https://github.com/scivision/python-performance

Performance benchmarks of Python, Numpy, etc. vs. other languages such as Matlab, Julia, Fortran.
https://github.com/scivision/python-performance

benchmark fortran gfortran julia matlab microbenchmark numba numpy performance

Last synced: 14 days ago
JSON representation

Performance benchmarks of Python, Numpy, etc. vs. other languages such as Matlab, Julia, Fortran.

• Host: GitHub
• URL: https://github.com/scivision/python-performance
• Owner: scivision
• License: mit
• Created: 2014-11-03T16:49:26.000Z (about 10 years ago)
• Default Branch: main
• Last Pushed: 2024-04-17T17:47:06.000Z (7 months ago)
• Last Synced: 2024-05-02T03:09:31.403Z (7 months ago)
• Topics: benchmark, fortran, gfortran, julia, matlab, microbenchmark, numba, numpy, performance
• Language: Python
• Homepage: https://scivision.dev/speed-of-matlab-vs-python-numpy-numba/
• Size: 708 KB
• Stars: 27
• Watchers: 4
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.txt

Awesome Lists containing this project

README

        
# Python Performance

![Actions Status](https://github.com/scivision/python-performance/workflows/ci/badge.svg)

All benchmarks are platform-independent (run on any computing device with appropriate hardware).

CuPy tests require an NVIDIA GPU with CUDA toolkit installed.

## Install

This command prepares Python prereqs:

```sh

pip install -r requirements.txt

```

C and Fortran benchmarks requires building first using CMake.

```sh

cmake --workflow --preset default

```

## Usage

Iterative benchmarks, here using the pisum algorithm:

```sh

python Pisum.py

```

![Pi (Machin) benchmark](./gfx/pisum.png)

---

Matrix Multiplication benchmarks:

```sh

python Matmul.py

```

![Matrix Multiplication benchmark](./gfx/matmul.png)

### Hypotenuse

For **Python**,

[numpy.hypot()](https://numpy.org/doc/stable/reference/generated/numpy.hypot.html)

is faster up to about a hundred elements, then

[numpy.sqrt(x**2 + y**2)](https://numpy.org/doc/stable/reference/generated/numpy.sqrt.html)

becomes slightly faster.

The benefit of `hypot()` is to not overflow for arguments near REALMAX.

For example, in Python:

```python

from math import sqrt, hypot

a=1e154; hypot(a,a); sqrt(a**2+a**2);

1.414213562373095e+154

inf

```

For **Fortran**, observe that with Gfortran compiler that `sqrt(x**2 + y**2)` is slightly faster than `hypot(x,y)` in general across the tested array sizes.

Execute the Hypot speed test by:

```sh

python Hypot.py

```

![hypot() vs rsq()](./gfx/hypot.png)

## Notes

### Julia

Julia binaries are often downloaded to a particular directory.

Python doesn't pickup `.bash_aliases`, which is commonly used to point to Julia.

### MKL selection

https://software.intel.com/en-us/articles/intel-mkl-link-line-advisor

We give a hint to CMake where your MKL libraries on.

For example:

```sh

MKLROOT=/opt/intel/mkl cmake ..

```

Of course this option can be combined with `FC`.

You can set this environment variable permanently for your convenience

(normally you always want to use MKL) by adding to your `~/.bashrc` the

line:

```sh

export MKLROOT=/opt/intel/mkl

```