microbenchmark

A minimal Python library for writing and running benchmarks.

microbenchmark gives you simple building blocks — Scenario, ScenarioGroup, and BenchmarkResult — that you can embed directly into your project or call from CI. No separate CLI package to install; .cli() is built in. You write a Python file, call .run() or .cli(), and you are done.

Key features:

A Scenario wraps any callable with an optional argument list and runs it n times, collecting per-run timings.
The arguments() helper captures both positional and keyword arguments for the benchmarked function.
A ScenarioGroup lets you combine scenarios and run them together with a single call.
BenchmarkResult holds every individual duration and gives you mean, median, best, worst, and percentile views.
Results can be serialized to and restored from JSON.
One dependency: printo (from the mutating organization), used for argument and function display in CLI output.

Installation

pip install microbenchmark

Quick start

from microbenchmark import Scenario

def build_list():
    return list(range(1000))

scenario = Scenario(build_list, number=500)  # name auto-derived as 'build_list'
result = scenario.run()

print(len(result.durations))
#> 500
print(result.mean)   # example — actual value depends on your hardware
#> 0.000012
print(result.median)
#> 0.000011
print(result.best)
#> 0.000010
print(result.worst)
#> 0.000018

arguments

The arguments class (lowercase by design) captures positional and keyword arguments for the benchmarked function. Import it directly:

from microbenchmark import arguments

Or use the short alias a — handy when writing compact benchmark scripts:

from microbenchmark import a

Both arguments and a refer to the same class. Create an instance by calling it like a function:

from microbenchmark import arguments

args = arguments(3, 1, 2)
print(args.args)
#> (3, 1, 2)
print(args.kwargs)
#> {}

args_with_kw = arguments(3, 1, 2, key=str)
print(args_with_kw.args)
#> (3, 1, 2)
print(args_with_kw.kwargs)
#> {'key': <class 'str'>}

The a alias is particularly useful when combining many scenarios inline:

from microbenchmark import Scenario, a

scenario = Scenario(sorted, a([3, 1, 2]), name='sort')
result = scenario.run()

arguments has a readable repr:

from microbenchmark import arguments

print(arguments(1, 2, key='value'))
#> arguments(1, 2, key='value')

print(arguments())
#> arguments()

`match(function)`

Checks whether function can be called with the arguments captured in this arguments instance. Returns True if the call is compatible with the function's signature, False otherwise.

from microbenchmark import arguments

args = arguments(1, 2)
print(args.match(lambda a, b: None))
#> True

print(args.match(lambda a: None))
#> False

Works with keyword arguments too:

from microbenchmark import arguments

args = arguments(key='value')
print(args.match(lambda *, key: None))
#> True

print(args.match(lambda: None))
#> False

Note on unintrospectable callables: For some exotic C-extension functions whose signatures Python cannot inspect at all, the check is silently skipped and True is returned. The function will be validated at runtime when the benchmark actually runs. Most common callables — including standard built-in functions like len — have introspectable signatures and are checked normally.

Scenario

A Scenario describes a single benchmark: the function to call, what arguments to pass, and how many times to run it.

Constructor

Scenario(
    function,
    arguments=None,
    *,
    name=None,
    doc='',
    number=1000,
    timer=time.perf_counter,
)

function — the callable to benchmark.
arguments — an arguments instance that holds the positional and keyword arguments passed to function on every call. None (the default) means the function is called with no arguments. Supports both positional and keyword arguments.
name — a short label for this scenario. If omitted, the name is derived automatically from function.__name__. For lambdas, the derived name will be '<lambda>'.
doc — an optional longer description.
number — how many times to call function per run. Must be at least 1; passing 0 or a negative value raises ValueError.
timer — a zero-argument callable that returns the current time as a float. Defaults to time.perf_counter. Supply a custom clock to get deterministic measurements in tests:

Signature validation: The constructor automatically checks that function can be called with the provided arguments. If the signatures are incompatible, a sigmatch.SignatureMismatchError is raised immediately — before the benchmark runs:

from microbenchmark import Scenario, arguments
from sigmatch import SignatureMismatchError

try:
    Scenario(lambda a: None, arguments(1, 2))
except SignatureMismatchError as e:
    print('caught:', e)
#> caught: Scenario arguments arguments(1, 2) are incompatible with the signature of <lambda>

For the rare callables whose signatures Python cannot introspect at all, the validation is silently skipped. See arguments.match() for details.

from microbenchmark import Scenario

tick = [0.0]
def fake_timer():
    tick[0] += 0.001
    return tick[0]

scenario = Scenario(lambda: None, name='noop', number=5, timer=fake_timer)
result = scenario.run()
print(result.mean)
#> 0.001

from microbenchmark import Scenario, arguments

scenario = Scenario(
    sorted,
    arguments([3, 1, 2]),
    name='sort_three_items',
    doc='Sort a list of three integers.',
    number=10000,
)
print(scenario.name)
#> sort_three_items
print(scenario.doc)
#> Sort a list of three integers.
print(scenario.number)
#> 10000

When name is omitted, it is derived from the function:

from microbenchmark import Scenario

def my_function():
    return list(range(100))

scenario = Scenario(my_function)
print(scenario.name)
#> my_function

For keyword arguments, pass them through arguments:

from microbenchmark import Scenario, arguments

scenario = Scenario(
    sorted,
    arguments([3, 1, 2], key=lambda x: -x),
    name='sort_descending',
)
result = scenario.run()

For functions that take multiple positional arguments:

from microbenchmark import Scenario, arguments

scenario = Scenario(pow, arguments(2, 10), name='power')
result = scenario.run()
print(result.mean)
#> 0.000001  # example — very fast operation

`run(warmup=0)`

Runs the benchmark and returns a BenchmarkResult.

The optional warmup argument specifies how many calls to make before timing begins. Warm-up calls execute the function but are not timed and their results are discarded. Warmup is useful when your function has one-time initialization costs — cache warming, lazy imports, JIT compilation — that you do not want to measure. Without warmup, the first few runs may be outliers that skew the mean.

from microbenchmark import Scenario

scenario = Scenario(lambda: list(range(100)), name='build', number=1000)
result = scenario.run(warmup=100)
print(len(result.durations))
#> 1000

`cli(argv=None)`

Turns the scenario into a small command-line program. Call scenario.cli() as the entry point of a script and it will parse sys.argv, run the benchmark, and print the result to stdout.

Pass argv as a list of strings to override sys.argv — useful when calling .cli() programmatically or from the microbenchmark command-line tool.

Supported arguments:

--number N — override the scenario's number for this run.
--max-mean THRESHOLD — exit with code 1 if the mean time (in seconds) exceeds THRESHOLD. Useful in CI.
--histogram — append an ASCII histogram of per-call timings inside the border. The histogram is 8 rows tall and fills the available inner width. The x-axis is clipped at the p99 value to prevent extreme outliers from compressing the bulk of the distribution.
--help — print usage information and exit.

Output format (each scenario is wrapped in a Unicode border):

╭──────────────────────────────────╮
│ benchmark: <name>                │
│ call:      <function>(<args>)    │
│ doc:       <doc>                 │
│ runs:      <number>              │
│ mean:      <mean>s               │
│ median:    <median>s             │
│ best:      <best>s               │
│ worst:     <worst>s              │
│ p95 mean:  <p95.mean>s           │
│ p99 mean:  <p99.mean>s           │
│ total:     <total_duration>s     │
│ fn total:  <functions_duration>s │
╰──────────────────────────────────╯

The doc: line is omitted when doc is empty. The call: line shows the function name and its arguments. Times are in seconds. Labels are padded to the same width for alignment. total: is the wall-clock time of the whole benchmark loop; fn total: is the sum of per-call timings (math.fsum(durations)). The border width adapts to the terminal width (minimum 20 columns).

If --max-mean is supplied and the actual mean exceeds the threshold, the output is printed in full and then a failure line is added before the process exits with code 1:

FAIL: mean <actual>s exceeds --max-mean <threshold>s

# benchmark.py
from microbenchmark import Scenario

def build_list():
    return list(range(1000))

scenario = Scenario(build_list, doc='Build a list of 1000 integers.', number=500)

if __name__ == '__main__':
    scenario.cli()

$ python benchmark.py
╭─────────────────────────────────────────────────╮
│ benchmark: build_list                           │
│ call:      build_list()                         │
│ doc:       Build a list of 1000 integers.       │
│ runs:      500                                  │
│ mean:      0.000012s                            │
│ median:    0.000011s                            │
│ best:      0.000010s                            │
│ worst:     0.000018s                            │
│ p95 mean:  0.000011s                            │
│ p99 mean:  0.000012s                            │
│ total:     0.006100s                            │
│ fn total:  0.006000s                            │
╰─────────────────────────────────────────────────╯

Use --histogram to append an ASCII distribution chart below the metrics:

$ python benchmark.py --histogram
╭─────────────────────────────────────────────────╮
│ benchmark: build_list                           │
│ call:      build_list()                         │
│ doc:       Build a list of 1000 integers.       │
│ runs:      500                                  │
│ mean:      0.000012s                            │
│ median:    0.000011s                            │
│ best:      0.000010s                            │
│ worst:     0.000018s                            │
│ p95 mean:  0.000011s                            │
│ p99 mean:  0.000012s                            │
│ total:     0.006100s                            │
│ fn total:  0.006000s                            │
│                                                 │
│ █████████████████████████████████████████████   │
│ █████████████████████████████████████████████   │
│ ████████████████████████████████████████████    │
│ ████████████████████████████   ████████████     │
│ ████████████████████ ██████    █████████        │
│ ██████████████ ████   ████      ██████          │
│ ████████ ████   ██     ██        ████           │
│ ████      ██     █               ██             │
╰─────────────────────────────────────────────────╯

Use --number to override the run count for this invocation. Use --max-mean to set a CI threshold:

$ python benchmark.py --max-mean 0.000001
╭─────────────────────────────────────────────────╮
│ benchmark: build_list                           │
│ call:      build_list()                         │
│ doc:       Build a list of 1000 integers.       │
│ runs:      500                                  │
│ mean:      0.000012s                            │
│ median:    0.000011s                            │
│ best:      0.000010s                            │
│ worst:     0.000018s                            │
│ p95 mean:  0.000011s                            │
│ p99 mean:  0.000012s                            │
│ total:     0.006100s                            │
│ fn total:  0.006000s                            │
╰─────────────────────────────────────────────────╯
FAIL: mean 0.000012s exceeds --max-mean 0.000001s
$ echo $?
1

ScenarioGroup

A ScenarioGroup holds a flat collection of scenarios and lets you run them together.

Creating a group

There are four ways to create a group.

Direct construction — pass any number of scenarios to the constructor. Passing no scenarios creates an empty group:

from microbenchmark import Scenario, ScenarioGroup

s1 = Scenario(lambda: None, name='s1')
s2 = Scenario(lambda: None, name='s2')

group = ScenarioGroup(s1, s2)
empty = ScenarioGroup()
print(len(empty.run()))
#> 0

The + operator between two scenarios produces a ScenarioGroup:

from microbenchmark import Scenario

s1 = Scenario(lambda: None, name='s1')
s2 = Scenario(lambda: None, name='s2')
group = s1 + s2
print(type(group).__name__)
#> ScenarioGroup

Adding a scenario to an existing group, or vice versa — the result is always a new flat group with no nesting:

from microbenchmark import Scenario, ScenarioGroup

s1 = Scenario(lambda: None, name='s1')
s2 = Scenario(lambda: None, name='s2')
s3 = Scenario(lambda: None, name='s3')
group = ScenarioGroup(s1, s2)
extended = group + s3     # ScenarioGroup + Scenario
also_ok  = s3 + group     # Scenario + ScenarioGroup
print(len(extended.run()))
#> 3

Adding two groups together produces a single flat group:

from microbenchmark import Scenario, ScenarioGroup

s1 = Scenario(lambda: None, name='s1')
s2 = Scenario(lambda: None, name='s2')
s3 = Scenario(lambda: None, name='s3')
g1 = ScenarioGroup(s1)
g2 = ScenarioGroup(s2, s3)
combined = g1 + g2
print(len(combined.run()))
#> 3

`run(warmup=0)`

Runs every scenario in order and returns a list of BenchmarkResult objects. The order of results matches the order the scenarios were added. The warmup argument is forwarded to each scenario individually.

from microbenchmark import Scenario, ScenarioGroup

s1 = Scenario(lambda: None, name='s1')
s2 = Scenario(lambda: None, name='s2')
group = ScenarioGroup(s1, s2)
results = group.run(warmup=50)
for result in results:
    print(result.scenario.name)
#> s1
#> s2

`cli(argv=None)`

Runs all scenarios and prints their results to stdout. Each scenario block is displayed in a nested Unicode border. All inner blocks are wrapped together in a single outer border.

Supported arguments:

--number N — passed to every scenario.
--max-mean THRESHOLD — exits with code 1 if any scenario's mean exceeds the threshold.
--histogram — append an ASCII histogram of per-call timings inside each inner border. The histogram is 8 rows tall and fills the available inner width. The x-axis is clipped at the p99 value to prevent extreme outliers from compressing the bulk of the distribution.
--help — print usage information and exit.

# benchmarks.py
from microbenchmark import Scenario, ScenarioGroup

s1 = Scenario(lambda: list(range(100)), name='range_100')
s2 = Scenario(lambda: list(range(1000)), name='range_1000')

group = s1 + s2

if __name__ == '__main__':
    group.cli()

$ python benchmarks.py
╭────────────────────────────────────────────────────╮
│ ╭──────────────────────────────────────────────╮  │
│ │ benchmark: range_100                         │  │
│ │ call:      range_100()                       │  │
│ │ runs:      1000                              │  │
│ │ mean:      0.000003s                         │  │
│ │ median:    0.000003s                         │  │
│ │ best:      0.000002s                         │  │
│ │ worst:     0.000005s                         │  │
│ │ p95 mean:  0.000003s                         │  │
│ │ p99 mean:  0.000003s                         │  │
│ │ total:     0.003200s                         │  │
│ │ fn total:  0.003000s                         │  │
│ ╰──────────────────────────────────────────────╯  │
│ ╭──────────────────────────────────────────────╮  │
│ │ benchmark: range_1000                        │  │
│ │ call:      range_1000()                      │  │
│ │ runs:      1000                              │  │
│ │ mean:      0.000012s                         │  │
│ │ median:    0.000011s                         │  │
│ │ best:      0.000010s                         │  │
│ │ worst:     0.000018s                         │  │
│ │ p95 mean:  0.000011s                         │  │
│ │ p99 mean:  0.000012s                         │  │
│ │ total:     0.012500s                         │  │
│ │ fn total:  0.012000s                         │  │
│ ╰──────────────────────────────────────────────╯  │
╰────────────────────────────────────────────────────╯

BenchmarkResult

BenchmarkResult is a dataclass that holds the outcome of a single benchmark run.

Fields

scenario: Scenario | None — the Scenario that produced this result, or None if the result was restored from JSON.
durations: tuple[float, ...] — per-call timings in seconds, one entry per call, in the order they were measured.
total_duration: float — wall-clock time of the entire benchmark loop in seconds, measured from just before the first call to just after the last call. Warmup time is not included. Must be provided at construction time.
mean: float — arithmetic mean of durations, computed with math.fsum to minimize floating-point error. Computed automatically from durations.
functions_duration: float — sum of all per-call timings (math.fsum(durations)). Computed automatically from durations.
median: float — median of durations. Computed lazily on first access and cached for the lifetime of the result object.
best: float — the shortest individual timing. Computed automatically.
worst: float — the longest individual timing. Computed automatically.
is_primary: bool — True for results returned directly by run(), False for results derived via percentile(). Preserved during JSON round-trips.

The mean, best, worst, and functions_duration fields are read-only computed values (not accepted as constructor arguments). The total_duration field is an input: pass it to the constructor. The median, p95, and p99 properties are cached lazily.

from microbenchmark import Scenario

result = Scenario(lambda: None, name='noop', number=100).run()
print(len(result.durations))
#> 100
print(result.is_primary)
#> True
print(isinstance(result.median, float))
#> True
print(isinstance(result.total_duration, float))
#> True
print(isinstance(result.functions_duration, float))
#> True

`percentile(p)`

Returns a new BenchmarkResult containing only the ceil(len(durations) * p / 100) fastest timings, sorted by duration ascending. The returned result has is_primary=False. p must be in the range (0, 100]; passing 0 or a value above 100 raises ValueError.

Percentiles help you focus on the typical case by trimming outliers. If your benchmark includes occasional GC pauses or scheduling jitter, the p95 or p99 view shows what most calls actually experience. is_primary=False marks results that are derived from raw data rather than measured directly; this distinction is preserved during JSON round-trips.

from microbenchmark import Scenario

result = Scenario(lambda: None, name='noop', number=100).run()
trimmed = result.percentile(95)
print(trimmed.is_primary)
#> False
print(len(trimmed.durations))
#> 95

You can call percentile() on a derived result too:

from microbenchmark import Scenario

result = Scenario(lambda: None, name='noop', number=100).run()
print(len(result.percentile(90).percentile(50).durations))
#> 45

`p95` and `p99`

Convenient cached properties that return percentile(95) and percentile(99) respectively. The value is computed once and cached for the lifetime of the result object.

from microbenchmark import Scenario

result = Scenario(lambda: None, name='noop', number=100).run()
print(len(result.p95.durations))
#> 95
print(result.p95.is_primary)
#> False
print(result.p95 is result.p95)   # cached — same object returned each time
#> True

`to_json()` and `from_json()`

to_json() serializes the result to a JSON string. It stores durations, is_primary, total_duration, and the scenario's name, doc, and number. The functions_duration field is derived and not stored.

from_json() is a class method that restores a BenchmarkResult from a JSON string produced by to_json(). Because the original callable cannot be serialized, the restored result has scenario=None. The mean, best, worst, median, and functions_duration fields are recomputed from durations on restoration.

Backward compatibility: JSON produced by older versions of microbenchmark that do not include total_duration can still be loaded. When total_duration is absent, from_json() falls back to math.fsum(durations) (equivalent to functions_duration), meaning total_duration will equal functions_duration (overhead is treated as zero).

from microbenchmark import Scenario, BenchmarkResult

result = Scenario(lambda: None, name='noop', number=100).run()

json_str = result.to_json()
restored = BenchmarkResult.from_json(json_str)

print(restored.scenario)
#> None
print(restored.mean == result.mean)
#> True
print(restored.durations == result.durations)
#> True
print(restored.is_primary == result.is_primary)
#> True
print(restored.median == result.median)
#> True

CLI

The microbenchmark package installs a command-line tool of the same name. It lets you run any Scenario or ScenarioGroup object directly from the terminal without writing a wrapper script.

Usage

microbenchmark TARGET [OPTIONS]

TARGET is the fully-qualified import path to a Scenario or ScenarioGroup object in your module, in the form module.path:attribute:

microbenchmark my_pkg.bench:suite
microbenchmark my_pkg.bench:single_scenario --number 500
microbenchmark my_pkg.bench:suite --max-mean 0.001

The module must be importable — either installed in the current environment or located in the current working directory. If the module is a local file (e.g. bench.py), run the command from the directory that contains it:

microbenchmark bench:suite

Options

All options accepted by Scenario.cli() and ScenarioGroup.cli() are forwarded automatically:

--number N — override the iteration count for this run.
--max-mean THRESHOLD — exit with code 1 if any scenario's mean time (seconds) exceeds THRESHOLD.
--histogram — append an ASCII histogram of per-call timings inside the output border. The x-axis is clipped at the p99 value to prevent extreme outliers from compressing the bulk of the distribution.
--help — print usage and exit.

Exit codes

Code	Meaning
0	All benchmarks passed.
1	A benchmark exceeded `--max-mean`.
3	Invalid target specification or import error.

Example

Given a file bench.py in the current directory:

from microbenchmark import Scenario, arguments

scenario = Scenario(sorted, arguments([3, 1, 2]), name='sort', number=1000)

Run it directly:

$ microbenchmark bench:scenario --number 500

Comparison with alternatives

Feature	`microbenchmark`	`timeit` (stdlib)	`pytest-benchmark`
Per-call timings	yes	via `repeat(number=1)`	yes
Percentile views	yes	no	yes
Median	yes	no	yes
JSON serialization	yes	no	yes
Inject custom timer	yes	yes	no
Warmup support	yes	no	yes (calibration)
CI integration (`--max-mean`)	yes	no	via configuration
Keyword arguments	yes	yes	yes
`+` operator for grouping	yes	no	no
External dependencies	one (`printo`)	none	several
Embeddable in your own code	yes	yes	pytest plugin required

timeit from the standard library is great for interactive exploration, but it gives only a single aggregate number per call — you can get a list by using repeat(number=1), though the interface is not designed around it. pytest-benchmark is powerful and well-integrated into the pytest ecosystem, but it is tightly coupled to the test runner and brings its own dependencies. microbenchmark sits between the two: richer than timeit, lighter and more portable than pytest-benchmark, and not tied to any test framework.

Name		Name	Last commit message	Last commit date
Latest commit History 56 Commits
.github		.github
microbenchmark		microbenchmark
tests		tests
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
pyproject.toml		pyproject.toml
requirements_dev.txt		requirements_dev.txt
spec.md		spec.md

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

microbenchmark

Table of contents

Installation

Quick start

arguments

`match(function)`

Scenario

Constructor

`run(warmup=0)`

`cli(argv=None)`

ScenarioGroup

Creating a group

`run(warmup=0)`

`cli(argv=None)`

BenchmarkResult

Fields

`percentile(p)`

`p95` and `p99`

`to_json()` and `from_json()`

CLI

Usage

Options

Exit codes

Example

Comparison with alternatives

About

Uh oh!

Releases 3

Packages

Uh oh!

Contributors

Uh oh!

Languages

Folders and files

Latest commit

History

Repository files navigation

microbenchmark

Table of contents

Installation

Quick start

arguments

match(function)

Scenario

Constructor

run(warmup=0)

cli(argv=None)

ScenarioGroup

Creating a group

run(warmup=0)

cli(argv=None)

BenchmarkResult

Fields

percentile(p)

p95 and p99

to_json() and from_json()

CLI

Usage

Options

Exit codes

Example

Comparison with alternatives

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases 3

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

`match(function)`

`run(warmup=0)`

`cli(argv=None)`

`run(warmup=0)`

`cli(argv=None)`

`percentile(p)`

`p95` and `p99`

`to_json()` and `from_json()`

Packages