matlab-write-performance-tests

What problem does it solve?

Writing trustworthy MATLAB performance tests is hard because naive timing (or poorly structured tests) produces noisy, non-reproducible results that mask real regressions.

Core Features & Use Cases

• Statistically rigorous benchmarking using matlab.perftest.TestCase with warmup, multiple samples, and outlier handling.
• Proper measurement boundaries via keepMeasuring for fast code, or startMeasuring/stopMeasuring to isolate exactly what you want timed.
• Correct parameterization patterns using MethodSetupParameter for values consumed during setup, plus guidance to avoid common MATLAB perf-test gotchas.
• Best-practice test design including RNG seeding, avoiding setup inside the measured region, and using local variables to reduce measurement overhead.

What problem does it solve?

Aizes performance-test development for MATLAB code by generating clean, repeatable benchmark classes that work well with runperf and matlab.perftest.TestCase.

Quick Start

Use this skill to write a matlab.perftest.TestCase performance test for your function, benchmark it across multiple input sizes, and run it with runperf to get median timing and regression-ready results.