Comments Locked

13 Comments

Back to Article

  • alkalinetaupehat - Sunday, August 15, 2010 - link

    It's known and understood that the smaller-capacity versions in a line of SSD's will perform slower, so would it be possible to include those results in the SSD bench to illustrate the differences? I'm particularly interested since I just picked up the 64GB version of the Crucial C300.
  • 7Enigma - Wednesday, August 18, 2010 - link

    Agreed, I think a lot of us are still using these more as boot drives due to cost (80GB Intel G2 for me), and so it skews the results a bit if I'm looking at the 160GB data (20% faster sequential write for example).
  • semo - Sunday, August 15, 2010 - link

    Look at that lonely HDD at the bottom, trailing even jmicron drives!

    Wonder when the new gen SSD will come around in decent stock levels so that we can get this HDD BS out of the way already. Us tech folks have suffered enough progress bars and hour glasses.
  • todlerix - Sunday, August 15, 2010 - link

    add in a 60GB short stroke version of the v-raptor ;)
  • FunBunny2 - Sunday, August 15, 2010 - link

    I neglected to add on the original post.

    The ultimate use-case for SSD is in RDBMS, specifically fully normalized schemas. The test would be kind of simple to execute:
    using a standard DB (SQL Server in the windows world), with minimum buffers in the database (forcing hard I/O to the maximum extent), compare the SSD doing a cross join of two 1,000 row tables and the HDD doing a table scan of the 1,000,000 row single table. The data should be large enough to matter, say a 1k row of mixed numeric, string, and date data.

    The server and HDD would be the same across SSDs tested. Since the CPU needs to synthesize the rows from the join, the engine configuration should be reasonable as an implementation of the database in question. It shouldn't be naive' out-of-the-box on the one hand, nor fully hand tuned on the other. A configuration to be expected from an experienced professional with the database (again, SQL Server in this example).

    The problem with published tests (not just those here) is that they relate to application loading and OS startup. Not really relevant to serious SSD use.
  • numberoneoppa - Sunday, August 15, 2010 - link

    Though they're incredibly relevant to 99% of readers.
  • Hrel - Sunday, August 15, 2010 - link

    I'd REALLY REALLY like to see results added for a mechanical disk stripped RAID set up thrown in there. Like two Seagate/WD 7200rpm 32MB cache drives in stripped RAID. Also Seagates MomemtusXT drive thrown in there. THANKS!
  • marraco - Monday, August 16, 2010 - link

    Ad, I add, Use RAID partitions with the same size that common SSD: 80 Gb, 120 GB, 180 Gb. Those small partitions speed up the read write speed, and HDD latencies.

    The name of the trick is short-stroke.

    It do not gives the same advantages than SSD, but enhance HDD performance, meanwhile allowing to have additional storage on other partitions.

    I made a 120 GB RAID 0 on first partition on a Samsung Spinpoint 1 Tb, and the rest of the disk is in RAID1.
  • mapesdhs - Monday, August 16, 2010 - link


    I've added a couple of RAID0 results to my page:

    http://www.sgidepot.co.uk/diskdata.html

    2 x 1TB 7200rpm Samsung SpinPoint F3 SATA gives 291MB/sec max.

    2 x 450GB 15K Hitachi (HP) HUS154545VLS300 SAS gives 294MB/sec max.

    For reference, one Samsung F3 gives 145MB/sec, while one Hitachi gives
    163MB/sec (faster than the WD VR 600GB).

    I've been testing with HDTach, but I can use Iometer instead if anyone's interested.

    Next week I'll be able to test with more than 2 Hitachis. Sorry I can't test atm with
    more than 2 Samsungs, am only using XP so it's limited by the 2TB OS issue. 4 x
    Hitachi will work ok though.

    Ian.
  • thorr2 - Monday, August 16, 2010 - link

    It would be great to see all of the drives in a RAID-1 setup to see how the performance compares. Purchasing two cheap SSD's may allow for much more storage and better performance than a single high-end SSD. Since this would mainly be used for the OS and applications, it is not that big of a deal if you lose everything (assuming you keep any important data on other mirrored or RAID-5+ drives).
  • white2011 - Friday, August 20, 2010 - link

    Good job Anand, THANKS !
    Can you add for hard disk beside WD like seagate 7200.12 and others too ? thanks
  • Joffer - Friday, August 20, 2010 - link

    What about adding a Hybrid disk - Seagate Momentus XT
  • pcunite - Wednesday, September 1, 2010 - link

    Test SSD serving mySQL / Apache application

Log in

Don't have an account? Sign up now