Experimental new ZCM file compressor!

[Nania Francesco]

ZCM Archiver by Nania Francesco Antonio (Italy)
Demo software. Only for testing

released version 0.93 !!


link:
http://heartofcomp.altervista.org/index.htm
Good testing....!

[jethro]

Impressive! Seriously, very good compression and speed is good too. Thank you Nania

[Shelwien]

http://nishi.dreamhosters.com/u/zcm01.htm

Compression is a little better than ccmx, but 3x slower.

Apparently IntelC was used to compile it... did you define things like
#define _CRT_SECURE_NO_DEPRECATE
#define _CRT_DISABLE_PERFCRIT_LOCKS
?

Also, are counters properly aligned by cache lines?

[Nania Francesco]

CCMX = 3 x faster than ZCM ???? Not true !
2 x faster is possible !

[SvenBent]

always compression time numbers.... im more into decompression times. how is zcm compared to ccm(x) ?

[Matt Mahoney]

20'th on LTCB. Nice job. http://mattmahoney.net/dc/text.html#1694

[Sebastian]

Nice job

I wonder what's the trick on fp.log? lpaq variants and your zcm compress it <400kb but even zpaq on max only manages 400kb.

[Skymmer]

CCMX = 3 x faster than ZCM ???? Not true !
2 x faster is possible !

Why not? Here is the little test performed on 3DMark06.dat file from 3DMark 2006 3d benchmark software. The file itself its an uncompressed container with different resources inside. Mostly textures in DDS format. Plus misc stuff like shaders, models, sounds in WAV\OGG formats, etc.
Benchmarked on i5-2500K @ 3700 MHz.

                               size          comp.time      memory.eat
                           =============     =========     =============
Original                   1 320 871 306        ---             ---
NZ 0.07 x64 -cc -m2g         379 483 484      1409.93      2 187 345 920
NZ 0.09 x64 -cc -m2g         382 553 817      1439.23      2 158 702 592
NZ 0.08 x64 -cc -m2g         382 697 890      1400.42      2 187 300 864
CCMx 1.30c 7                 418 449 265       402.40      1 367 261 184
CCM  1.30c 7                 420 388 544       345.70      1 367 265 280
ZCM c7                       425 177 147      1171.56      1 644 777 472
BiT 0.7 -p=5                 428 650 913      1117.70        682 151 936

Comparing to CCMx its 2.91 times slower, has 20.3% higher memory consumption and gives 6.4 MB worse result. Comparing to CCM it even 3.38 times slower.
Well, quite unlucky file for ZCM. Anyway, results in LTCB and SqueezChart are quite good, so... Nice work. Thanks, Nania!

always compression time numbers.... im more into decompression times. how is zcm compared to ccm(x) ?

ZCM is the CM type of compressor, so its symmetric. I mean decompression time is almost equal to compression one

[Matt Mahoney]

20,857,925 enwik8-7.ccmx, 107 sec
20,093,284 enwik8-c7.zcm, 330 sec

Tested on a 2.0 GHz Intel T3200, Vista SP2 32 bit.

[SvenBent]

ZCM is the CM type of compressor, so its symmetric. I mean decompression time is almost equal to compression one

*sigh* total useless

Nania stop working on this nonsens and start making some nice usable LZ based compression stuff


P.S.
yes it was sarcasme

[m^2]

In the Shelwein's table, CCMX was less than 2x faster, not 3x, I guess that's what Francesco wanted to say.

[Shelwien]

Either way, I actually expected that a better CM would appear, now that we know precisely how ccm works and all.
I even demonstrated one possible improvement by adding vecrc to ccm, and I can easily list a few more options.
But now, 2.5 years later, we still don't have anything better.

[Nania Francesco]

ZCM v.0.02 released

- added option 0 (require 20MB)
- improved better compression !
- more faster !

Comment:
Algorithm: Z-CM
.. Please ..kindly it does not compare it to you to CCM or CCMX ! Thanks !
link:
http://heartofcomp.altervista.org

[Shelwien]

http://nishi.dreamhosters.com/u/zcm02a.htm

at -1: 16.6% faster, 2.9% worse compression (vs v01)
at -7: 10.5% faster, 0.18% worse compression

Also m1x2 used for comparison instead of ccm - same compression if we'd compare
m1 -4 and zcm -1, but zcm is 2x slower.

Note that m1 is completely unrelated to ccm, and also open-source.

If you don't need objective benchmarks, please say so, as it takes time to run them.
Btw, I also think that your coder is promising - otherwise I'd just ignore it.

[Nania Francesco]

ZCM v.0.2 SFC benchnark

Option c0
A10.jpg 842468->826648
AcroRd32.exe 3870784->1216639
english.dic 4067439->570948
FlashMX.pdf 4526946->3671250
FP.LOG 20617071->388703
MSO97.DLL 3782416->1671436
ohs.doc 4168192->759259
rafale.bmp 4149414->791560
vcfiu.hlp 4121418->524038
world95.txt 2988578->519312
original=53134726 compressed=10939793 ENC=45.891998 sec. DEC=45.483002 sec.

Option c7
A10.jpg 842468->824511
AcroRd32.exe 3870784->1137017
english.dic 4067439->481353
FlashMX.pdf 4526946->3637043
FP.LOG 20617071->375936
MSO97.DLL 3782416->1535570
ohs.doc 4168192->734007
rafale.bmp 4149414->737553
vcfiu.hlp 4121418->493442
world95.txt 2988578->439967
original=53134726 compressed=10.396.399 ENC=68.375000 sec. DEC=67.971001 sec.

[Matt Mahoney]

Compression is slightly worse but faster. http://mattmahoney.net/dc/text.html#1694

[Nania Francesco]

Thanks Matt for testing ZCM v.0.02!
HI!
Best Regards!

[Nania Francesco]

ZCM v.0.03 released

- improved better compression !
- more faster !

link:
http://heartofcomp.altervista.org/ZCM/home.htm

[joerg]

Gratulation Nania!

in my first test
the compression is very good
the relation compression/time is very good

please add the "fast cm" - algorithms from rings 1.3 or rings 1.5c
as "very fast mode" to this program

and it will THE BEST NEW FILE COMPRESSOR in 2011 !!

best regards

[Nania Francesco]

Thanks joerg!
Test in SFC (Intel Core 2 duo E660 2.4GHZ)
OPTION c0
A10.jpg 842468->826527
AcroRd32.exe 3870784->1215497
english.dic 4067439->579219
FlashMX.pdf 4526946->3669778
FP.LOG 20617071->392526
MSO97.DLL 3782416->1669778
ohs.doc 4168192->758168
rafale.bmp 4149414->792896
vcfiu.hlp 4121418->526811
world95.txt 2988578->514261
original=53134726 compressed=10945461 ENC=44.139999 sec. DEC=43.796001 sec.

Option c7
A10.jpg 842468->824319
AcroRd32.exe 3870784->1134577
english.dic 4067439->472460
FlashMX.pdf 4526946->3635874
FP.LOG 20617071->379430
MSO97.DLL 3782416->1531287
ohs.doc 4168192->733240
rafale.bmp 4149414->735129
vcfiu.hlp 4121418->492559
world95.txt 2988578->435596
original=53134726 compressed=10.374.471 ENC=64.499001 sec. DEC=64.843002 sec.

[Nania Francesco]

ZCM v.0.04 released

- improved better compression (not in text files) !
- more faster (20-30 %)!

link:
http://heartofcomp.altervista.org/ZCM/home.htm

@Matt and @Shelwien please .. if possible ... test ZCM v.0.04! Thanks

[Biozynotiker]

Hi Nania,

the download link is still v0.03!
And please add any progress to the screen output, if not already done in V0.04.

[Shelwien]

http://nishi.dreamhosters.com/u/zcm04.htm

[Nania Francesco]

@ Shelwien
Thanks !! No comment ?

@ Biozynotiker
Please update the page !

[Shelwien]

> No comment ?

Good work. Compression got worse so its hard to compare, but if we'd match
levels with similar compression in v03 and v04, the latter seems to be 20-25% faster.
But in the same sense, m1 is still 30% faster. Please continue

> Please update the page !

He probably meant that zcm.exe still prints "ZCM v. 0.03 (C) 2011, Nania Francesco Antonio (Italy)"

[Matt Mahoney]

http://mattmahoney.net/dc/text.html#1693

v0.04 is faster but loses some compression. It says v0.03 (2011) when run but is a different program.

[Nania Francesco]

@ Shelwien : Thanks for comment's !
@ Matt: Thanks for test!

Excuse me for banner error !

Best regards !

[Nania Francesco]

ZCM version 0.11 released
- added data detection !
- better compression !
- more faster (5-15 %)!

link:
http://heartofcomp.altervista.org/ZCM/home.htm

@Matt and @Shelwien please ..

.. if possible ... test last ZCM version ! Thanks !

[Nania Francesco]

SFC test for example

[Pedofinder]

it can't compress directory?