Software Tuning, Performance Optimization & Platform Monitoring
Discussion around monitoring and software tuning methodologies, Performance Monitoring Unit (PMU) of Intel microprocessors, and platform monitoring
Announcements
This community is designed for sharing of public information. Please do not share Intel or third-party confidential information here.
1622 Discussions

New entropy coding - combining speed of Huffman with accuracy of arithmetic coding

Jarek_D_
Beginner
239 Views

There is a new entropy coding approach: asymmetric numeral systems (ANS), which can be seen as simpler alternative for arithmetic coding: using a single natural number as the state, instead of two to represent the range. One of its variant is as computationally demanding as Huffman coding (large alphabet), but handles fractional amounts of bits - giving it accuracy (compression rate) like for arithmetic coding.

As there are reference implementations for Intel processor of data compression algorithms ( http://software.intel.com/sites/products/documentation/hpc/ipp/ipps/ipps_ch13/ch13_0_Intro.html ), maybe it would be worth to create them also for ANS?

Some implementation: https://github.com/Cyan4973/FiniteStateEntropy

Preprint: http://arxiv.org/pdf/1311.2540

With best regards,

Jarek Duda

0 Kudos
0 Replies
Reply