sign up now