compression-algorithms-benchmark.clj

#!/usr/bin/env boot

;; BOOT_CLOJURE_NAME=org.clojure/clojure
;; BOOT_CLOJURE_VERSION=1.8.0
;; BOOT_VERSION=2.5.5
;; BOOT_JVM_OPTIONS="-Xms1g -Xmx1g -XX:+UseG1GC -XX:+AggressiveOpts -server -Dclojure.compiler.direct-linking=true"

(set-env! :repositories #(conj % ["bintray" {:url "http://dl.bintray.com/nitram509/jbrotli"}]))

(set-env! :dependencies '[[criterium "0.4.4"]
                          [org.xerial.snappy/snappy-java "1.1.2.6"]
                          [com.github.luben/zstd-jni "1.1.0"]
                          [net.jpountz.lz4/lz4 "1.3.0"]])

(set! *warn-on-reflection* true)

(import 'java.util.Random)
(import 'org.xerial.snappy.Snappy)
(import 'net.jpountz.lz4.LZ4Factory)
(import 'net.jpountz.lz4.LZ4Compressor)
(import 'net.jpountz.lz4.LZ4FastDecompressor)
(import 'com.github.luben.zstd.Zstd)

(require '[criterium.core :refer [bench quick-bench with-progress-reporting]])

;; --- Helpers

(def data "
  Lorem ipsum dolor sit amet, consectetur adipiscing
  elit. Pellentesque bibendum neque ut feugiat vulputate. Donec non mi
  maximus, fermentum mi vitae, egestas orci. Aliquam lobortis, nunc sit
  amet aliquet auctor, eros lorem euismod ante, eu aliquet felis libero
  eget neque. Phasellus blandit, lacus ac rhoncus condimentum, nulla
  risus ultrices ex, et venenatis enim dolor nec velit. Nunc dapibus nec
  sapien eu tristique. Donec vel suscipit purus. Pellentesque mollis
  eros in hendrerit faucibus. Vestibulum ante ipsum primis in faucibus
  orci luctus et ultrices posuere cubilia Curae; Donec a eleifend
  odio. Morbi nec elementum quam. Donec gravida metus vel sem aliquam
  pellentesque. Donec hendrerit nisl in mauris dapibus, nec auctor
  mauris dignissim. Mauris ante mi, pretium nec interdum quis, maximus
  quis tortor. Lorem ipsum dolor sit amet, consectetur adipiscing
  elit. Curabitur non nibh bibendum, viverra enim luctus, commodo dolor.

  Proin aliquet, risus at convallis imperdiet, sem ligula facilisis
  ante, vitae consectetur justo massa quis velit. Fusce eu urna porta,
  pharetra felis in, malesuada augue. Quisque sit amet arcu a turpis
  blandit molestie. In ac lacus eu odio pulvinar faucibus. In vulputate
  at libero a tempor. Etiam eget neque aliquet, accumsan est quis,
  convallis tortor. Cum sociis natoque penatibus et magnis dis
  parturient montes, nascetur ridiculus mus. Nullam hendrerit erat quam,
  sed imperdiet dui efficitur eget. Sed elementum, eros sed dapibus
  lacinia, elit augue iaculis leo, at gravida tortor elit non
  mi. Praesent volutpat lacus molestie risus convallis
  malesuada. Curabitur sed ex eu felis venenatis auctor. Fusce convallis
  turpis ac nibh commodo luctus. Sed commodo dui non arcu mattis
  sagittis. Ut at porta nulla, in sagittis sapien. Duis lectus ligula,
  dignissim id dolor id, porttitor posuere tellus.

  Nulla sit amet venenatis tellus. Donec non enim mi. Vestibulum purus
  leo, dictum sed ligula quis, lacinia vehicula urna. Aenean at libero
  varius, semper urna at, placerat sem. Suspendisse feugiat, diam vel
  porttitor hendrerit, lectus arcu varius ipsum, non eleifend sem dui
  sit amet ligula. Ut consectetur diam ante, eget varius justo finibus
  vel. Phasellus at porttitor nisl, ac hendrerit dui. Nullam egestas
  malesuada arcu quis efficitur. Nulla consectetur, tellus ut fringilla
  rhoncus, sapien purus accumsan nisl, quis dictum odio nunc sed
  lacus. Integer fringilla maximus consectetur. Praesent eu diam sed
  justo blandit scelerisque. Nam vel mollis sapien, tempus ullamcorper
  odio. Phasellus in lacus at odio euismod convallis. Integer sed
  condimentum eros, eu rhoncus quam. Praesent rutrum sem ante. Curabitur
  a ornare quam.

  Lorem ipsum dolor sit amet, consectetur adipiscing
  elit. Pellentesque bibendum neque ut feugiat vulputate. Donec non mi
  maximus, fermentum mi vitae, egestas orci. Aliquam lobortis, nunc sit
  amet aliquet auctor, eros lorem euismod ante, eu aliquet felis libero
  eget neque. Phasellus blandit, lacus ac rhoncus condimentum, nulla
  risus ultrices ex, et venenatis enim dolor nec velit. Nunc dapibus nec
  sapien eu tristique. Donec vel suscipit purus. Pellentesque mollis
  eros in hendrerit faucibus. Vestibulum ante ipsum primis in faucibus
  orci luctus et ultrices posuere cubilia Curae; Donec a eleifend
  odio. Morbi nec elementum quam. Donec gravida metus vel sem aliquam
  pellentesque. Donec hendrerit nisl in mauris dapibus, nec auctor
  mauris dignissim. Mauris ante mi, pretium nec interdum quis, maximus
  quis tortor. Lorem ipsum dolor sit amet, consectetur adipiscing
  elit. Curabitur non nibh bibendum, viverra enim luctus, commodo dolor.

  Proin aliquet, risus at convallis imperdiet, sem ligula facilisis
  ante, vitae consectetur justo massa quis velit. Fusce eu urna porta,
  pharetra felis in, malesuada augue. Quisque sit amet arcu a turpis
  blandit molestie. In ac lacus eu odio pulvinar faucibus. In vulputate
  at libero a tempor. Etiam eget neque aliquet, accumsan est quis,
  convallis tortor. Cum sociis natoque penatibus et magnis dis
  parturient montes, nascetur ridiculus mus. Nullam hendrerit erat quam,
  sed imperdiet dui efficitur eget. Sed elementum, eros sed dapibus
  lacinia, elit augue iaculis leo, at gravida tortor elit non
  mi. Praesent volutpat lacus molestie risus convallis
  malesuada. Curabitur sed ex eu felis venenatis auctor. Fusce convallis
  turpis ac nibh commodo luctus. Sed commodo dui non arcu mattis
  sagittis. Ut at porta nulla, in sagittis sapien. Duis lectus ligula,
  dignissim id dolor id, porttitor posuere tellus.

  Nulla sit amet venenatis tellus. Donec non enim mi. Vestibulum purus
  leo, dictum sed ligula quis, lacinia vehicula urna. Aenean at libero
  varius, semper urna at, placerat sem. Suspendisse feugiat, diam vel
  porttitor hendrerit, lectus arcu varius ipsum, non eleifend sem dui
  sit amet ligula. Ut consectetur diam ante, eget varius justo finibus
  vel. Phasellus at porttitor nisl, ac hendrerit dui. Nullam egestas
  malesuada arcu quis efficitur. Nulla consectetur, tellus ut fringilla
  rhoncus, sapien purus accumsan nisl, quis dictum odio nunc sed
  lacus. Integer fringilla maximus consectetur. Praesent eu diam sed
  justo blandit scelerisque. Nam vel mollis sapien, tempus ullamcorper
  odio. Phasellus in lacus at odio euismod convallis. Integer sed
  condimentum eros, eu rhoncus quam. Praesent rutrum sem ante. Curabitur
  a ornare quam.
  ")

(defn generate-data
  []
  (.getBytes ^String data "UTF-8"))

;; --- Initialization

(def ^LZ4Factory lz4-factory (LZ4Factory/fastestInstance))
(def ^LZ4Compressor lz4-compressor (.fastCompressor lz4-factory))
(def ^LZ4FastDecompressor lz4-decompressor (.fastDecompressor lz4-factory))
(def zstd-level 1)

;; --- Benchmark Data

(def uncompressed-data (generate-data))
(def original-size (count uncompressed-data))

(def snappy-compressed-data (Snappy/compress uncompressed-data))
(def lz4-compressed-data (.compress lz4-compressor uncompressed-data))
(def zstd-compressed-data (Zstd/compress uncompressed-data ^long zstd-level))

;; --- Benchmark Functions

(defn bench-snappy-compression
  []
  (Snappy/compress ^bytes uncompressed-data))

(defn bench-snappy-uncompression
  []
  (Snappy/uncompress ^bytes snappy-compressed-data))

(defn bench-lz4-compression
  []
  (.compress lz4-compressor ^bytes uncompressed-data))

(defn bench-lz4-uncompression
  []
  (.decompress lz4-decompressor
               ^bytes lz4-compressed-data
               ^int original-size))

(defn bench-zstd-compression
  []
  (Zstd/compress ^bytes uncompressed-data ^long zstd-level))

(defn bench-zstd-uncompression
  []
  (Zstd/decompress ^bytes zstd-compressed-data ^int original-size))

(defn -main
  [& args]
  (println "\nSize comparison:")
  (println "Original: " (count uncompressed-data) "bytes")
  (println "Snappy:   " (count snappy-compressed-data) "bytes")
  (println "LZ4:      " (count lz4-compressed-data) "bytes")
  (println "Zstd:     " (count zstd-compressed-data) "bytes")

  (do
    (println "\nsnappy:compression")
    (quick-bench (bench-snappy-compression) :verbose)

    (println "\nsnappy:uncompression")
    (quick-bench (bench-snappy-uncompression) :verbose)

    (println "\nlz4:compression")
    (quick-bench (bench-lz4-compression) :verbose)

    (println "\nlz4:uncompression")
    (quick-bench (bench-lz4-uncompression) :verbose)

    (println "\nzstd:compression")
    (quick-bench (bench-zstd-compression) :verbose)

    (println "\nzstd:uncompression")
    (quick-bench (bench-zstd-uncompression) :verbose)
    )
  )

result.txt

Size comparison:
Original:  5500 bytes
Snappy:    1888 bytes
LZ4:       1871 bytes
Zstd:      1239 bytes

snappy:compression
amd64 Linux 4.4.22-1-lts 4 cpu(s)
OpenJDK 64-Bit Server VM 25.102-b14
Runtime arguments: -Xms1g -Xmx1g -XX:+UseG1GC -XX:+AggressiveOpts -Dclojure.compiler.direct-linking=true -Dboot.app.path=/home/niwi/bin/boot
Evaluation count : 38148 in 6 samples of 6358 calls.
      Execution time sample mean : 15.939866 µs
             Execution time mean : 15.940166 µs
Execution time sample std-deviation : 169.443316 ns
    Execution time std-deviation : 177.073406 ns
   Execution time lower quantile : 15.733521 µs ( 2.5%)
   Execution time upper quantile : 16.134124 µs (97.5%)
                   Overhead used : 8.128448 ns

snappy:uncompression
amd64 Linux 4.4.22-1-lts 4 cpu(s)
OpenJDK 64-Bit Server VM 25.102-b14
Runtime arguments: -Xms1g -Xmx1g -XX:+UseG1GC -XX:+AggressiveOpts -Dclojure.compiler.direct-linking=true -Dboot.app.path=/home/niwi/bin/boot
Evaluation count : 215526 in 6 samples of 35921 calls.
      Execution time sample mean : 2.789171 µs
             Execution time mean : 2.789251 µs
Execution time sample std-deviation : 43.668779 ns
    Execution time std-deviation : 47.591622 ns
   Execution time lower quantile : 2.741830 µs ( 2.5%)
   Execution time upper quantile : 2.842026 µs (97.5%)
                   Overhead used : 8.128448 ns

lz4:compression
amd64 Linux 4.4.22-1-lts 4 cpu(s)
OpenJDK 64-Bit Server VM 25.102-b14
Runtime arguments: -Xms1g -Xmx1g -XX:+UseG1GC -XX:+AggressiveOpts -Dclojure.compiler.direct-linking=true -Dboot.app.path=/home/niwi/bin/boot
Evaluation count : 64470 in 6 samples of 10745 calls.
      Execution time sample mean : 9.387533 µs
             Execution time mean : 9.387533 µs
Execution time sample std-deviation : 120.745726 ns
    Execution time std-deviation : 121.575286 ns
   Execution time lower quantile : 9.297546 µs ( 2.5%)
   Execution time upper quantile : 9.581689 µs (97.5%)
                   Overhead used : 8.128448 ns

Found 1 outliers in 6 samples (16.6667 %)
        low-severe       1 (16.6667 %)
 Variance from outliers : 13.8889 % Variance is moderately inflated by outliers

lz4:uncompression
amd64 Linux 4.4.22-1-lts 4 cpu(s)
OpenJDK 64-Bit Server VM 25.102-b14
Runtime arguments: -Xms1g -Xmx1g -XX:+UseG1GC -XX:+AggressiveOpts -Dclojure.compiler.direct-linking=true -Dboot.app.path=/home/niwi/bin/boot
Evaluation count : 131106 in 6 samples of 21851 calls.
      Execution time sample mean : 4.622484 µs
             Execution time mean : 4.622318 µs
Execution time sample std-deviation : 62.585152 ns
    Execution time std-deviation : 67.411418 ns
   Execution time lower quantile : 4.543771 µs ( 2.5%)
   Execution time upper quantile : 4.713306 µs (97.5%)
                   Overhead used : 8.128448 ns

zstd:compression
amd64 Linux 4.4.22-1-lts 4 cpu(s)
OpenJDK 64-Bit Server VM 25.102-b14
Runtime arguments: -Xms1g -Xmx1g -XX:+UseG1GC -XX:+AggressiveOpts -Dclojure.compiler.direct-linking=true -Dboot.app.path=/home/niwi/bin/boot
Evaluation count : 33018 in 6 samples of 5503 calls.
      Execution time sample mean : 18.310950 µs
             Execution time mean : 18.310950 µs
Execution time sample std-deviation : 203.716216 ns
    Execution time std-deviation : 215.136814 ns
   Execution time lower quantile : 18.020925 µs ( 2.5%)
   Execution time upper quantile : 18.562508 µs (97.5%)
                   Overhead used : 8.128448 ns

zstd:uncompression
amd64 Linux 4.4.22-1-lts 4 cpu(s)
OpenJDK 64-Bit Server VM 25.102-b14
Runtime arguments: -Xms1g -Xmx1g -XX:+UseG1GC -XX:+AggressiveOpts -Dclojure.compiler.direct-linking=true -Dboot.app.path=/home/niwi/bin/boot
Evaluation count : 73530 in 6 samples of 12255 calls.
      Execution time sample mean : 8.244839 µs
             Execution time mean : 8.244839 µs
Execution time sample std-deviation : 129.557439 ns
    Execution time std-deviation : 131.562930 ns
   Execution time lower quantile : 8.134237 µs ( 2.5%)
   Execution time upper quantile : 8.444877 µs (97.5%)
                   Overhead used : 8.128448 ns