Crate mobench_runner 

mobench-runner

A lightweight benchmarking harness designed for mobile platforms.

This crate provides the core timing infrastructure for the mobench ecosystem. It’s designed to be minimal and portable, with no platform-specific dependencies, making it suitable for compilation to Android and iOS targets.

Overview

The runner executes benchmark functions with:

• Configurable warmup iterations
• Precise nanosecond-resolution timing
• Simple, serializable results

Usage

Most users should use this crate via mobench-sdk. Direct usage is typically only needed for custom integrations:

use mobench_runner::{BenchSpec, run_closure, BenchError};

// Define a benchmark specification
let spec = BenchSpec::new("my_benchmark", 100, 10)?;

// Run the benchmark
let report = run_closure(spec, || {
    // Your benchmark code
    let sum: u64 = (0..1000).sum();
    std::hint::black_box(sum);
    Ok(())
})?;

// Analyze results
let mean_ns = report.samples.iter()
    .map(|s| s.duration_ns)
    .sum::<u64>() / report.samples.len() as u64;

println!("Mean: {} ns", mean_ns);

Types

Type	Description
BenchSpec	Benchmark configuration (name, iterations, warmup)
BenchSample	Single timing measurement in nanoseconds
BenchReport	Complete results with all samples
BenchError	Error conditions during benchmarking

Crate Ecosystem

This crate is part of the mobench ecosystem:

• mobench-sdk - Core SDK with build automation
• mobench - CLI tool
• mobench-macros - #[benchmark] proc macro
• mobench-runner (this crate) - Timing harness

Structs

BenchReport

Complete benchmark report with all timing samples.

BenchSample

A single timing sample from a benchmark iteration.

BenchSpec

Benchmark specification defining what and how to benchmark.

Enums

BenchError

Errors that can occur during benchmark execution.

Functions

run_closure

Runs a benchmark by executing a closure repeatedly.