silk.bench.report.

Turns the raw samples of a benchmark run into the report a renderer draws.

When to use#

Call Reports.make on a silk.bench.record.RunResult you read from a result file, to render a saved run again or with another noise threshold. silk.bench.bench calls it for every suite it runs.

Details#

A combination of parameters whose programs print result lines gets one table for each name the lines carry, in the unit of the lines: ns/op when a line has it, or else the first unit written. The medians of the other units are in the row. Where a larger value is better, as for req/s, a ratio above 1 has the verdict faster. When the lines of two implementations carry different checks for one name, the table is marked and no row of it has a ratio. Every other combination gets one table of process times, as follows.

Each implementation gets one row for each benchmark. The row's statistics cover the measured runs that exited with status 0: warm-up runs and failed runs are left out, and failed runs are counted in the row. Every row except the reference's gets its ratio to the reference with a 95% interval and a verdict, from silk.bench.statistics. A suite with more than one benchmark also gets the geometric mean of each implementation's ratios.

Import as Reports with import silk.bench.report { Reports }.

Public declarations: 1.

Reports#

silk
pub struct Reports

The owner type of the report functions. It has no fields and is never constructed.

Implementation Reports#

silk
impl Reports

Associated function compared#

silk
pub effect fn compared(result: &mut RunResult, earlier: RunResult, label: string) -> bool
    ! OutOfMemoryError
    ? &mut Allocator

Adds the implementations of the earlier run earlier to result under the name label, with their samples for the benchmarks both runs have, and makes the earlier reference the reference. An earlier run with one implementation adds it as label; one with several adds each as <label>:<name>. Returns false, and adds nothing, if result already has an implementation named label.

Errors#

OutOfMemoryError if the Allocator cannot supply the names.

Associated function make#

silk
pub effect fn make(result: &RunResult, noise: f64) -> Report
    ! OutOfMemoryError
    ? &mut Allocator

Returns the report of result, with verdicts at the noise threshold noise, a fraction such as 0.01 for 1%.

Errors#

OutOfMemoryError if the Allocator cannot supply the report.

Category: Benchmarks.

Profiles: aarch64-apple-darwin, aarch64-unknown-linux-gnu, aarch64-unknown-linux-gnu-no-libc, x86_64-unknown-linux-gnu, x86_64-unknown-linux-gnu-no-libc, wasm32-unknown-unknown, wasm32-wasip1, wasm32-wasip2, x86_64-pc-windows-msvc, wasm64-unknown-unknown.

View the source of this page