Cuckoo filter

algebird-benchmark/src/main/scala/com/twitter/algebird/benchmark/CuckooFilterBenchmarkQuery.scala

@@ -0,0 +1,38 @@
+package com.twitter.algebird
+package benchmark
+
+import com.twitter.algebird.benchmark.CuckooFilterBenchmarkQuery.CuckooFilterState
+import org.openjdk.jmh.annotations._
+
+object CuckooFilterBenchmarkQuery {
+
+  @State(Scope.Benchmark)
+  class CuckooFilterState {
+
+    @Param(Array("100", "1000", "10000"))
+    var nbrOfElements: Int = 0
+
+    @Param(Array("100", "300"))
+    var bucketNumber: Int = 0
+
+    @Param(Array("10", "50"))
+    var fingerprintBucket: Int = 0
+
+    var cf: CF[String] = _
+
+    @Setup(Level.Trial)
+    def setup(): Unit = {
+      val randomStrings =
+        CuckooFilterCreateBenchmark.createRandomString(nbrOfElements, 10)
+      cf = CuckooFilter[String](fingerprintBucket, bucketNumber)
+        .create(randomStrings: _*)
+    }
+  }
+}
+
+class CuckooFilterBenchmarkQuery {
+
+  @Benchmark
+  def queryCuckooFilter(cuckooFilterState: CuckooFilterState): Boolean =
+    cuckooFilterState.cf.lookup("1")
+}

algebird-benchmark/src/main/scala/com/twitter/algebird/benchmark/CuckooFilterCreateBenchmark.scala

@@ -0,0 +1,41 @@
+package com.twitter.algebird
+package benchmark
+
+import com.twitter.algebird.benchmark.CuckooFilterCreateBenchmark.CuckooFilterState
+import org.openjdk.jmh.annotations._
+
+import scala.util.Random
+
+object CuckooFilterCreateBenchmark {
+  def createRandomString(nbString: Int, sizeStr: Int): Seq[String] =
+    Seq.fill(nbString)(Random.nextString(sizeStr))
+
+  @State(Scope.Benchmark)
+  class CuckooFilterState() {
+
+    @Param(Array("100", "1000", "10000"))
+    var nbrOfElements: Int = 0
+
+    @Param(Array("100", "300"))
+    var bucketNumber: Int = 0
+
+    @Param(Array("10", "50"))
+    var fingerprintBucket: Int = 0
+
+    var randomStrings: Seq[String] = _
+
+    @Setup(Level.Trial)
+    def setup(): Unit =
+      randomStrings = createRandomString(nbrOfElements, 10)
+  }
+
+}
+
+class CuckooFilterCreateBenchmark {
+  @Benchmark
+  def createCuckooFilter(cuckooFilterState: CuckooFilterState): CF[String] = {
+    val cfMonoid = CuckooFilter[String](cuckooFilterState.fingerprintBucket, cuckooFilterState.bucketNumber)
+    val cf = cfMonoid.create(cuckooFilterState.randomStrings: _*)
+    cf
+  }
+}