Benchmark class¶

Provide micro benchmarking functionality

class benchmark.Benchmark(name: str | None = None, runs: int = 1000)¶

Create and run a microbenchmark.

Use this class as a context manager. Create the Benchmark and set the test subject with set_user_code(). Rayne automatically measures and prints the test subject’s execution time when the context manager exits.

with Benchmark() as benchmark:
    benchmark.set_user_code(fibonacci, n=10)

name¶

Rayne assigns a name to each Benchmark. Rayne uses the name in the output to differentiate multiple benchmarks within a single benchmarking script. The default name is the string representation of the test subject.

>>> with Benchmark() as benchmark:
>>>     benchmark.set_user_code(fibonacci, n=10)
>>> print(benchmark.name)
fibonacci

You can assign a custom name to the Benchmark.

Initialization

>>> with Benchmark(name="Recursive") as benchmark:
>>>     benchmark.set_user_code(fibonacci, n=10)
>>> print(benchmark.name)
Recursive

Post Initialization

>>> with Benchmark() as benchmark:
>>>     benchmark.name = "Recursive"
>>>     benchmark.set_user_code(fibonacci, n=10)
>>> print(benchmark.name)
Recursive

runs¶

Rayne runs the test subject several times to build a statistical model of the execution time. The runs attribute controls how many times Rayne runs the test subject. You can modify this value before the Benchmark context manager exits. For example, to run the test subject 2000 times:

Initialization

with Benchmark(runs=2000) as benchmark:
    benchmark.subject = fibonacci

Post Initialization

with Benchmark() as benchmark:
    benchmark.runs = 2000
    benchmark.subject = fibonacci

property run_time: float¶: The average execution time, in nanoseconds, of the subject function.

set_user_code(function, **kwargs)¶

Set the user code to benchmark.

Parameters:

function (Callable) – The benchmark measures the execution time of this function.
kwargs – The benchmark passes these arguments to function.

class benchmark.BenchmarkResults(name: str, run_times: List[int])¶: Benchmark name and associated run times.