Commit graph

10 commits

Author SHA1 Message Date
9f3c1d0168
remove markov chain from arithmetic compression in preparition for move to front transform 2025-12-04 19:33:49 -08:00
4a963eb383
the start of a rewrite using what was learned from bzip2 2025-11-28 17:47:27 -08:00
da83f9a5d0
benchmark, test suite, and run length encoding 2025-10-27 13:15:58 -07:00
2123636291
swap to word4 instead of word8 sizes
This isn't beter yet, but I have a hunch it's to to the Maybe in
the compressed word4 list type, which I can get rid of
2025-01-08 19:45:34 -08:00
46749895bc
parallel compression and decompression
I've selected a chunk size of 8000 bytes rather arbitrarily, but it
seems to work just fine. The bigger problem is that the simplified
final fraction makes the running fraction in decompression's foldl
more complicated as time goes on, making it around 50 times slower
than compression. It takes just under an hour to decompress the
Great Gatsby.
2024-12-23 08:29:31 -05:00
791fff6107
add arithmetic coding 2024-12-21 12:28:00 -05:00
Jack Wines
74b18d29a0
Add LICENSE 2024-04-15 20:35:54 -07:00
43afd380c4
decompression works
I also added a cli interface
2024-04-15 19:18:24 -07:00
b8cb909fc9
intmap-based priority queue change
speedup is significant. Now does >5MB RAW images with ease
2024-04-14 23:34:16 -07:00
874d0b161d
compression works
Word16 seems to give the best compression ratio. It takes a very long
time to run on >5MB files. Decompression is not implemented yet, but I'm
confident it works.
2024-04-13 02:13:24 -07:00