Struct tiny_bench::BenchmarkConfig

pub struct BenchmarkConfig {
    pub measurement_time: Duration,
    pub num_resamples: usize,
    pub num_samples: usize,
    pub warm_up_time: Duration,
    pub dump_results_to_disk: bool,
    pub max_iterations: Option<u64>,
}

Struct containing all of the configuration options for a benchmark.

Fields

measurement_time: Duration

How long the bench ‘should’ run, num_samples is prioritized so benching will take longer to be able to collect num_samples if the code to be benched is slower than this time limit allowed.

num_resamples: usize

How many resamples should be done

num_samples: usize

Recommended at least 50, above 100 https://en.wikipedia.org/wiki/Bootstrapping_(statistics)#Recommendations doesn’t seem to yield a significantly different result

warm_up_time: Duration

How long the bench should warm up

dump_results_to_disk: bool

Puts results in target/tiny-bench/label/.. if target can be found. used for comparing previous runs

max_iterations: Option<u64>

Sets a hard ceiling on max iterations, overriding the heuristic calculations for iteration count. A rule of thumb; if this is used, the results are unlikely to be statistically significant.

Trait Implementations

impl Default for BenchmarkConfig

fn default() -> Self

Returns the “default value” for a type.