PACKET v.0.01 new fast compressor !

[Nania Francesco]

Packet file compressor.

Packet is new experimental LZP+ARI very faster compressor.
Requires 4 MB Memory free.

Copyright ? 2008 by Nania Francesco Antonio (Italy ME)
All rights reserved. Use only for TESTING.


LINK:
http://www.winturtle.netsons.org/packet.zip

[encode]

Thank you!

[LovePimple]

Thanks Francesco!

[LovePimple]

Quick test...

A10.jpg > 845,534
AcroRd32.exe > 2,054,947
english.dic > 909,388
FlashMX.pdf > 4,122,762
FP.LOG > 1,230,593
MSO97.DLL > 2,410,700
ohs.doc > 1,081,920
rafale.bmp > 1,557,830
vcfiu.hlp > 1,190,190
world95.txt > 1,070,231

Total = 16,474,095 bytes


ENWIK8 > 37,637,275 bytes

Compression Time: 7.748 Seconds

[SvenBent]

Quick test

Quake2.bin (quake2 cd image w/audio tracks) 733.071.360 bytes


Commpressor - compressed size - decomp time
RZM 0.07h - 461.162.427 - 00:02:00.938
Packet 0.01 - 592.710.674 - 00:02:47.522


Time measured with acutimer v1.2

RZM compress much better and is faster to decompress.
just i only measured decompression time because i find it more important than compression time

-- edit --
fixed that i had reversed the decomrpession time.
my conclusion was correct. rzm deccompressed faster

[SvenBent]

Quick teste again


Extremeties (quake2 expasion cd image, only data) 389.808.128 bytes


'Compressor - Compressed size - Comp time - Decomp time
RZM 0.07H - 113.568.796 bytes - 00:05:06.413 - 00:00:42.426
Packet 0.01 - 235.540.043 bytes - 00:00:38.943 - 00:01:57.675


same results as before
RZM compresses alot better and decomrpesses faster
Packet is faster to compress

and yes Packet was actually slower in decompression than compression

[Nania Francesco]

I don't believe that Packet is so slow in decompression but is currently surely it! I believe is a problem that can be resolved! I have to resolve him/it really in all of my programs in the phase of decompression! However Packet is born as an experiment of improvement of LZP and demonstrative! In fact I have intention in the briefest time to publish on my site a description of my system Advanced LZP and to make him/it of everybody available!

[SvenBent]

I will b back soon with some more test of the PAcket decompression time.

to be honnest i worked on both the rzm and packet on the same time.
I'm running af quadcore q6600 (@3ghz and) and drives in raid 0 so i don't think it the reason. But i will to a complete serial testing of decompression time right now.


just one question. does packet works with big files ?

[SvenBent]

new test under propoer condition

Nebula  emulator + 3 rom files (43,4 MB)

Packet 0.01     23.1 MB   00:00:01.641
RZM 0.07h       15.8 MB   00:00:04.705
Tor v0.4 -12    19.0 MB   00:00:01.366


WarCraft 3 install dir with saves in 7.z container (1,36 GB)

Packet 0.01     1.31 GB   00:02:44.287
RZM 0.07h       1.16 GB   00:04:33.038
Tor v0.4 -12    1.17 GB   00:01:27.952

Seems that packet IS faster then rzm at decompression. my prior results mus be incorrect

I tried using Tor as well
Tor seem to compress better and decompress faster.

once again it didn't measure compression time. because they are not important to me

[encode]

Packet is new experimental LZP+ARI very faster compressor.

Of course, potentially LZP should be slower than ROLZ/LZ77 at decompression. RZM just uses more expensive LZ-output encoding...

[Black_Fox]

Thanks Francesco!

to be honnest i worked on both the rzm and packet on the same time. I'm running af quadcore q6600 (@3ghz and) and drives in raid 0 so i don't think it the reason

As observed some time ago with CCM, even running each program on different core(s) and providing it with different RAM area can result in (de)compression slowdown...

[Nania Francesco]

For SvenBent! currently the only ones two programs of mine that support >2GB are only WinTURTLE and FlashZIP! Hi!

[Nania Francesco]

Packet v.0.02 Released

In New Version:
- Improved better WAVE compression.
- Improved faster compression/decompresson.
- Support for NTFS files [>2GB].

LINK:
http://www.winturtle.netsons.org/packet.zip

[LovePimple]

Thanks Francesco!

[LovePimple]

Quick test...

A10.jpg > 845,535
AcroRd32.exe > 2,054,948
english.dic > 909,389
FlashMX.pdf > 4,122,763
FP.LOG > 1,230,594
MSO97.DLL > 2,410,701
ohs.doc > 1,081,921
rafale.bmp > 1,557,831
vcfiu.hlp > 1,190,191
world95.txt > 1,070,232

Total = 16,474,105 bytes



ENWIK8 > 37,637,276 bytes

Compression Time: 8.733 Seconds

[LovePimple]

Another quick test...

Test file: Scribble.wav (9,249,480 bytes)

PACKET v0.01 compresses scribble.wav to 74,108 bytes.
PACKET v0.02 compresses scribble.wav to 71,704 bytes.

Improved better WAVE compression.

The 'improved WAVE compression' is working.

[Matt Mahoney]

packet 0.02 compresses 1 byte larger but wins for a smaller packet.exe.zip.
http://cs.fit.edu/~mmahoney/compression/text.html#3345

[Shelwien]

Matt,

Wonder if you could test fpaq0pv4,v4A,v4B from these threads:
http://encode.ru/forums/index.php?ac...ic=671#msg9437
http://encode.ru/forums/index.php?ac...ic=679#msg9654

PM is really invisible here, it seems...

[Nania Francesco]

Thanks for testing packet Matt ! Hi !

[Nania Francesco]

Currently Packet uses the engine of FlashZip and doesn't have anything to whether to see with Fpaq0p that would be indeed more powerful and slower ! In the next version I will double the speed of decompression and I will improve the compression!

[Matt Mahoney]

Matt,

Wonder if you could test fpaq0pv4,v4A,v4B from these threads:
http://encode.ru/forums/index.php?ac...ic=671#msg9437
http://encode.ru/forums/index.php?ac...ic=679#msg9654

PM is really invisible here, it seems...

I added v4 and v4a, but v4b gave me this error on my Athlon-64 3500+

Fatal Error: This program was not built to run on the processor in your system. The allowed processors are: Intel(R) Pentium(R) 4 and compatible Intel processors. Enables new optimizations in addition to Intel processor-specific optimizations.

http://cs.fit.edu/~mmahoney/compression/text.html#5586

Edit: PM works, I just didn't think to check it. I haven't been to the forum lately due to lots of other work.

[Nania Francesco]

Packet v.0.03 file compressor. Only for testing
Copyright ? 2008 by Nania Francesco Antonio (Italy). All rights reserved.
Packet is new experimental LZ77+ARI order[0] very faster compressor.
Requires 3 MB Memory free for compression and 1 MB for decompression.
New
- Improved better compression.
- Low Memory.
- Very faster decompresson.
- Support for NTFS files.

Instructions: Packet [options] [mode] infile outfile
options:
d -> decompress
c [mode] -> use for normal compression [FAST]
c x [mode] -> use for best compression [SLOW]
mode:
1 - 6 -> set mode only in compression [1=fast - 6=slow]
LINK:
http://www.winturtle.netsons.org/packet.zip

[LovePimple]

Thanks Francesco!

[Shelwien]

@Matt:
> I added v4 and v4a, but v4b gave me this error on my Athlon-64 3500+
> Fatal Error: [...] The allowed processors are: Intel(R) Pentium(R) 4 and compatible Intel processors.

Ah. So here we have a good example for that thread: http://encode.ru/forum/showthread.php?t=32

> http://cs.fit.edu/~mmahoney/compression/text.html#5586

Thanks, but please test 4B too. Here I patched out the "GenuineIntel" checks, hope it'll help:
http://shelwien.googlepages.com/fpaq0pv4B1.exe

Also here's my E2160 fpaq0p comparison.
http://shelwien.googlepages.com/fpaq0pv4B1.htm
Actually I was going to test "packet" instead, but its author likes to overwrite his archives
and testing only the last version seemed boring

> PM works, I just didn't think to check it.

That's why I called it "invisible". Wonder if Ilya can make it blink if there's a PM or something.

[Black_Fox]

Actually I was going to test "packet" instead, but its author likes to overwrite his archives
and testing only the last version seemed boring

Versions 0.01 to 0.03 can be found here http://www.mediafire.com/?9medvcms91y

[Shelwien]

Thanks, Black_Fox

But, now, this is really something: http://shelwien.googlepages.com/packet003.htm

Seems that v0.03 fails to restore offsets after E8 in "normal" mode,
and completely loses most of matches (replaced with zeroes) in
"x" mode.

But instead, its fast

[Nania Francesco]

Packet v.0.03b file compressor. Only for testing
Copyright ? 2008 by Nania Francesco Antonio (Italy). All rights reserved.
Corrected All Bugs
LINK:
http://www.winturtle.netsons.org/packet.zip

[Shelwien]

E2160 @ 9x360=3.24, DDR2-800 5-5-5-18 @ 900
http://shelwien.googlepages.com/packet003b.htm
Works now.

[Nania Francesco]

I wanted to know Shelwien thing think of it of this program ?

[Shelwien]

Sorry, I don't think much of whatever LZ.
Its very wide and theoretically interesting topic, but I don't see any practical application for most of new LZ engines.
Well, if an archiver appeared which would compress significantly better than rar,
and work faster at the same time - maybe I'd consider using it.
But for now even 7-zip isn't enough as rar has better management of LZH/ppmd/delta, so there're still a lot of cases where 7-zip is unable to win without major option tweaking.

As to "packet", well, it seems comparable to rar -m1, while being 2x+ faster in decoding.
Also options seem to mainly affect the encoding speed