/FFdecsa/ChangeLog

  1- created
  2
  3- released 0.0.1
  4
  5- simplified s, A, B
  6
  7- released 0.0.2
  8
  9- simplified nxt=
 10
 11- released 0.0.3
 12
 13- removed commented code
 14- code formatting
 15
 16- released 0.0.4
 17
 18- kk now unsigned char
 19- removed 64 bit ints
 20
 21- released 0.0.5
 22
 23- created decrypt_2ts
 24
 25- released 0.0.6
 26
 27- renamed files
 28- created decrypt_many_ts, removed others
 29- external interface has 2 functions only: set_cws() and decrypt_many_ts()
 30- reformatted code
 31- reimplemented s12,s34,s56,s7
 32- unsigned char become int for table optimization
 33
 34- released 0.0.7
 35
 36- optional icc compiler
 37- kk now 0..55
 38- decrypt_many_ts really works (no parallelism yet)
 39- added get_cws() to interface
 40- created stream.c
 41- created key_schedule_stream, using iA[] and iB[]
 42
 43- released 0.0.8
 44
 45- decrypt_many_ts() makes a group, sorts the packets, processes them
 46- preliminar stream_cypher_group() created
 47- parallel computing activated
 48- huge speed increase (+500%) thanks to stream_cypher_group()
 49
 50- released 0.0.9
 51
 52- block_cypher_group() created (no parallelism yet)
 53
 54- released 0.0.10
 55
 56- block_cypher_group() has 56 simple iterations
 57- block_cypher_group() doesn't shift registers anymore
 58
 59- released 0.0.11
 60
 61- some parallelization on block_cypher_group()
 62
 63- released 0.0.12
 64
 65- better parallelization of block_cypher_group()
 66
 67- released 0.0.13
 68
 69- block_cypher() was still called by error when N=23
 70- speed is now 109Mbit/s on AMD XP2000+ CPU
 71
 72- released 0.0.14
 73
 74- stream_cypher_group() has a init and normal variant
 75- A[0]-A[9] instead of A[1]-A[10], same for B
 76- implemented virtual shift of A and B
 77- speed is now 117Mbit/s on AMD XP2000+ CPU
 78
 79- released 0.0.15
 80
 81- better optimization of E and F in the stream cypher
 82- speed is now 119Mbit/s on AMD XP2000+ CPU
 83
 84- released 0.0.16
 85
 86- removed some debug overhead
 87- speed is now 120Mbit/s on AMD XP2000+ CPU
 88
 89- released 0.0.17
 90
 91- don't move packets with residue anymore
 92- speed is now 123Mbit/s on AMD XP2000+ CPU
 93
 94- released 0.0.18
 95
 96- solved alignment problems
 97- search groupable packets even beyond ungroupable ones
 98  (more speed in some real world cases)
 99- created decrypt_many_ts2(), useful with circular buffers
100
101- released 0.0.19
102
103- removed old code
104
105- released 0.0.20
106
107- partially converted code to size-independent group
108- icc doesn't work with optimizations on
109
110- released 0.1.1
111
112- merge loops on block_decypher (speed++ gcc, speed-- icc)
113- transposition are now functions (speed-- icc)
114- icc works again (compiler bug work around?)
115
116- released 0.1.2
117
118- better use of COPY8 &co
119- better flags for gcc
120- removed old code
121
122- released 0.1.3
123
124- int and not char in block cypher (speed++++++ gcc, speed-- icc)
125
126- released 0.1.4
127
128- group abstraction finally implemented
129- support for group width 64
130
131- released 0.1.5
132
133- group 64 mmx implemented (speed++ gcc)
134
135- released 0.1.6
136
137- more parallelism in block cypher (speed++ gcc)
138- transposition before and after block (disabled because of no speed gain yet)
139
140- released 0.1.7
141
142- more parallelism in block cypher (speed++ gcc)
143- transposition before and after block enabled (speed++ gcc)
144- gcc options (unrolled 500) speed gcc++
145
146- released 0.1.8
147
148- reworked FFN_ALL_* constants (speed++++ gcc) 
149
150- released 0.1.9
151
152- transposition in block as inlined functions
153- group abstraction working well
154
155- released 0.1.10
156
157- group 128 sse implemented, but batch is 64 mmx (not faster than group 64 mmx)
158
159- released 0.1.11
160
161- lot of code polishing and dead code elimination
162- better and more debug output
163
164- released 0.1.12
165
166- name change: FFdecsa
167
168- released 0.2.0
169
170- separated test cases
171- corrected all group_modes (now called parallel_modes)
172- parallel 128 8 char implemented
173- parallel 64 long implemented
174- parallel 128 2 long implemented
175- parallel 128 2 mmx implemented (incredibly slow, the compiler is very confused)
176- parallel 128 16 charA implemented (very slow compilation)
177- parallel 128 16 char implemented
178- renamed softcsa* to FFdecsa*
179
180- released 0.2.1
181
182- new external interface (based on ranges)
183
184- released 0.2.2
185
186- can be compiled with g++ too
187- using g++ the code is 3% faster!
188- external interface: function name changing and new functions
189- a group of ranges is now called a cluster
190- renamed autogenerated files
191
192- released 0.2.3
193
194- written docs
195- removed unneeded files
196- added Copyright and license notes
197- reworked "logic"
198
199- released 0.3.0
200
201- Makefile reworked
202- misc fixes
203- added vdr patch
204
205- released 1.0.0 (public release)
206