PIMPLE2 is here!

[encode]

Dedicated to data compression maniacs.

Memory usage: <128 MB
SFC: 10,640,280 bytes

Link:
pimple2.zip (77 KB)

No comments!

Please test it carefully! Also I very interested in comparison with things like WinUDA, PAQ6, ... (for speed/compression/mem use)

Enjoy!

[Black_Fox1]

Thanks encode Will do!

[LovePimple]

Dedicated to data compression maniacs.

Thats me!

Thanks Ilia!

[Nania Francesco]

Thanks Ilia! Pimple is very good Compressor!

[LovePimple]

ENWIK8 > 20,871,457 bytes

[Matt Mahoney]

enwik9 = 180,251,530 bytes. It moves up 1 spot from pimple with 1/4 as much memory. Compression time is about 5 hours.
http://cs.fit.edu/~mmahoney/compression/text.html# 1803

I think it would benefit from more memory on large files like this, but it is the best result for only 128 MB (I did not try other compressors at this setting. The comparison is for all compressors at max compression).

[encode]

Thanks for testing!

Initially PIMPLE2 was designed as super-high compression mode for PIM archiver. That's why I restrict memory usage to such number. During testing, I've found that it too slow for practical usage and not added it to the PIM.

[Piotr Tarsa]

Matt Mahoney
imo, your memory indications aren't very accurate. for example you've reported that TarsaLZP Jul 30 2007 used 42 mb of memory while it used more (about 60 mb). also memory indication for winace (only four megabytes) looks suspicious.

[LovePimple]

Initially PIMPLE2 was designed as super-high compression mode for PIM archiver. Thats why I restrict memory usage to such number. During testing, Ive found that it too slow for practical usage and not added it to the PIM.

Do you plan to improve PIMPLE2 by increasing memory usage?

[encode]

Do you plan to improve PIMPLE2 by increasing memory usage?

Currently Im busy with LZPM, but later, I hope I will improve PIMPLE2s speed and compression. To be honest, I tested PIMPLE2 with a larger model - in many cases this only hurts compression. Note that larger model not always means higher compression. A smaller model discards obsolete statistics more efficiently.

[LovePimple]

Currently Im busy with LZPM, but later, I hope I will improve PIMPLE2s speed and compression.