Benchmarks
==========
Setup
-----
There are a number Python modules providing APIs for histogram
calculations. Here we see how pygram11 performs in comparison to
numpy_, fast-histogram_, and boost-histogram_. Tests were performed on
an Intel i7-8850H 2.60Gz processor (6 physical cores, 12 threads).
Fast-histogram does not provide calculations for variable width bins,
so, when benchmarking variable width bins, we only compare to NumPy
and boost-histogram.
Results
-------
The results clearly show that pygram11 is most useful for input arrays
exceeding about 5,000 elements. This makes sense because the pygram11
backend has a clear and simple overhead: to take advantage of N
available threads we make N result arrays, fill them individually
(splitting the loop over the input data N times), and finally combine
the results (one per thread) into a final single result that is
returned.
For one dimensional histograms with fixed width bins pygram11 becomes
the most performant calculation for arrays with about 5,000 or more
elements (up to about 3x faster than the next best option and over 10x
faster than NumPy). Fast-histogram is a bit more performant for
smaller arrays, while pygram11 is always faster than NumPy and
boost-histogram.
.. image:: /_static/benchmarks/fixed1d.png
:width: 60%
:align: center
For two dimensional histograms with fixed width bins pygram11 becomes
the most performant calculation for arrays with about 10,000 or more
elements (up to about 3x faster than the next best option and almost
100x faster than NumPy). Fast-histogram is again faster for smaller
inputs, while pygram11 is always faster than NumPy and *almost* always
faster than boost-histogram.
.. image:: /_static/benchmarks/fixed2d.png
:width: 60%
:align: center
For one dimensional histograms with variable width bins pygram11
becomes the most performant option for arrays with about 10,000 or
more elements (up to about 8x faster than the next best option and
about 13x faster than NumPy).
.. image:: /_static/benchmarks/var1d.png
:width: 60%
:align: center
For two dimensional histograms with variable width bins pygram11
becomes the most performant option for arrays with about 5,000 or more
elements (up to 10x faster than the next best option).
.. image:: /_static/benchmarks/var2d.png
:width: 60%
:align: center
.. _fast-histogram: https://github.com/pybind/pybind11
.. _numpy: http://www.numpy.org/
.. _boost-histogram: https://boost-histogram.readthedocs.io/en/latest/