Versions tested: 1.12
Optimal parameters: -o4
Links: http://www.compressc...
Authors: Michael Schindler
Algorithms: ST/BWT+ARI
Notable peformances: -


Szip uses a limited order Sort transform [ST]. This transformation is related to the Burrows-Wheeler transformation used in blocksorting compression methods. The difference is a fast, deterministic behaviour in time at some compression loss. And you can easily do it in hardware. [...]

As next step it uses a probability model for blocksorted files.

The last step is entropy encoding using a range coder. More info, including a GNU GPL sourcecode, can be found on This coder is based on an article published in 1979 and is belived to be patent-free.

the program documentation


The program is featured in the BWT comparison.

(Have info that should be added here? E-mail.)

The program uses full BWT with -o0 -switch.

Qualified configurations
Ver Rating CPR DPR S.E. R.E. Ratio C. kB/s D. kB/s
1.12 16 29 2 70 0 2.978 7851 8433
1.12 18 50 2 90 0 2.934 16341 8355
1.12 18 47 2 95 0 2.915 16799 9237

Version info

1.12, tar: yes, mode: 32-bit, crc32: 16177ae2, decoder: 27 kB, tested: 2010 Apr 16
compression: szip.exe <args> <src> <cfile>
decompression: szip.exe -d <cfile> <src>