Enum SampleOrder 

pub(crate) enum SampleOrder {
    Blocked,
    RoundRobin,
}

The order the suite takes its samples in.

This changes nothing about what is measured – the same rows run the same number of times either way, each row’s report is the same shape, and a whole suite takes the same 564 seconds under either – only when each sample is taken. docs/VM_ARCHITECTURE.md, “What the measurement itself costs”, is the round that measured which one to prefer, and by how little.

Variants

Blocked

Every sample of one row, then every sample of the next.

The order this harness used until issue #205, and the reason a short row was the least reliable thing in the suite: pure on the VM runs in about 1.4 ms, so fifteen samples of it are twenty milliseconds of measurement taken at one instant of a suite that lasts nine and a half minutes. Whatever the machine was doing in that instant is the whole of that row’s answer, and nothing in the row’s own spread can say so.

Kept because the round that replaced it was run against it, and a result nobody can reproduce is a result nobody can check.

RoundRobin

One sample of every row, then a second of every row, and so on.

The same total work, rearranged so that each row’s series is spread over the whole suite instead of one instant of it. A machine that drifts over the suite then drifts through every row’s series rather than between one row’s series and another’s, so the median of a series is a summary of the session rather than of a moment in it, and the interquartile range beside it starts including the drift instead of being blind to it.

Trait Implementations

impl Clone for SampleOrder

fn clone(&self) -> SampleOrder

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for SampleOrder

impl Debug for SampleOrder

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Eq for SampleOrder

impl PartialEq for SampleOrder

fn eq(&self, other: &SampleOrder) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for SampleOrder