This was popularized by Yann Collet's work on <a href="https://github.com/Cyan4973/FiniteStateEntropy" rel="nofollow">https://github.com/Cyan4973/FiniteStateEntropy</a>
Almost all current implementations (except TurboANX) are in general not better than huffman coding.
See: <a href="https://sites.google.com/site/powturbo/entropy-coder" rel="nofollow">https://sites.google.com/site/powturbo/entropy-coder</a><p>Better test this yourself on your own data with TurboBench:
<a href="https://github.com/powturbo/TurboBench" rel="nofollow">https://github.com/powturbo/TurboBench</a>
detailed list of implementations found <a href="http://encode.ru/threads/2078-List-of-Asymmetric-Numeral-Systems-implementations" rel="nofollow">http://encode.ru/threads/2078-List-of-Asymmetric-Numeral-Sys...</a>