JMH Benchmark weird performance because @Param annotation

Question

I try to use JMH with parameters, but I get weird results when @Param annotation is used. Here is case with hard-coded function argument:

@BenchmarkMode(Mode.Throughput)
@OutputTimeUnit(TimeUnit.SECONDS)
@Warmup(iterations = 3, time = 3, timeUnit = TimeUnit.SECONDS)
@Measurement(iterations = 4, time = 4, timeUnit = TimeUnit.SECONDS)
@Fork(1)
@State(Scope.Benchmark)
public class MyBenchmark {

    NumberToBinaryArray array = new NumberToBinaryArray();

    @Benchmark
    public void testMethod() {
        array.toBinaryArray2(0);
    }
}

This implementation has performance: 2_312_967_772 ops/sec (approx. is 2 billions ops/sec). But if I use arguments for JMH performance slowed down:

public class MyBenchmark {

    NumberToBinaryArray array = new NumberToBinaryArray();

    @Param({"0"})   <= changed here
    private int arg;   

    @Benchmark
    public void testMethod() {
        array.toBinaryArray2(arg);   <= changed here
    }
}

Perfomance is 484_748_276 ops/sec. Why so ?

Answer 1

I found out a reason. This so because of not final field (arg).