Announcement

Collapse
No announcement yet.

You were warned so don't get mad at me.

Collapse
This topic is closed.
X
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    I agree on the limited space. I recently bought a 30GB OCZ and a 64GB Kingston just to play with - the one machine that really needs the performance is the Windows box, and that is rapidly approaching half a terabyte with all the games and crap that I keep on it.

    I did test installs of both Windows and Linux to see whether the performance was really there - and it is, though Windows likes to insert pauses for the hell of it, no matter how fast the disk is.

    However, I do see that there are beginning to appear "booster" units, to which you attach a normal hard disk and an SSD. The booster then caches part of the hard disk to the SSD, and thus gains a pile of performance without losing the capacity. The one I've seen on sale isn't very sophisticated, but I'm sure that will improve, especially if someone tinkers with it.

    I also strongly agree that you must go for quality. The biggest trick is to find out which controller is used. Outside of the Intel drives, the Indilinx controller is the one to go for in general, but the new JMicron 61x series (which my Kingston uses) is perfectly respectable. The Kingston drives are remarkably cheap at about €2 per GB, the Indilinx-based drives are about twice that.

    The Samsung and the old JMicron 602 controllers are to be avoided. So is the Toshiba controller, which is a rebadged 602 with tweaked firmware. The Samsung is apparently reliable, but a very poor performer.

    The Sandforce and Crucial/Micron controllers are as yet unproven, so avoid them for the next few months if you like reliability, but they should be excellent choices for performance once they've worked the bugs out.

    Comment


    • #17
      SSDs

      Quoth Chromatix View Post
      I recently bought a 30GB OCZ and a 64GB Kingston
      How long ago were they made? Even six(6) months makes a big difference in the firmware used inside the drive.
      Please read: http://www.anandtech.com/storage/showdoc.aspx?i=3531 and http://www.anandtech.com/storage/showdoc.aspx?i=3631 if you have not before.

      Windows likes to insert pauses for the hell of it, no matter how fast the disk is.
      The pauses are probably because the type of SSDs/firmware you use, it is explained why in the above reviews and how to avoid the problems.

      beginning to appear "booster" units, to which you attach a normal hard disk and an SSD.
      I think it is better to have a good SSD as your boot drive and a good fast Hard Drive as your data/work drive, if one fails the other is untouched.

      I also strongly agree that you must go for quality. The biggest trick is to find out which controller is used. Outside of the Intel drives, the Indilinx controller is the one to go for in general, but the new JMicron 61x series (which my Kingston uses) is perfectly respectable. The Kingston drives are remarkably cheap at about €2 per GB, the Indilinx-based drives are about twice that.

      The Samsung and the old JMicron 602 controllers are to be avoided. So is the Toshiba controller, which is a rebadged 602 with tweaked firmware. The Samsung is apparently reliable, but a very poor performer.

      The Sandforce and Crucial/Micron controllers are as yet unproven, so avoid them for the next few months if you like reliability, but they should be excellent choices for performance once they've worked the bugs out.
      Oh, did you read the reviews already? The above suggests you already know what I was trying to point out. But the pauses you see with windows should not be there if your drives handles small writes properly.
      Last edited by earl colby pottinger; 03-10-2010, 05:00 AM.

      Comment


      • #18
        When I mentioned I work with SSD's, I don't buy them.

        I build them.



        I work for one of the companies mentioned above. I'm not going to say which one though.

        And I would not personally use an SSD. At least not yet. Give it a couple years for the technology to mature, then I'll be all over them. Its still in the very early stages for that technology. In terms of reliability a HDD is far beyond an SSD, with a vastly lower failure rate.

        SSD failure rates are in the double digits, by the way. HDD's? Maybe 1-2%.

        Comment


        • #19
          Why SSDs

          Quoth Hyndis View Post
          When I mentioned I work with SSD's, I don't buy them.

          I build them. I work for one of the companies mentioned above.
          SSD failure rates are in the double digits, by the way. HDD's? Maybe 1-2%.
          Just because you work for a company that does not build good SSDs does not mean you have to buy your SSDs from them. Get a good one from one of your competitors.

          Bet you can't find a URL showing the failure rates you claim for any of the better models.

          page 3 of http://download.intel.com/it/pdf/Ent...ate_Drives.pdf is of interest here.

          PS. While we are getting off-topic I really would like to see an URL to high failure (double digits) as I never seen any given for the Intel drive I bought.
          Last edited by earl colby pottinger; 03-10-2010, 04:33 PM.

          Comment


          • #20
            You won't find it widely published. Its the dirty little secret of SSD's.

            Yes, the MTBF is extremely high, but the actual failure rates tend to be far higher due to issues with the controller chips and/or firmware failures.

            With a HDD you usually get some warning if there is mechanical failure, such as the infamous clicking noise. When an SSD goes there is almost never any warning. Its just gone. Everything on it gone unless you have a backup.

            Comment


            • #21
              Published rates????

              Quoth Hyndis View Post
              You won't find it widely published. Its the dirty little secret of SSD's.

              Yes, the MTBF is extremely high, but the actual failure rates tend to be far higher due to issues with the controller chips and/or firmware failures.

              With a HDD you usually get some warning if there is mechanical failure, such as the infamous clicking noise. When an SSD goes there is almost never any warning. Its just gone. Everything on it gone unless you have a backup.
              Hard to believe, not the causes but the rates you are trying to claim.

              There are all sorts of on-line news sites looking for dirt to pull in readers, take another look at the first review of SSDs, he really slagged some of the drives.

              Your claim basic come down to "It's a secret, but trust me!", sorry if it was true then there are too many competitive companies who have a lot to gain presently by exposing the facts.

              They were not able to hide the access speed slow-down seen in early drives, it was big news, so how are they hiding a high failure rate?

              Again, stop looking at your own companies product, too many data base companies who do not build/sell these drives are buying and testing them in bulk for this to be hidden for long.

              See:
              http://www.google.ca/search?q=failure+rate+of+SSDs
              http://www.bit-tech.net/news/2008/03...te_unfounded/1
              http://hothardware.com/News/SSDs-Hav...-Failure-Rate/
              For your side of claims.
              http://www.forum.crucial.com/t5/Soli...ptops/m-p/6126

              But it seems the claim was first started by a company doing poor research and ran with it from there.

              Comment


              • #22
                Quoth earl colby pottinger View Post
                How long ago were they made? Even six(6) months makes a big difference in the firmware used inside the drive.

                The pauses are probably because the type of SSDs/firmware you use, it is explained why in the above reviews and how to avoid the problems.
                On the contrary, these are pauses with the hard disk light off. They also occur at times when no writes are (or at last should) be happening, and I had precisely zero disk performance troubles with the same drive under Linux.

                As far as I can tell, the firmware doesn't make as much of a difference as the underlying controller design. The one exception would be TRIM support, which in turn is only relevant if the OS uses it (which WinXP doesn't) and the disk has ever been nearly full.

                Comment


                • #23
                  Rooted?

                  Quoth Chromatix View Post
                  On the contrary, these are pauses with the hard disk light off. They also occur at times when no writes are (or at last should) be happening, and I had precisely zero disk performance troubles with the same drive under Linux..
                  Are you sure you have not been rooted then?

                  I have always hated MS Windows. From Windows 3.11 to Windows 7 I have always noted the big pauses caused by writes to the hard drive, problems that never showed up on my Amigas, Beos or Haiku machines.

                  Windows in particular is always writing back data to the hard drive, then acting like it never buffered the data, resulting in the system pausing at random.

                  However, once I added a SSD to my netbook, as far as I can detect all that pausing went away. Note: I am using Window 7 Starter, maybe you have a lot more services running in the background to slow you down, but beware being rooted!

                  Comment

                  Working...
                  X