1 | |
1 | |
|
|
2 | TODO: try unaligned copy again in decompressor |
|
|
3 | TODO: allow size-optimised binaries by avoiding unrolling |
|
|
4 | TODO: implement lzf_c_best in lzf. |
|
|
5 | TODO: fix lzf_c_best. |
|
|
6 | |
|
|
7 | 3.7 (unreleased) |
|
|
8 | - add lzf_c_best.c, a slower but better compressor. |
|
|
9 | - switch to a multiplicative hash (developed with Steinar Gunderson), |
|
|
10 | which is faster on modern cpus and compresses a bit better. The old |
|
|
11 | hash function which uses only shifts is still available. |
|
|
12 | - allow user configurable hash table slots, which makes it possible |
|
|
13 | to use e.g. 16 bit offsets for a smaller hashtable (if your data is |
|
|
14 | always < 64kb). |
|
|
15 | - use _WIN32, not WIN32, when testing for windows (fails with bcc), |
|
|
16 | patch by Tamas Tevesz. |
|
|
17 | - try to port to win64+gcc, which is non-posix but doesn't have _int64 |
|
|
18 | (reported by Maciej Adamczyk). |
|
|
19 | |
|
|
20 | 3.6 Mon Feb 7 17:37:31 CET 2011 |
2 | - fixed hash calculation in C♯ version (Tiago Freitas Leal). |
21 | - fixed hash calculation in C♯ version (Tiago Freitas Leal). |
|
|
22 | - unroll copy for small sizes, use memcpy for larger sizes, |
|
|
23 | greatly speeding up decompression in most cases. |
|
|
24 | - finally disable rep movsb - it's a big loss on modern intel cpus, |
|
|
25 | and only a small win on amd cpus. |
|
|
26 | - improve C++ compatibility of the code. |
|
|
27 | - slightly improve compressor speed. |
|
|
28 | - halved memory requirements for compressor on 64 bit architectures, |
|
|
29 | which can improve the speed quite a bit on older cpus. |
3 | |
30 | |
4 | 3.5 Fri May 1 02:28:42 CEST 2009 |
31 | 3.5 Fri May 1 02:28:42 CEST 2009 |
5 | - lzf_compress did sometimes write one octet past the given output |
32 | - lzf_compress did sometimes write one octet past the given output |
6 | buffer (analyzed and nice testcase by Salvatore Sanfilippo). |
33 | buffer (analyzed and nice testcase by Salvatore Sanfilippo). |
7 | |
34 | |