Announcement

Collapse
No announcement yet.

F2FS With Linux 5.10 Brings Many Improvements And A Few More Features

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

  • F2FS With Linux 5.10 Brings Many Improvements And A Few More Features

    Phoronix: F2FS With Linux 5.10 Brings Many Improvements And A Few More Features

    The Flash-Friendly File-System (F2FS) that is of growing prominence on Android-powered mobile phones and other flash-storage-only Linux systems has some promising improvements with Linux 5.10...

    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
    I wonder if they will be able to increase the FS size beyond 16 TB, as that is already too small for larger zoned devices. At least it's on their TODO list, so hopefully there will be progress.

    Comment


    • #3
      Originally posted by Space Heater View Post
      I wonder if they will be able to increase the FS size beyond 16 TB, as that is already too small for larger zoned devices. At least it's on their TODO list, so hopefully there will be progress.
      Back to the OP:
      > "... growing prominence on Android-powered mobile phones and other flash-storage-only Linux systems has some promising improvements with Linux 5.10."

      Sixteen terabytes, on Android mobile phones is not enough?
      Old timers here are amazed at the hopes & expectations of baby operating systems, including the first Linux. "Baby" has outpaced every other operating system. Concepts now exist which were totally unimaginable just ten years ago. Another ten years ... again must be unimaginable.

      Ten years from now, the "normal" people will be surprised how simple minded they were in 2020. Perhaps we will move from alphanumeric coding to another more advanced biocarbon coding?

      Comment


      • #4
        Originally posted by gregzeng View Post

        Back to the OP:
        > "... growing prominence on Android-powered mobile phones and other flash-storage-only Linux systems has some promising improvements with Linux 5.10."

        Sixteen terabytes, on Android mobile phones is not enough?
        I'm not talking about mobile phones, f2fs' log-based design makes it not only ideal for NAND but also zoned storage devices in general.

        Back to my post (with emphasis added):
        Originally posted by Space Heater View Post
        I wonder if they will be able to increase the FS size beyond 16 TB, as that is already too small for larger zoned devices.
        f2fs is currently the only upstream filesystem that supports zoned storage drives such as HM-SMR HDDs, btrfs is getting support eventually but probably not for a few more releases. Western Digital is selling 20TB HM-SMR drives right now, which means that today no linux filesystem can natively use the entire drive (e.g. not with dm-zoned).
        Last edited by Space Heater; 16 October 2020, 02:40 AM.

        Comment


        • #5
          Originally posted by gregzeng View Post

          Sixteen terabytes, on Android mobile phones is not enough?
          I use it on the desktop with 970 EVO in order to unload disk's controller with write sector juggling as much as possible and lenghten the drive life.
          It works great. But it would be nice to see it on systems with more than 16TB...

          Comment


          • #6
            I would be happy to see F2FS functionality being a superset of NILFS2 functionality. I do not know if that is (relatively) easily technically achievable.

            Comment


            • #7
              Originally posted by Old Grouch View Post
              I would be happy to see F2FS functionality being a superset of NILFS2 functionality. I do not know if that is (relatively) easily technically achievable.
              It depends on low-level implementations. IIRC F2FS is less strict about never writing to the already written location, which means amongst other things, longer mount time on big medium that has seen a lot of writes - system has to go through all the writes in order to find latest generation of particular file.
              If it can't do writes back to already written location, it has to sacrifice some of the performance so that it distributes its write load as much as possible.

              Comment

              Working...
              X