Using black holes

You can use an instance of Blackhole explicitly, but it's really useful if you're going to consume several values with the black hole. In another case, this approach just affects readability and it would be better to just return the value. The next example demonstrates correct and incorrect cases:

@BenchmarkMode(Mode.AverageTime)
@OutputTimeUnit(TimeUnit.NANOSECONDS)
@State(Scope.Thread)
public class MyBenchmark {

    double x1 = Math.PI;
    double x2 = Math.PI * 2;

    @Benchmark
    public double baseline() {
        return Math.log(x1);
    }

    @Benchmark
    public double measureWrong() {
        Math.log(x1);
        return Math.log(x2);
    }

    @Benchmark
    public double measureRight_1() {
        return Math.log(x1) + Math.log(x2);
    }

    @Benchmark
    public void measureRight_2(Blackhole bh) {
        bh.consume(Math.log(x1));
        bh.consume(Math.log(x2));
    }

}

The output shows how the JVM eliminates the first line of the measureWrong() method:

 Benchmark                         Mode  Cnt Score Error Units
 MyBenchmark.baseline              avgt  5   24.385 ± 1.559 ns/op
 MyBenchmark.measureRight_1        avgt  5   43.861 ± 4.813 ns/op
 MyBenchmark.measureRight_2        avgt  5   47.041 ± 4.800 ns/op
 MyBenchmark.measureWrong          avgt  5   24.447 ± 2.333 ns/op