Announcement

Collapse
No announcement yet.

Matrox HD benchmark.

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Matrox HD benchmark.

    Hey Guys,

    How reliable is the HD benchmark included with the matrox video tools?
    I only ask because I have a Seagate cheetah on a 2940uw which according to This benchmark is considerably slower than my Seagate medalist UDMA drive.
    I have another benchmarking utility which gives more expected results, i.e. The scsi disk show's an average of about 17mb a second and the Udma drive show's an average of around 5mb a second.

    Is it something to do with the way that video capture writes to the drive?

    *confused*

    Cheers.

    Trav

  • #2
    I'm confused as well.
    I 've used the benchmark my self using a Quantum Atlas II differential drive 9gb. it's supposed to be AV drive. I sometimes get a (4-5) MB/s benchmark on this drive and my Fujitsu dma drive a 3 mb/s.

    I don't think it's accurate at all. According to the specs of my drive I should get better performance on this drive. All my settings for this drive on the controller are 40mb etc.
    I don't get any drop frames even capturing at max settings w/ RR 704x480 6:1 capture.

    any suggestion on other HD benchmark utilities out there or Sug.

    Thanks for posting this ?
    Good luck

    Comment


    • #3
      My IBM 10Krpm 18.2GB U2 SCSI HDD according to Matrox's benchmarked was slower my friends Quantum Bigfoot 4.5Krpm 13GB IDE HDD (which had DMA on). I got the benchmark to be more realistic by turning on write cache on my Adaptec 2940U2W controller which you can turn on permantly if you are using BIOS 2.20 or you can use Adaptec EZ-SCSI to turn on write cache.

      Before
      IDE 5MBps
      SCSI 4.5MBps
      after
      SCSI 13.5MBps

      Comment


      • #4
        The Matrox HD Benchmark is notoriously unreliable, and should not be taken as a true value of your HDs potential. Note the word "potential", because unless you have the right settings tweaked it is very possible that you are not making the most of your drive.

        The main purpose of the HDTest is to confirm that your HDs are capable of writing sustained datarates. The values are written into registry and used to determine what selections of frame size/rate and compression are available to you. As long as you are able to select full frame (704*576/704*480) and full rate (25fps PAL/30fps NTSC) and you can check 6.6:1 compression, then you don't have a problem. HD doesn't skow your drives to its own calculation or anything.

        If you want a fuller analysis of drive performance you can use several free/shareware utilities. Try Scisoft Sandra for starters

        Comment


        • #5
          That's SISOFT's SANDRA ;-)

          At any rate it will give results much closer to reality than thos of the HDBenchmark. SANDRA can be found on most D/L sites and it includes much more than just a drive test. I liked it enough to go to the Professional version.

          At any rate if the HDBenchmark defines a rate of at least 3.2 mB/s you will be able to use all your MJPeg settings.

          The rub comes when trying to use RGB captures. Your drive may be capable of them but PC-VCR won't present all the possible options if HDBenchmark misreports.

          There is a fix: run HDBenchmark so the registry entries are created, then change them by plugging in SANDRA's numbers ;-)

          Dr. Mordrid

          Comment


          • #6
            I had a 17gig UDMA33 HD installed and ran SiSoft Sandra on it prior to the installation of a Promise Fasttrak IDE RAID card/additional 17gig drive.
            After setting up the Fasttrak RAID as Striped Sandra reported no speed increase while Matrox HDbench reported a 2x increase.

            I had timed a few file copy operations prior to and after the FastTrak install and the results of these were consistent with the Matrox bench results.
            I am suspect of Sandra's and HDbench's ability to test HD speed although in my case HDbench does seem to be the most accurate.

            [This message has been edited by RealTime (edited 04 November 1999).]

            Comment


            • #7
              In SANDRA did were you checking the composite numbers or did you dig into the list for the sustained sequential write result?

              Also; I'm using 5.30 Professional (full version)

              Dr. Mordrid


              [This message has been edited by DrMordrid (edited 05 November 1999).]

              Comment


              • #8
                Sustained results. I am using Sandra 99.3.5.0 Professional. I just ran it again and am getting results closer to what I would expect Sequential Read = 18MB/s Sequential write = 20MB/s . I'm now convinced that I screwed up . I'm now thinking that I was looking at my system drive results. In my defence, It was late...

                Comment

                Working...
                X