Type Members

1.  trait AbsoluteNoise extends Noise

   A mixin measurer which adds an absolute amount of Gaussian noise to the measurement.

   A mixin measurer which adds an absolute amount of Gaussian noise to the measurement.

   A random value is sampled from a Gaussian distribution for each measurement x. This value is then multiplied with Key.noiseMagnitude and added to the measurement. The default value for the noise magnitude is 0.0 - it has to be set manually for tests requiring artificial noise. The resulting value is clamped into the range x - magnitude, x + magnitude.

   $noise

2.  abstract class BaseMemoryFootprint extends Measurer[Double]
3.  case class BoxingCount(primitives: Class[_ >: Primitive]*) extends InvocationCount with Product with Serializable

   Counts autoboxed by a Scala compiler values.

   Counts autoboxed by a Scala compiler values.

   primitives

   primitive types whose autoboxing will be counted.

4.  class Default extends Timer with IterationBasedValue

   A default measurer executes the test as many times as specified and returns the sequence of measured times.

5.  class GarbageCollectionCycles extends Measurer[Int]
6.  class IgnoringGC extends Timer with IterationBasedValue

   A measurer that discards measurements during which it detects GC cycles.

   A measurer that discards measurements during which it detects GC cycles.

   Assume that M measurements are requested. To prevent looping forever, after the number of measurement failed due to GC exceeds the number of successful measurements by more than M, the subsequent measurements are accepted regardless of whether GC cycles occur.

7.  trait IterationBasedValue extends AnyRef

   Mixin for measurers whose benchmarked value is based on the current iteration.

8.  class MemoryFootprint extends BaseMemoryFootprint with OutlierElimination[Double]

   Measures the total memory footprint of an object created by the benchmarking snippet.

   Measures the total memory footprint of an object created by the benchmarking snippet.

   Eliminates outliers.

9.  case class MethodInvocationCount(matcher: InvocationCountMatcher) extends InvocationCount with Product with Serializable

   Counts invocations of arbitrary method(s) specified by org.scalameter.execution.invocation.InvocationCountMatcher.

10.  trait Noise extends Measurer[Double]

    A measurer which adds noise to the measurement.

11.  trait OutlierElimination[V] extends Measurer[V]

    A mixin measurer which detects outliers (due to an undetected GC or JIT) and requests additional measurements to replace them.

    A mixin measurer which detects outliers (due to an undetected GC or JIT) and requests additional measurements to replace them. Outlier elimination can also eliminate some pretty bad allocation patterns in some cases. Only outliers from above are considered.

    When detecting an outlier, up to Key.exec.outliers.suspectPercent% (with a minimum of 1) of worst times will be considered. For example, given Key.exec.outliers.suspectPercent = 25 the times:

    10, 11, 10, 12, 11, 11, 10, 11, 44

    times 12 and 44 are considered for outlier elimination.

    Given the times:

    10, 12, 14, 55

    the time 55 will be considered for outlier elimination.

    A potential outlier (suffix) is removed if removing it increases the coefficient of variance by at least Key.exec.outliers.covMultiplier times.

12.  trait PeriodicReinstantiation[V] extends Measurer[V] with IterationBasedValue

    A mixin measurer which causes the value for the benchmark to be reinstantiated every Key.exec.reinstantiation.frequency measurements.

    A mixin measurer which causes the value for the benchmark to be reinstantiated every Key.exec.reinstantiation.frequency measurements. Before the new value has been instantiated, a full GC cycle is invoked if Key.exec.reinstantiation.fullGC is true.

13.  type Primitive = Boolean with Char with Byte with Short with Int with Long with Float with Double
14.  trait RelativeNoise extends Noise

    A mixin measurer which adds an amount of Gaussian noise to the measurement relative to its mean.

    A mixin measurer which adds an amount of Gaussian noise to the measurement relative to its mean.

    An observations sequence mean m is computed. A random Gaussian value is sampled for each measurement x in the observations sequence. It is multiplied with m / 10.0 times Key.noiseMagnitude (default 0.0). Call this multiplication factor f. The resulting value is clamped into the range x - f, x + f.

    The bottomline is - a 1.0 noise magnitude is a variation of 10% of the mean.

    $noise

15.  trait Timer extends Measurer[Double]