Charset benchmark

gistfile1.java

package charset;

import java.io.UnsupportedEncodingException;
import java.nio.charset.StandardCharsets;
import java.util.Random;

import org.openjdk.jmh.annotations.GenerateMicroBenchmark;
import org.openjdk.jmh.annotations.Scope;
import org.openjdk.jmh.annotations.State;

@State(Scope.Benchmark)
public class CharsetBenchmark {

  String ref = "Hello Charset Benchmark";

  // seeding the same randoms for both benchmarks
  Random r1 = new Random(123456789l);
  Random r2 = new Random(123456789l);

  @GenerateMicroBenchmark
  public byte[] _a_preJava7CharsetLookup() {
    try {
      switch (r1.nextInt(4)) {
        case 0:
          return ref.getBytes("UTF-8");
        case 1:
          return ref.getBytes("ISO-8859-1");
        case 2:
          return ref.getBytes("US-ASCII");
        case 3:
          return ref.getBytes("UTF-16");
      }
    } catch (UnsupportedEncodingException e) {
      // errm this cannot happen. the JLS requires these charsets to exist!
    }
    return null;
  }

  @GenerateMicroBenchmark
  public byte[] _b_postJava7CharsetLookup() {
    switch (r2.nextInt(4)) {
      case 0:
        return ref.getBytes(StandardCharsets.UTF_8);
      case 1:
        return ref.getBytes(StandardCharsets.ISO_8859_1);
      case 2:
        return ref.getBytes(StandardCharsets.US_ASCII);
      case 3:
        return ref.getBytes(StandardCharsets.UTF_16);
    }
    return null;
    // No Exception! YAY!
  }

}

/**
 * <pre>
 * Benchmark                                        Mode   Samples         Mean   Mean error    Units
 * c.CharsetBenchmark._a_preJava7CharsetLookup     thrpt         5     3956.537      144.562   ops/ms
 * c.CharsetBenchmark._b_postJava7CharsetLookup    thrpt         5     7138.064      179.101   ops/ms
 * </pre>
 */

@shipilev correctly points out, that this benchmark does not scale with concurrent threads. Its a toy benchmark, also because in reality you are not randomly choosing charsets. I chose the randomness to cheat: it maximizes the cost of looking up charsets, because the last used charset will be cached in a way it is faster to access.