Announcement

Collapse
No announcement yet.

LZ4 v1.9.4 Achieves 20~70% Speedups For Some CPUs & Configurations

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

  • LZ4 v1.9.4 Achieves 20~70% Speedups For Some CPUs & Configurations

    Phoronix: LZ4 v1.9.4 Achieves 20~70% Speedups For Some CPUs & Configurations

    LZ4 v1.9.4 is out today as the first point release in nearly two years for this BSD-licensed, speedy, lossless compression algorithm...

    https://www.phoronix.com/news/LZ4-1.9.4-Released

  • #2
    For filesystems that do transparent compression, do they use the upstream code or a copied one in their repos?

    Besides the direct benchmark, I would like to know if this improvements reach the end user indirectly by the lib consumers

    Comment


    • #3
      Originally posted by andrei_me View Post
      For filesystems that do transparent compression, do they use the upstream code or a copied one in their repos?

      Besides the direct benchmark, I would like to know if this improvements reach the end user indirectly by the lib consumers
      Linux kernel has its own vendored in copy of these algorithms. Usually they're pretty out of date, too. It'll probably take some time for these improvements to land.

      Comment


      • #4
        Can't wait for this to trickle down to zram!

        Comment


        • #5
          Default zram still uses lzo+rle, would be great to have some tests with different algorithms.

          Comment


          • #6
            Originally posted by oibaf View Post
            Default zram still uses lzo+rle, would be great to have some tests with different algorithms.
            I found some benchmarks on Reddit showing lz4 being lot faster than lzo+rle, it just compresses slightly less.
            ​​​​​​I wonder why anyone would choose this over lz4...

            Comment


            • #7
              Originally posted by andrei_me View Post
              For filesystems that do transparent compression, do they use the upstream code or a copied one in their repos?

              Besides the direct benchmark, I would like to know if this improvements reach the end user indirectly by the lib consumers
              OpenZFS is, for the most part, using upstream LZ4 based on version 1.9.3

              Comment


              • #8
                Originally posted by MastaG View Post

                I found some benchmarks on Reddit showing lz4 being lot faster than lzo+rle, it just compresses slightly less.
                ​​​​​​I wonder why anyone would choose this over lz4...
                You're using BRTFS and the BTRFS devs said "what's the point?" in regards to LZ4 because LZO was good enough.

                Comment


                • #9
                  Originally posted by skeevy420 View Post

                  You're using BRTFS and the BTRFS devs said "what's the point?" in regards to LZ4 because LZO was good enough.
                  Well I meant for use with zram.
                  Why would one choose for lzo+rle over lz4 while the latter being much faster with just slightly less compression.

                  Comment


                  • #10
                    Originally posted by skeevy420 View Post

                    You're using BRTFS and the BTRFS devs said "what's the point?" in regards to LZ4 because LZO was good enough.
                    They were most likely referring to zstd as good enough, rather than lzo. You can even specify a compression ratio using the compress=zstd:$ratio mount option (or use force-compress and get ~10% more compression). There is really no point in using lzo nowadays. For zswap/zram swap lz4 is much faster and for storage/zram file sytem zstd offers better compression at acceptable performance.

                    See also centos_st.pngO
                    Source https://linuxaria.com/article/linux-...-bzip2-vs-lzma

                    Comment

                    Working...
                    X