State-of-the-art lossless data compression in C++. The goal is to provide clean APIs and really fast implementation. It includes compression codecs (Run Length coding, Exp Golomb coding, Huffman, Range, LZ4, Snappy, ANS, Context Mixers, PAQ derivatives), bit stream manipulation, and transforms such as Burrows-Wheeler (BWT) and Move-To-Front, etc ...
For more details, check https://github.com/flanglet/kanzi/wiki.
Credits
Matt Mahoney, Yann Collet, Jan Ondrus, Yuta Mori, Ilya Muravyov, Neal Burns, Fabian Giesen, Jarek Duda
Disclaimer
Use at your own risk. Always keep a backup of your files.
i7-7700K @4.20GHz, 32GB RAM, Ubuntu 18.04
g++ version 7.3.0
Kanzi version 1.5 C++ implementation. Block size is 100 MB.
| Compressor | Encoding (sec) | Decoding (sec) | Size |
|---|---|---|---|
| Original | 211,938,580 | ||
| Kanzi -l 1 | 1.6 | 1.1 | 80,003,837 |
| Kanzi -l 1 -j 12 | 0.6 | 0.4 | 80,003,837 |
| Gzip 1.6 | 6.0 | 1.0 | 68,227,965 |
| Gzip 1.6 -9 | 14.3 | 1.0 | 67,631,990 |
| Kanzi -l 2 | 4.0 | 2.7 | 63,878,466 |
| Kanzi -l 2 -j 12 | 1.5 | 1.0 | 63,878,466 |
| Zstd 1.3.3 -13 | 11.9 | 0.3 | 58,789,182 |
| Brotli 1.0.5 -9 | 94.3 | 1.4 | 56,289,305 |
| Lzma 5.2.2 -3 | 24.3 | 2.4 | 55,743,540 |
| Kanzi -l 3 | 8.4 | 5.8 | 55,594,153 |
| Kanzi -l 3 -j 12 | 3.3 | 2.0 | 55,594,153 |
| Bzip2 1.0.6 -9 | 14.1 | 4.8 | 54,506,769 |
| Zstd 1.3.3 -19 | 45.2 | 0.4 | 53,977,895 |
| Kanzi -l 4 | 12.2 | 10.8 | 51,795,306 |
| Kanzi -l 4 -j 12 | 4.4 | 3.5 | 51,795,306 |
| Kanzi -l 5 | 15.6 | 15.5 | 49,455,342 |
| Kanzi -l 5 -j 12 | 5.5 | 5.0 | 49,455,342 |
| Lzma 5.2.2 -9 | 65.0 | 2.4 | 48,780,457 |
| Kanzi -l 6 | 24.9 | 24.7 | 46,485,165 |
| Kanzi -l 6 -j 12 | 7.7 | 7.6 | 46,485,165 |
| Tangelo 2.4 | 83.2 | 85.9 | 44,862,127 |
| zpaq v7.14 m4 t1 | 107.3 | 112.2 | 42,628,166 |
| zpaq v7.14 m4 t12 | 108.1 | 111.5 | 42,628,166 |
| Kanzi -l 7 | 56.5 | 57.5 | 41,838,503 |
| Kanzi -l 7 -j 12 | 22.0 | 21.8 | 41,838,503 |
| Tangelo 2.0 | 302.0 | 310.9 | 41,267,068 |
| Kanzi -l 8 | 79.0 | 79.0 | 40,844,691 |
| Kanzi -l 8 -j 12 | 31.0 | 30.8 | 40,844,691 |
| zpaq v7.14 m5 t1 | 343.1 | 352.0 | 39,112,924 |
| zpaq v7.14 m5 t12 | 344.3 | 350.4 | 39,112,924 |
i7-7700K @4.20GHz, 32GB RAM, Ubuntu 18.04
g++ version 7.3.0
Kanzi version 1.4 C++ implementation. Block size is 100 MB. 1 thread
| Compressor | Encoding (sec) | Decoding (sec) | Size |
|---|---|---|---|
| Original | 100,000,000 | ||
| Kanzi -l 1 | 1.21 | 0.66 | 35,611,290 |
| Kanzi -l 2 | 2.23 | 1.63 | 28,468,601 |
| Kanzi -l 3 | 4.21 | 2.96 | 25,517,555 |
| Kanzi -l 4 | 5.44 | 4.15 | 22,512,813 |
| Kanzi -l 5 | 7.62 | 6.12 | 21,934,022 |
| Kanzi -l 6 | 14.69 | 12.72 | 20,791,492 |
| Kanzi -l 7 | 22.48 | 22.44 | 19,613,190 |
| Kanzi -l 8 | 29.53 | 29.72 | 19,284,434 |
The C++ code can be built on Windows with Visual Studio and Linux with g++ and clang++. Porting to other operating systems should be straightforward.
This version of VS uses C++03 (AFAIK). Unzip the file "Kanzi_VS2008.zip" in place. The project generates a Windows 32 binary. Multithreading is not supported with this version.
This version of VS uses C++11 (AFAIK). Unzip the file "Kanzi_VS2015.zip" in place. The project generates a Windows 64 binary. Multithreading is supported with this version.
Go to the source directory and run 'mingw32-make.exe'. The Makefile contains all the necessary targets. Tested successfully on Win64 with mingw-w64 g++ 8.1.0. Multithreading is supported.
Go to the source directory and run 'make'. The Makefile contains all the necessary targets. Tested successfully on Ubuntu 18.04 with g++ 7.3.0, clang++ 6.0.0 and icc 19.0.0.117. Multithreading is supported with g++ version 5.0.0 or newer.