Announcement

Collapse
No announcement yet.

AHCI vs. IDE Linux Performance Benchmarks

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

  • AHCI vs. IDE Linux Performance Benchmarks

    Phoronix: AHCI vs. IDE Linux Performance Benchmarks

    Hitting OpenBenchmarking.org this weekend are some interesting benchmarks comparing performance of AHCI vs. IDE modes under Linux from an AMD Fusion system...

    http://www.phoronix.com/vr.php?view=MTE0OTE

  • #2
    For an evolution of a standard it sure looks dissapointing from those benchmarks

    Comment


    • #3
      Suicide

      It is too bad that libata maintainer Andre Hedrick recently committed suicide.

      https://lwn.net/Articles/508222/

      Comment


      • #4
        Originally posted by uid313 View Post
        It is too bad that libata maintainer Andre Hedrick recently committed suicide.

        https://lwn.net/Articles/508222/
        Ouch...who's stepping forward to take his place, as this is an important library?

        Comment


        • #5
          I suppose this is a typical 5400 rpm HDD so it's kind of expected that it doesn't make much difference. Try the same thing with a fast SSD and see if the results are the same.

          Comment


          • #6
            Originally posted by DeepDayze View Post
            Ouch...who's stepping forward to take his place, as this is an important library?
            I don't know. Perhaps Jeff Garzik?

            https://lwn.net/Articles/508467/

            Comment


            • #7
              Originally posted by devius View Post
              I suppose this is a typical 5400 rpm HDD so it's kind of expected that it doesn't make much difference. Try the same thing with a fast SSD and see if the results are the same.
              I agree with that

              Comment


              • #8
                With disk drives, AHCI is not so much about performance, but about SATA (like hotswapping.) With fast disks, there's a bit of a performance advantage with non-sequential reads due to NQC, but it's not much (it's limited compared to SCSI.) However, it also helps the drive's read head live longer as it travels less.

                Note that SSDs don't have heads, so NQC is useless here. Access to the higher bandwidth however, is useful for fast SSDs.

                Comment


                • #9
                  Your description is wrong, first it's NCQ and it is not limited to reads:

                  http://en.wikipedia.org/wiki/Native_Command_Queuing

                  Comment


                  • #10
                    Originally posted by devius View Post
                    I suppose this is a typical 5400 rpm HDD so it's kind of expected that it doesn't make much difference. Try the same thing with a fast SSD and see if the results are the same.
                    You've got it backwards. For slower drives, NCQ should help a lot, as heads won't have to seel as much. However, in my experience the difference is no noticeable.
                    For SSDs, NCQ means nothing.
                    Other AHCI improvements are not about speed (afaik).

                    Comment


                    • #11
                      Originally posted by Kano View Post
                      Your description is wrong, first it's NCQ and it is not limited to reads
                      No, it was spot on. These were just typos since I wrote it very fast and I didn't keep track of the latest SSD developments ;-)

                      Comment


                      • #12
                        In my experience using AHCI makes a huge impact once you use software RAID devices - especially when it comes to responsiveness under load.
                        In IDE mode, heavy IO would have simply bogged down my dual Xeon machines.
                        w/ AHCI, even while running a tar backup w/ pbzip2, I can still use the machine.

                        ... Though to be honest, I never took the time to benchmark it.

                        - Gilboa
                        DEV: Intel S2600C0, 2xE52658V2, 32GB, 4x2TB + 2x3TB, GTX780, F21/x86_64, Dell U2711.
                        SRV: Intel S5520SC, 2xX5680, 36GB, 4x2TB, GTX550, F21/x86_64, Dell U2412..
                        BACK: Tyan Tempest i5400XT, 2xE5335, 8GB, 3x1.5TB, 9800GTX, F21/x86-64.
                        LAP: ASUS N56VJ, i7-3630QM, 16GB, 1TB, 635M, F21/x86_64.

                        Comment

                        Working...
                        X