Announcement

Collapse
No announcement yet.

USB 2.0 vs. USB 3.0 Flash Drives On Linux

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

  • USB 2.0 vs. USB 3.0 Flash Drives On Linux

    Phoronix: USB 2.0 vs. USB 3.0 Flash Drives On Linux

    With the current Linux USB stack and file-systems, do USB 3.0 flash drives provide much of a performance gain over USB 2.0 flash drives? In this article are some brief benchmarks from USB 2.0 and USB 3.0 Corsair Flash Voyagers.

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Meanwhile, Windows 7 SP1 does not support USB 3.0 out of the box.

    Comment


    • #3
      Michael, how about trying the USB 3.0 stick in a USB 2.0 slot on that motherboard and seeing what difference just the interface makes?

      Comment


      • #4
        Originally posted by willmore View Post
        Michael, how about trying the USB 3.0 stick in a USB 2.0 slot on that motherboard and seeing what difference just the interface makes?
        Which might also help differentiate the underlying hardware between the two models. USB3.0 might require faster onboard processing due to having to handle more data quickly, so inherently gives a slight (or massive?) advantage over USB2.0. I really need to test this myself, especially in RAID. There was a great little article floating around the internet too about RAIDing USB (in my case was for HTPC, and wanted local transcoding otions with the discs being the bottleneck)

        But I lost it.
        Hi

        Comment


        • #5
          Wow. This was one of the worst benchmarks in Phoronix history. The title should be Corsair X vs Corsair XX. If you really wanted to test the difference between USB 2.0 and 3.0 on linux you should have gotten a USB3.0 HDD dock, put a very fast SSD in there and first connect it to a USB 2.0 port, and then to a USB 3.0 port. That would present meaningful results instead of completely irrelevant results.

          Comment


          • #6
            Originally posted by devius View Post
            Wow. This was one of the worst benchmarks in Phoronix history. The title should be Corsair X vs Corsair XX. If you really wanted to test the difference between USB 2.0 and 3.0 on linux you should have gotten a USB3.0 HDD dock, put a very fast SSD in there and first connect it to a USB 2.0 port, and then to a USB 3.0 port. That would present meaningful results instead of completely irrelevant results.
            Except, it would only show the throughput for the specific drive in the specific USB device plugged into the motherboards specific controller. Just as he has done but with something we all use. This is a coffee break article and I seriously doubt it was supposed to be comprehensive and as an intelligent individual you possibly could have picked that up. Your response is unnecessarily aggravated.
            Hi

            Comment


            • #7
              Having read an article in Germany's c't magazine (a serious one) just yesterday there were tests of USB HDDs. Here it was noticed that there were a lot of problems still with USB 3. It is a complex interaction of device, USB-chipset and sadly also cables. Cables still seem to be an issue here. (Of course cables can always be but there are still low quality cables sold as it seems, even delivered in packages with devices).
              There was one example where an external HDD would only reach USB 3.0 speed on a NEC controller chip but not on any other controller.
              So in reality your mileage may vary quite a bit (if your USB connection becomes the bottleneck).
              Stop TCPA, stupid software patents and corrupt politicians!

              Comment


              • #8
                Originally posted by Adarion View Post
                Having read an article in Germany's c't magazine (a serious one) just yesterday there were tests of USB HDDs. Here it was noticed that there were a lot of problems still with USB 3. It is a complex interaction of device, USB-chipset and sadly also cables. Cables still seem to be an issue here. (Of course cables can always be but there are still low quality cables sold as it seems, even delivered in packages with devices).
                There was one example where an external HDD would only reach USB 3.0 speed on a NEC controller chip but not on any other controller.
                So in reality your mileage may vary quite a bit (if your USB connection becomes the bottleneck).
                That's pretty much been my experience with USB3.0 as well. It just doesn't seem to be reliable insofar as having a close range of transfer speed ie 20-25Mb/s as opposed to the 10-50Mb/s I'm seeing. Are there any sources you know of with brands that utilise good controllers?
                Hi

                Comment


                • #9
                  Sandisk Extreme 64 GB at 183 MByte/s

                  I have an SSD Samsung 256GByte, Kubuntu 12.10 and a Sandisk 64 GByte Extreme and over Usb 3. I am getting extraordinary results like 183 MByte/s on large files copy. E.g. copying a virtualbox appliance.

                  Really miss Virtual Box having usb 3 support because my laptop has no usb 2 ports.

                  Comment


                  • #10
                    Originally posted by stiiixy View Post
                    Are there any sources you know of with brands that utilise good controllers?
                    No, not really. For my own part I don't even have USB 3 (yet). As far as I know the AMD USB 3 builtins (in their SBs) are quite performant. Other than that there ought to be NEC, Renesas, probably VIA, intel, TI and others. In the report of the c't (computer and tech) magazine it was more seen as the fault of the USB/SATA bridges used on the external HDDs. I could have a look at the magazine again and check which caused problems and which were good.
                    Thing is: some brands use whatever they can buy cheap off the market so internals in the housing might change. Especially cheaper models might change internals. Things like that can be nasty if you check Linux compatibility and it seems to work but next month they use a different chip or something and things break.
                    Iirc c't even mentioned some HDD people using discs from the competition.
                    Stop TCPA, stupid software patents and corrupt politicians!

                    Comment

                    Working...
                    X