Hook 1.4 , ADMC compression return!

[Nania Francesco]

Hook 1.4 is released!
improved better compression, more faster in decompression!

new instructions:

To compress: hook c memsize in out
To decompress: hook d in out

Example compress: Hook c 256 in out
Example decompress: Hook d in out

only from
http://heartofcomp.altervista.org/

[LovePimple]

Thanks Francesco!

[Skymmer]

If somebody interested in tests, I've attached the XLS file which contains comparing of Hook v1.3 and v1.4 on 16 different files.
Some notes: 2 columns per each version - time in sec. and size in bytes.
RED color means worser and GREEN means better. Some values are BOLDed. Its the best achievements. Tested with 256 MB.

[LovePimple]

Thanks for sharing your test results.

[Nania Francesco]

Thanks for all test's!

[pat357]

Thanks for all test's!

Congrats to your new Hook version !
Why does the new Hook perform a bit less on .WAV files ?
Would it be possible to implement your Flashzip "filters" for WAV files in Hook ?

[Nania Francesco]

The problem of Hook 1.3 is in Wave compression! Hook 1.4 is more stable but in the future version surely I will insert something similar to the engine of Flashzip!

[Skymmer]

Thanks for all test's!

Thanks for sharing your test results.

You're welcome !

By the way, if we are talking about Dynamic Markov compression here, can somebody enlighten me about it. As I see its not so popular among developers and AFAIK there are only two programs based on it. Hook and Ocamyd. So why it is ? Is it just hard to implement or its just "weak" cause for example Hook takes not so high ranks in MFC test.

[Black_Fox]

Ocamyd's pretty strong, so that's surely not an issue. It may be a mere popularity problem, like when everybody uses range/arithmetic coding and only few use asymetric binary system, though they are similar in resulting performance...

[osmanturan]

By the way, if we are talking about Dynamic Markov compression here, can somebody enlighten me about it. As I see its not so popular among developers and AFAIK there are only two programs based on it. Hook and Ocamyd. So why it is ? Is it just hard to implement or its just "weak" cause for example Hook takes not so high ranks in MFC test.

I keep out DMC from my "toys". Because, DMC needs too much memory for effective compression IMO. Though, it's pretty fast. But, memory requirement is a big problem for me. BTW, as a note, technically DMC=Bitwise PPM.