asv

library0.6.5pypypiunverified

Airspeed Velocity (asv) is a tool for benchmarking Python packages over their lifetime. It is primarily designed to benchmark a single project over its lifetime using a given suite of benchmarks. The results are displayed in an interactive web frontend that requires only a basic static webserver to host. The current version is 0.6.5, released in September 2025, and it maintains a relatively active release cadence with several updates per year.

testing devops observability

pip install asv

github ↗package ↗docs ↗

INSTALL

IMPORT

SIG · ASV

asv

testingpythonv0.6.5

Install

3.1s avg

Import

—

Disk

36MB

Pass rate

10/ 10

Env Coverage10 / 10

glibc

3.9–3.13

musl

3.9–3.13

Install & Compatibility

Where this runs

tested against v0.6.5 · pip install

no network on importno background threads

Install × environment matrix

Each cell = how many times install + import succeeded across repeated harness runs. Partial = flaky.
glibc = Debian/Ubuntu slim · musl = Alpine Linux

musl

py 3.10–3.920 runs

installs and imports cleanly · install 0.0s · import 0.000s · 38.6MB

glibc

py 3.10–3.920 runs

installs and imports cleanly · install 3.1s · import 0.000s · 37MB

36MB installed

● package 36MB

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

asv

✓ asv <command>

✗ import asv; asv.run()

ASV is primarily a command-line interface (CLI) tool. While it uses Python for defining benchmarks, its core functionality is exposed via shell commands like `asv run`, `asv publish`, etc., rather than direct programmatic Python imports in user applications.

To get started, navigate to your project directory (or a new directory for benchmarks) and run `asv quickstart` to set up the basic configuration and an example benchmark file. Then, define your benchmarks in Python files within the `benchmarks/` directory, typically as classes with `time_` or `mem_` prefixed methods. Finally, execute `asv run` to collect benchmark data and `asv publish` to generate the interactive web report, which can be viewed with `asv browse`.

# 1. Initialize a new ASV benchmark suite # Run this in your project's root or a dedicated benchmark directory. asv quickstart # 2. Add a benchmark to benchmarks/benchmarks.py (example content below) # Create or edit 'benchmarks/benchmarks.py' with content like: # import numpy as np # # class TimeSuite: # def setup(self): # self.x = np.random.rand(1000, 1000) # # def time_sum_array(self): # np.sum(self.x) # # def time_dot_product(self): # np.dot(self.x, self.x) # # def teardown(self): # self.x = None # # class MemSuite: # def mem_full_array(self): # return np.random.rand(1000, 1000) # 3. Run the benchmarks asv run # 4. Publish the results to a web frontend asv publish # 5. Open the results in a web browser asv browse

asv --version

Debug

Known issues

breakingChanging benchmark names, class names, or parameter names within your `benchmarks.py` files can invalidate historical data in ASV's reports. This makes it difficult to compare new results with past performance for the renamed benchmarks, essentially starting a new series.

fix

Carefully consider benchmark naming conventions from the start. If changes are necessary, acknowledge that historical data for that specific benchmark series will be lost or fragmented. Documentation advises on pitfalls of changing `benchmark_name` attributes.

affects: All versions

gotchaASV is fundamentally a command-line tool (`asv quickstart`, `asv run`, `asv publish`, etc.). It's not typically imported as a Python library for direct programmatic use in application code. Attempting `import asv` to access benchmark running logic directly is not the intended usage model for most users.

fix

Interact with ASV via its command-line interface. Benchmark definitions are Python code, but the execution and reporting are driven by the `asv` command.

affects: All versions

gotchaWhen configuring benchmark dependencies, `asv.conf.json` can specify `environment_matrix` settings (e.g., Python versions, specific package versions). If you also use `conda` or `mamba` with an `environment.yml` file, the explicit dependencies listed in `environment_matrix` within `asv.conf.json` will override those in `environment.yml`.

fix

Be mindful of the precedence: `environment_matrix` in `asv.conf.json` takes priority. Ensure your dependency specifications are consistent or intentionally configured across both files if you use both methods.

affects: >=0.6.1

gotchaASV relies on Git or Mercurial for tracking project history and associating benchmark results with specific commits. While it can run benchmarks locally, its primary strength—historical performance tracking—requires an initialized Git or Mercurial repository.

fix

Ensure your project is under version control (Git or Mercurial) and that the `repo` path in `asv.conf.json` correctly points to your repository.

affects: All versions

gotchaThe acronym 'ASV' is also used in other contexts, such as Adaptive Servo-Ventilation (medical devices for sleep apnea) and ASV Inventions (motorcycle parts). When searching for or discussing the Python benchmarking tool, ensure the context is clear to avoid confusion.

fix

Specify 'Python asv' or 'Airspeed Velocity' when referring to this library.

affects: All versions

Upgrade

Version history

0.6.5latest on PyPI · released Sep 13, 2025

Audit

Dependencies

pythonrequiredRequired Python version.

Agent activity

36 hits · last 30 days

node

OpenAI (training)

Resources

githubhttps://github.com/airspeed-velocity/asv ↗

packagehttps://pypi.org/project/asv/ ↗

docshttps://asv.readthedocs.io/en/stable/ ↗