TRIM Performance

In practice, SandForce based drives running a mainstream client workload do very well and typically boast low average write amplification. However if subjected to a workload composed entirely of incompressible writes (e.g. tons of compressed images, videos and music) you can back the controller into a corner.

To simulate this I filled the drive with incompressible data, ran a 4KB (100% LBA space, QD32) random write test with incompressible data for an hour and a half, and then ran AS-SSD (another incompressible data test) to see how low performance could get:

Kingston HyperX 3K - Resiliency - AS SSD Sequential Write Speed - 6Gbps
  Clean After 1.5hr Torture After TRIM
Kingston HyperX 3K 240GB 312.4 MB/s 103.9 MB/s 245.8 MB/s

I usually run this test for only 20 minutes but after seeing an unusually resilient performance by the 240GB drives I decided to extend the test period to a full 90 minutes. Performance does drop pretty far at that point, down to 103MB/s. TRIMing the drive does restore some performance but not all. If you have a workload that uses a lot of incompressible data (e.g. JPGs, H.264 videos, software encrypted data, highly random datasets etc...) then SandForce just isn't for you.

Power Consumption

SandForce boasts fairly low power consumption, particularly at idle. Even with incompressible data the HyperX 3K's power draw is competitive:

Drive Power Consumption - Idle

Drive Power Consumption - Sequential Write

Drive Power Consumption - Random Write

 

AnandTech Storage Bench 2011 - Light Workload
Comments Locked

44 Comments

View All Comments

  • colonelclaw - Tuesday, April 10, 2012 - link

    Just out of interest, when you talk about filling the drive and then 'TRIMing the drive', how exactly do you do this? I thought TRIM was automatic (in the right OS), or is there some command that can be run to 'TRIM' a drive?
  • Coup27 - Tuesday, April 10, 2012 - link

    In Windows 6, TRIM is triggered when a file is actually deleted, so a format, empty the recycle bin or shift+del a file. When AT test SSDs they are connected as a storage drive, so it is easy to torture the drive and then TRIM it by formatting it within disk management.
  • Coup27 - Tuesday, April 10, 2012 - link

    ^ Windows 7.

    Aside from that, Intel and Samsung have toolbox programs which allow you to manually force a TRIM and GC pass.
  • colonelclaw - Tuesday, April 10, 2012 - link

    Thanks Coup!
  • hrrmph - Tuesday, April 10, 2012 - link

    BIG THANKS for including the information that it has a toolbox and that the toolbox doesn't support secure erase!
  • hechacker1 - Tuesday, April 10, 2012 - link

    What's Kingston's warranty on this? That lets me know about what they expect in terms of support.
  • itsmepj - Monday, May 28, 2012 - link

    fast is fast.. but not reliable..... i bough mine in Jan 2012... last week sent back for warranty, cannot detect..... all data gone.... yet replacement unit have to wait for 4-8 weeks time from Taiwan.... very bad backup service..... sad
  • killabee_me - Saturday, June 2, 2012 - link

    I would love to see Anandtech review Sandisk Extreme SSD 240GB .

    The word is that Sandisk managed to fix one of the biggest problems with Sandforce - the TRIM actually works 100% and the performance is back to the advertized one even after hours of torture tests with incompressible data.

    In addition the Extreme series use a very fast SanDisk's own 24nm Toggle Mode NAND and a custom optimized firmware, which results in one of the fastest SSD on the market. The price is extremely competitive as well.

    It would be interesting to see what it can do in AnandTech Storage Bench.
  • phillyry - Tuesday, August 7, 2012 - link

    Anand,

    Why have you assumed a write amplification of 10x ?

    Looking at xtremesystems dot org forums for SSD-Write-Endurance-25nm-Vs-34nm, it appears that 1x seems more common (with only the Samsung 470 showing 5x).

    Are these stress tests unrealistic and therefor somehow lowering write amplification?

    Could you please provide some insight as to how you came up with this figure.

    I thought that most SF and Micron based drives had write amp neer 1.0

    RSVP,

    phillyry
  • starcom - Tuesday, November 20, 2012 - link

    In the end, I buy the HyperX or HyperX 3K?

Log in

Don't have an account? Sign up now