Model size:
Amount of memory needed for decompression. Must include buffer size, plus, size of index table.
Largest optimised match:
Max optimize lookahead for Optimal Parsing. Zero disables optimizing. Note that looks like with zero param. WinRK uses a different structure for match search (hash chains?). Some testing results with world95.txt:
0: 561,889 bytes
1: 615,367 bytes [!!]
2: 610,408 bytes
...
512: 505,151 bytes
Audio model order:
Not used with ROLZ2.
Bit-stream model order:
Also not used with ROZL2.
Other options are self explanatory. http://www.encode.su/images/smilies/wink.gif
