TRIM Performance

Our IO consistency tests give a good idea of the drive's garbage collection but for TRIM testing we still rely on the good old HD Tach method. To begin the test, I ran HD Tach on a secure erased drive to get the baseline performance:

Next I tortured the drive for 30 minutes with 4KB random writes (QD=32, 100% LBA) and TRIM'ed the drive after the torture:

TRIM works as expected, although there is some weird behavior at the 50% mark. It seems like the Hawk features a performance mode similar to OCZ's Vector and Vertex 4. The idea is rather simple: you only write to half of the pages until they are full. The gain is lower latencies as your drive is basically operating with 50% over-provisioning, but the downside is that the pages need to be reorganized once more than half of the drive has been filled. In the graph the write speed drops to ~100MB/s but this is because the drive is organizing the pages while it's being written to, which results in quite bad performance. After I had run HD Tach, I let the drive idle for about 10 minutes and IOmeter showed that the write speed had returned to +300MB/s. However, it did drop to ~100MB/s quite quickly so the reorganizing might take a while (with Vector it only took a few minutes) and I would advice to let the drive idle for an hour or so if you fill more than half of it.

AnandTech Storage Bench 2011 - Light Workload Power Consumption
Comments Locked

26 Comments

View All Comments

  • Kristian Vättö - Tuesday, June 25, 2013 - link

    That's different. Apple doesn't hide the fact that they've updated the iPad, the least they do is release a press release. I agree that calling third-gen iPad the "iPad 3" and then going back to just "iPad" wasn't the best move from Apple, but it was understandable as the naming would quickly get stupid (imagine e.g. iPad 11). Apple has always been doing that with the MacBook Pros for instance.

    Strontium did nothing to let the public know that they've changed suppliers, hence the suspicion. No press release, no mention in the product page, nothing. Like I said in the article, we are the first to report about this - I even thought I'm reviewing a SandForce drive until I took the drive apart.
  • gamoniac - Tuesday, June 25, 2013 - link

    Great review. Not a bad drive at all. Unfortunately it is overshadowed by the GB vs GiB and Hawk vs Python issues. I don't mind owning one if reliability is their focus. Now the other deciding factors are warranty and customer service. If they can build up a good reputation, they will have a good following.
  • toytanks - Wednesday, June 26, 2013 - link

    I just purchased a the Samsung 840 Pro 256GB. Does anyone know how much of a difference it would make to use this instead, in terms of battery life?
  • LB-ID - Wednesday, June 26, 2013 - link

    Absolutely not, and you're a fool soon parted from your money if you do.
  • zjozjom - Wednesday, June 26, 2013 - link

    for all you thinking samsung 840 pro the best drive, getting brain washed only to see the destroyer benchmark owns it =(

    yeahhh look at the inconsistent write for sandisk extreme II under hd tach yet it destroys samsung 840 pro due to ncache LOL gj sandisk, samsung stop brain wash people thanks, people stop being retards thanks yeah
  • hemlock44 - Thursday, June 27, 2013 - link

    Should they ever plan to sell in Italy, I strongly suggest a name change, because "Strontium" recalls a scatological term here :-)

Log in

Don't have an account? Sign up now