Fix jmh

Author: nbirillo

src/jmh/kotlin/cub/BenchmarkRunner.kt

@@ -9,35 +9,41 @@ import java.util.concurrent.TimeUnit
  * to see the difference between first iterations and the next ones
  * To see the difference, just check Fork: N of 3 sections, e.g.
  *
- * # Fork: 1 of 3
- * Iteration   1: 27041458.000 ns/op
- * Iteration   2: 1750.000 ns/op
- * Iteration   3: 1750.000 ns/op
- * Iteration   4: 625.000 ns/op
- * Iteration   5: 1125.000 ns/op
- * Iteration   6: 791.000 ns/op
- * Iteration   7: 958.000 ns/op
- * Iteration   8: 792.000 ns/op
- * Iteration   9: 750.000 ns/op
- * Iteration  10: 1084.000 ns/op
+    # Run progress: 50.00% complete, ETA 00:00:00
+    # Fork: 1 of 3
+    # Warmup Iteration   1: 27579584.000 ns/op
+    # Warmup Iteration   2: 1333.000 ns/op
+    # Warmup Iteration   3: 667.000 ns/op
+    Iteration   1: 625.000 ns/op
+    Iteration   2: 833.000 ns/op
+    Iteration   3: 917.000 ns/op
+    Iteration   4: 625.000 ns/op
+    Iteration   5: 875.000 ns/op
+    Iteration   6: 667.000 ns/op
+    Iteration   7: 708.000 ns/op
+    Iteration   8: 625.000 ns/op
+    Iteration   9: 1584.000 ns/op
+    Iteration  10: 4292.000 ns/op
  *
- * Here we can see that the first iteration takes more time than all others.
+ * Here we can see that after several Warmup iterations time is table .
  * We see different time between attempts, it is ok since the processor spents time for time measurement operations,
  * it affects to the results, but the time here is in nano secs, the difference is not significant.
 **/
 open class BenchmarkRunner {
     companion object {
         private const val BENCHMARK_ITERATIONS = 10
         private const val BENCHMARK_FORKS = 3
-        private const val BENCHMARK_WARMUPS = 0
+        private const val BENCHMARK_WARMUPS = 3
     }
     @Benchmark
     @BenchmarkMode(Mode.SingleShotTime)
     @OutputTimeUnit(TimeUnit.NANOSECONDS)
     @Measurement(iterations = BENCHMARK_ITERATIONS)
+    @Warmup(iterations = BENCHMARK_WARMUPS)
     @Fork(value = BENCHMARK_FORKS, warmups = BENCHMARK_WARMUPS)
     fun solveSudoku() {
         val solver = BacktrackingSudokuSolver()
         solver.solve()
     }
 }
+