Three versions of LZ4 decoders for 68k processors.
Written by Arnaud Carré ( @leonard_coder )
LZ4 technology by Yann Collet ( https://lz4.github.io/lz4/ )
The fastest version use 3722 bytes and is suited for CPU without instruction cache ( 68000 ) Prefer lz4_normal.asm for cache instruction CPU ( >= 68020 ) If you really need a small footprint (bootsector) you can use lz4_smallest version, that is 72bytes only
If you're using LZ77 decoder in your demo, you can switch to this for sure. LZ4 has better packing ratio than LZ77 and it depacks faster
| source code | decoder size | speed factor |
|---|---|---|
| lz4_smallest.asm | 72 bytes | x 1.0 |
| lz4_normal.asm | 180 bytes | x 1.53 |
| lz4_fastest.asm | 3722 bytes | x 2.36 |
Speed factor is measured on ATARI ST by depacking a 105KiB packed exe demo file (167KiB unpacked).
Measured on the same machine, lz4_fastest.asm is 6.81 times faster than ARJ and 4.45 times faster than UPX ( please note packing ratio is not as good as ARJ or UPX )
lz4_frame.asm can be used to depack a LZ4 frame. You can generate LZ4 frame using standard LZ4 command line tool, like: lz4.exe -9 --no-frame-crc <input_file> <output_file>