Random Read Performance

For full details of how we conduct our Iometer tests, please refer to this article.

Iometer - 4KB Random Read

Random read performance at low queue depths is decent, but not class-leading. Similar to what we saw in the Storage Bench tests, only Samsung has a notable advantage over other manufacturers.

Iometer - 4KB Random Read (Power)

Power consumption, on the other hand, is outstanding as the BX100 is the most efficient drive we have tested.

Crucial BX100 120GB

Since the SM2246EN is purely a client controller and fairly lightweight in general, it doesn't offer very aggressive scaling and tops out at 300MB/s, whereas many competing drives are able to achieve up to 400MB/s at the highest queue depths. Fortunately that's not relevant to client workloads, so the lack of high QD performance isn't that much of an issue.

 

Random Write Performance

Iometer - 4KB Random Write

Random write performance is also average, but the power consumption is again excellent and it should be noted that the power draw doesn't increase substantially with higher capacities (but neither does performance). 

Iometer - 4KB Random Write (Power)

Crucial BX100 120GB

Performance scaling at higher queue depths isn't that impressive, but the BX100 does scale very well at low queue depths and the performance at QD1 is very good, which is ultimately the most important queue depth for client workloads. 

AnandTech Storage Bench - Light Sequential Performance
Comments Locked

67 Comments

View All Comments

  • Hulk - Friday, April 10, 2015 - link

    Wow, that validates all of the other reviews I've been reading about this drive. Great performance, great value, and no TLC NAND worries. I'm getting one for my new laptop.
  • Uplink10 - Saturday, April 11, 2015 - link

    If you want a great value you better wait for a few months (maybe more than a few) till 3D NAND drives are going to come out from SanDisk, Micron, Intel, Toshiba. Then Samsing will not be the only company with 3D NAND drives and prices will probably going to come down.
  • Harry_Wild - Saturday, April 25, 2015 - link

    Thanks for the info! I will wait it out to get the best 3D NAND drives.
  • Hace - Friday, April 10, 2015 - link

    I'm kinda surprised you didn't draw more attention to the 850 EVO as a competitor, which is neck-in-neck with pricing.
  • digiguy - Friday, April 10, 2015 - link

    Yes, similar value for the money, and definitely more than the 840 EVO....
  • Kristian Vättö - Friday, April 10, 2015 - link

    I have to admit that I totally overlooked the 850 EVO as I was kinda assuming that its price would be higher. I put this review together on a very short notice as Ryan couldn't get the MacBook review finished for today, hence my mistake of not paying enough attention to the prices (even though I updated the table today...). Anyway, I've updated the conclusion to take the 850 EVO into account because as you said, the pricing is very close and it does provide a little higher performance.
  • Hulk - Friday, April 10, 2015 - link

    While the 850EVO is definitely a competitor for the BX100 the fact that it uses TLC NAND vs. MLC for the BX100 is a big deal for many people who aren't as yet convinced the 850EVO won't be affected by the read issue that is still a problem with the 840EVO. I realize the cell size for the 850 EVO's 3D NAND is much greater but as I wrote above many people, myself included, don't see the need to roll the dice on the 850EVO when the BX100 will provide basically similar performance.
  • just4U - Friday, April 10, 2015 - link

    The 850Evo is $20-$50 more on all models here in Canada. Not sure what stateside or Euro Pricing is like.
  • Margalus - Saturday, April 11, 2015 - link

    I just bought a 1TB EVO for $350 on Amazon, USA
  • repoman27 - Saturday, April 11, 2015 - link

    Kristian, hopefully you'll be helping Ryan with that MacBook review by covering the NVMe(!) SSD it ships with in depth. Looks to be a PCIe 2.0 x4 connected device dubbed "AP0256H", so possibly a semi-custom Apple controller based on the Marvell 88SS1093.

Log in

Don't have an account? Sign up now