Random Read Performance

One of the major changes in our 2015 test suite is the synthetic Iometer tests we run. In the past we used to test just one or two queue depths, but real world workloads always contain a mix of different queue depths as shown by our Storage Bench traces. To get the full scope in performance, I'm now testing various queue depths starting from one and going all the way to up to 32. I'm not testing every single queue depth, but merely how the throughput scales with the queue depth. I'm using exponential scaling, meaning that the tested queue depths increase in powers of two (i.e. 1, 2, 4, 8...). 

Read tests are conducted on a full drive because that is the only way to ensure that the results are valid (testing with an empty drive can substantially inflate the results and in reality the data you are reading is always valid rather than full of zeros). Each queue depth is tested for three minutes and there is no idle time between the tests. 

I'm also reporting two metrics now. For the bar graph, I've taken the average of QD1, QD2 and QD4 data rates, which are the most relevant queue depths for client workloads. This allows for easy and quick comparison between drives. In addition to the bar graph, I'm including a line graph, which shows the performance scaling across all queue depths. To keep the line graphs readable, each drive has its own graph, which can be selected from the drop-down menu.

I'm also plotting power for SATA drives and will be doing the same for PCIe drives as soon as I have the system set up properly. Our datalogging multimeter logs power consumption every second, so I report the average for every queue depth to see how the power scales with the queue depth and performance.

Iometer - 4KB Random Read

While the other SSDs hover at 60-90MB/s for random reads, the SM951 provides a rather noticeable upgrade at 108MB/s. 

Samsung SM951 512GB

Looking at the performance more closely reveals that the SM951 delivers better performance at all queue depths, although obviously the difference is at high queue depths where the SM951 can take advantage of the faster PCIe interface. The SM951 actually does over 150K IOPS when the MB/s is translated into throughput.

 

 

 

Iometer - 4KB Random Write

Random write performance is equally strong. The line graphs shows how the SM951 shifts the whole curve up, implying a performance increase at all queue depths. Especially the performance at queue depths of 1 and 2 are noticeably better than on other drives.

Samsung SM951 512GB
AnandTech Storage Bench - Light Sequential Performance
Comments Locked

128 Comments

View All Comments

  • Kristian Vättö - Tuesday, February 24, 2015 - link

    PCIe adapters are fine for review purposes and are in fact more easily serviceable than M.2 slots tend to be.
  • Samus - Tuesday, February 24, 2015 - link

    Thank you for pointing out all capacities are M2 2280 DOUBLE SIDED. That is missing from virtually all reviews, sales material and eCommerce sites for M2 drives. Newegg has been pretty good about taking pictures of both sides of many drives, though.

    This is important to me because I work with HP Elitebook's and the newest generation of the Elitebook 810 (G2) and 820\1040 all require single-sided. The 840 has room for M2 DS modules, though.
  • Flash13 - Tuesday, February 24, 2015 - link

    The company is not trustworthy! Buy at your own risk. Good Luck.
  • youtard - Tuesday, February 24, 2015 - link

    hurr!
  • Wardrop - Tuesday, February 24, 2015 - link

    Does this form factor work in standard desktop PCI-e 4x slots?
  • DanNeely - Tuesday, February 24, 2015 - link

    It's an m-2 plug, you need an adapter to fit it in a standard pcie slot.
  • wtallis - Wednesday, February 25, 2015 - link

    When discussing NVMe, please make it clear that the chipset and motherboard firmware only matter for booting off the drive; they don't need any updates to allow a compatible operating system to access the drive as something other than the boot volume.

    (As for what the motherboard firmware needs to gain in order to allow booting from NVMe devices, it's just a loadable UEFI device driver. Even if your motherboard doesn't have such a driver built-in, you could load it from some other storage device and then boot off the NVMe drive.)
  • ericgl21 - Wednesday, February 25, 2015 - link

    I wonder if Samsung (or any other OEM) would be willing to sell a 1TB m.2 NVMe PCIe3.0 x4 SSD with SLC NAND with a 2280 form factor?
    Many professionals would appreciate the speed and reliability that SLC NAND provides.
    Sure, it would cost a lot, but so do the Samsung SM951 and Intel P3700.

    If that's not possible with current 16nm manufacturing, then a 512GB would also be nice. :-)
    Just my 2 cents.
  • baii9 - Wednesday, February 25, 2015 - link

    reliable on nand? Reliable controller matters, high endurance nand matters, I think nand are "reliable' enough already(compare to that lovely controller).
  • IlikeSSD - Wednesday, February 25, 2015 - link

    looks like Samsung paid for not showing OCZ in consistency and mixed workload tests )))))

Log in

Don't have an account? Sign up now