Add JMH benchmarks for Lucene IntIntHashMap and Eclipse IntShortHashM…

…ap (#599)

Taskfile.yaml

@@ -186,9 +186,9 @@ tasks:
     cmds:
       - sbt -client elastiknn-api4s/test
       - sbt -client elastiknn-client-elastic4s/test
+      - sbt -client elastiknn-jmh-benchmarks/test
       - sbt -client elastiknn-lucene/test
       - sbt -client elastiknn-models/test
-      - sbt -client elastiknn-models-benchmarks/test
       - sbt -client elastiknn-plugin/test
 
   jvmUnitTestQuick:

build.sbt

@@ -38,7 +38,7 @@ lazy val `elastiknn-root` = project
     `elastiknn-client-elastic4s`,
     `elastiknn-lucene`,
     `elastiknn-models`,
-    `elastiknn-models-benchmarks`,
+    `elastiknn-jmh-benchmarks`,
     `elastiknn-plugin`,
     `elastiknn-plugin-integration-tests`
   )
@@ -101,13 +101,17 @@ lazy val `elastiknn-models` = project
     TestSettings
   )
 
-lazy val `elastiknn-models-benchmarks` = project
-  .in(file("elastiknn-models-benchmarks"))
-  .dependsOn(`elastiknn-models`, `elastiknn-api4s`)
+lazy val `elastiknn-jmh-benchmarks` = project
+  .in(file("elastiknn-jmh-benchmarks"))
+  .dependsOn(`elastiknn-models`, `elastiknn-api4s`, `elastiknn-lucene`)
   .enablePlugins(JmhPlugin)
   .settings(
     Jmh / javaOptions ++= Seq("--add-modules", "jdk.incubator.vector"),
-    TpolecatSettings
+    TpolecatSettings,
+    libraryDependencies ++= Seq(
+      "org.eclipse.collections" % "eclipse-collections" % "11.1.0",
+      "org.eclipse.collections" % "eclipse-collections-api" % "11.1.0"
+    )
   )
 
 lazy val `elastiknn-plugin` = project

elastiknn-jmh-benchmarks/src/main/scala/com/klibisz/elastiknn/jmhbenchmarks/HitCounterBenchmarks.scala

@@ -0,0 +1,63 @@
+package com.klibisz.elastiknn.jmhbenchmarks
+
+import org.openjdk.jmh.annotations._
+import org.apache.lucene.util.hppc.IntIntHashMap
+import org.eclipse.collections.impl.map.mutable.primitive.IntShortHashMap
+
+import scala.util.Random
+
+@State(Scope.Benchmark)
+class HitCounterBenchmarksFixtures {
+  val rng = new Random(0)
+  val numDocs = 60000
+  val numHits = 2000
+  val initialMapSize = 1000
+  val docs = (1 to numHits).map(_ => rng.nextInt(numDocs)).toArray
+}
+
+class HitCounterBenchmarks {
+
+  @Benchmark
+  @BenchmarkMode(Array(Mode.Throughput))
+  @Fork(value = 1)
+  @Warmup(time = 5, iterations = 5)
+  @Measurement(time = 5, iterations = 5)
+  def arrayCountBaseline(f: HitCounterBenchmarksFixtures): Unit = {
+    val arr = new Array[Int](f.numDocs)
+    for (d <- f.docs) arr.update(d, arr(d) + 1)
+    ()
+  }
+
+  @Benchmark
+  @BenchmarkMode(Array(Mode.Throughput))
+  @Fork(value = 1)
+  @Warmup(time = 5, iterations = 5)
+  @Measurement(time = 5, iterations = 5)
+  def hashMapGetOrDefault(f: HitCounterBenchmarksFixtures): Unit = {
+    val h = new java.util.HashMap[Int, Int](f.initialMapSize, 0.99f)
+    for (d <- f.docs) h.put(d, h.getOrDefault(d, 0) + 1)
+    ()
+  }
+
+  @Benchmark
+  @BenchmarkMode(Array(Mode.Throughput))
+  @Fork(value = 1)
+  @Warmup(time = 5, iterations = 5)
+  @Measurement(time = 5, iterations = 5)
+  def luceneIntIntHashMap(f: HitCounterBenchmarksFixtures): Unit = {
+    val m = new IntIntHashMap(f.initialMapSize, 0.99d)
+    for (d <- f.docs) m.putOrAdd(d, 1, 1)
+    ()
+  }
+
+  @Benchmark
+  @BenchmarkMode(Array(Mode.Throughput))
+  @Fork(value = 1)
+  @Warmup(time = 5, iterations = 5)
+  @Measurement(time = 5, iterations = 5)
+  def eclipseIntShortHashMapAddToValue(f: HitCounterBenchmarksFixtures): Unit = {
+    val m = new IntShortHashMap(f.initialMapSize)
+    for (d <- f.docs) m.addToValue(d, 1)
+    ()
+  }
+}

elastiknn-models-benchmarks/src/main/scala/com/klibisz/elastiknn/vectors/VectorOpsJmhBenchmark.scala → elastiknn-jmh-benchmarks/src/main/scala/com/klibisz/elastiknn/jmhbenchmarks/VectorOpsJmhBenchmark.scala

@@ -1,6 +1,7 @@
-package com.klibisz.elastiknn.vectors
+package com.klibisz.elastiknn.jmhbenchmarks
 
 import com.klibisz.elastiknn.api.Vec
+import com.klibisz.elastiknn.vectors._
 import org.openjdk.jmh.annotations._
 
 import scala.util.Random