perfbench

About

perfbench measures execution time of code snippets with Timeit and uses Plotly to visualize the results.

Feature

It is possible to select measurement modes.
It is possible to switch between layout sizes dynamically.
It is possible to switch between axes scales dynamically.
It is possible to switch between subplots dynamically.
The result of the benchmark can be saved locally as a html.
The result of the benchmark can be saved locally as a png. Requires installation oforca. When not to use the function, you do not need to install orca separately.

Compatibility

perfbench works with Python 3.5 or higher.

Dependencies

tqdm(4.6.1 or higher.)
cerberus(1.1 or higher.)
plotly(3.0.0 or higher)
notebook(6.0 or higher.)
ipywidgets(7.2 or higher.)

Installation

pip install perfbench

Usage

Plotting a single figure. Here is the demonstration.

import numpy as np
from perfbench import *


bm = Benchmark(
    datasets=[
        Dataset(
            factories=[
                lambda n: np.random.uniform(low=-1., high=1., size=n).astype(np.float64),
            ],
            title='float64'
        )
    ],
    dataset_sizes=[2 ** n for n in range(26)],
    kernels=[
        Kernel(
            stmt='np.around(DATASET)',
            setup='import numpy as np',
            label='around'
        ),
        Kernel(
            stmt='np.rint(DATASET)',
            setup='import numpy as np',
            label='rint'
        )
    ],
    xlabel='dataset sizes',
    title='around vs rint',
)
bm.run()
bm.plot()

Plotting multiple plots on a single figure. Here is the demonstration.

import numpy as np
from perfbench import *


bm = Benchmark(
    datasets=[
        Dataset(
            factories=[
                lambda n: np.random.uniform(low=-1., high=1., size=n).astype(np.float16),
            ],
            title='float16'
        ),
        Dataset(
            factories=[
                lambda n: np.random.uniform(low=-1., high=1., size=n).astype(np.float32),
            ],
            title='float32'
        ),
        Dataset(
            factories=[
                lambda n: np.random.uniform(low=-1., high=1., size=n).astype(np.float64),
            ],
            title='float64'
        )
    ],
    dataset_sizes=[2 ** n for n in range(26)],
    kernels=[
        Kernel(
            stmt='np.around(DATASET)',
            setup='import numpy as np',
            label='around'
        ),
        Kernel(
            stmt='np.rint(DATASET)',
            setup='import numpy as np',
            label='rint'
        ),
    ],
    xlabel='dataset sizes',
    title='around vs rint',
)
bm.run()
bm.plot()

Switching between layout sizes.

import numpy as np
from perfbench import *


bm = Benchmark(
    datasets=[
        Dataset(
            factories=[
                lambda n: np.random.uniform(low=-1., high=1., size=n).astype(np.float64),
            ],
            title='float64'
        )
    ],
    dataset_sizes=[2 ** n for n in range(26)],
    kernels=[
        Kernel(
            stmt='np.around(DATASET)',
            setup='import numpy as np',
            label='around'
        ),
        Kernel(
            stmt='np.rint(DATASET)',
            setup='import numpy as np',
            label='rint'
        )
    ],
    xlabel='dataset sizes',
    title='around vs rint',
    layout_sizes=[
        LayoutSize(width=640, height=480, label='VGA'),
        LayoutSize(width=800, height=600, label='SVGA'),
        LayoutSize(width=1024, height=768, label='XGA'),
        LayoutSize(width=1280, height=960, label='HD 720p'),
    ]
)
bm.run()
bm.plot()

Save as a html.

# same as above
bm.save_as_html(filepath='/path/to/file')

Save as a png.

# same as above
bm.save_as_png(filepath='/path/to/file', width=1280, height=960)

Other Here are a few examples.

License

This software is released under the MIT License, see LICENSE.

Metadata

Release files for perfbench 5.0.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for perfbench 5.0.0
File	Size	Uploaded	Provenance
perfbench-5.0.0.tar.gz	227.4 kB	Jan 9, 2021	Without provenance	Details

Release files / perfbench-5.0.0.tar.gz

Download URL	perfbench-5.0.0.tar.gz
Size	227.4 kB
Tags	Source
SHA-256 checksum How to use checksums	`1e3496824a59d32bb87a81b98da07de13b63cd1d68d9470338a1d815c7c914b7`
BLAKE2b-256 checksum How to use checksums	`395095897d4413c5b5db3b1a9acc75b2c62e900748d29e9893ac06039daa36f5`
Upload date	Jan 9, 2021
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/3.3.0 pkginfo/1.6.1 requests/2.25.1 setuptools/41.2.0 requests-toolbelt/0.9.1 tqdm/4.55.1 CPython/3.8.3`

Security for perfbench 5.0.0

Report project as malware

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.