| Versions tested: | 2.02c, 2.02s |
| Optimal parameters: | default |
| Links: | ppmy_3c.rar ppmy_3c_sse.rar |
| Authors: | Eugene Shelwien |
| Algorithms: | CM |
| Notable peformances: | - |
PPMY, ContextMixing-based data compressor: v2.02(3c) of 02/06/2001; v2.02(3c+SSE) of 26/02/2003
ppmy is a bytewise coder based on precise symbol frequencies (numbers of occurences) and linear mixing of distributions from context orders with weights determined by some adhoc function.author's description, email 2009 April 5ppmy with SSE is [like old ppmy] plus the mixed probability of symbol previously seen in maxorder context is processed with adaptive mapping (aka Secondary Symbol Estimation) with linear interpolation. All statistics and model logic are implemented in floats, which is rather unique by itself, although slow (it wasn't at the time, in fact).
(Have info that should be added here? E-mail.)
| Ver | Rating | CPR | DPR | S.E. | R.E. | Ratio | C. kB/s | D. kB/s | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| <none> | ||||||||||||
| 2.02c | 14 | 13 | 15 | 10 | 1 | 3.097 | 84 | 83 | ||||
| 2.02s | 15 | +7.1% | + | +6.7% | 3.118 | +0.7% | 85 | +1.2% | 85 | +2.4% | ||