Re: [benchmark] Histogram as archetype of pixel iteration and access

From: Mateusz Loskot (mateusz_at_[hidden])
Date: 2019-06-24 20:44:29

On Fri, 21 Jun 2019 at 23:57, Mateusz Loskot <mateusz_at_[hidden]> wrote:
>
> In our benchmarking prototype, I've added [1] new benchmark of various
> pixel iteration and access methods for a contiguous images composed
> of homogeneous pixels with single 8-bit integer channel (grayscale).
>
> The implementation is available in histogram_gray8.cpp [2] and
> currently it is only Celero-based and Google-based is coming soon.
> There is also CSV file [3] (viewable on GitHub as a nice table)
> with results based on MSVC 19.21 (VS2019).
>
> Apart from the purpose of benchmarking, the histogram_gray8.cpp
> provides a catalogue of available pixel iteration and access techniques
> in a single place, easy to compare.
> I think this particular aspect may be useful for our GSoC students.
>
> [1] https://github.com/boost-gil/benchmark-prototype/commit/7eb74b84a58012452287b49f67e12cf4da9f82c0
> [2] https://github.com/boost-gil/benchmark-prototype/blob/benchmark/benchmark/celero/histogram_gray8.cpp
> [3] https://github.com/boost-gil/benchmark-prototype/blob/benchmark/benchmark/celero/histogram_gray8_msvc1421.csv

FYI,
I have included HTML pages with sample charts
https://github.com/boostorg/gil/tree/benchmark/benchmark/celero/charts

Those were generated with Bokeh visualization library and a little utility
script transforming Celero's CSV to Bokeh's charts:
https://github.com/mloskot/pycelerograph
which I have just updated for Bokeh 1.2.0+

-- 
Mateusz Loskot, http://mateusz.loskot.net