ccm - latest version & how does it compare?

[avitar]

Can someone tell me where I can find the latest (best?) version - believe it is 1.0.6 with some improvements & fixes vs v1.0.3, which I have. The original links in Christian's email seem broken.

Also, how does it compare with latest 7zip, freearc etc.

John

[Bulat Ziganshin]

lovepimple.110mb.com - latest was best. it has better compression than lzma, but slower and only 64 mb dictionary. it's hard to compare with full 7-zip or freearc since it depends on many factors. overall, 7-zip and freearc is better due to m/t support and lot of extra codecs in freearc

[avitar]

1) Only found 1.03 on lovepimple so links to later versions eg 1.0.6 appreciated in case they'd be better as per the discussion at the time.

2) Just a quick check on my 2G laptop, with no extra switches on a 11Mbyte text file,
ccm 1.03 was 1 Mbyte, 7s
latest alpha freearc & latest alpha 7z were 1.5Mbytes, about 12s

Is this what's expected?

So just for this not v accurate & specific test still looks ok, but of course I agree re support & lots of switches arc& 7z still better choice, maybe.

John

[schnaader]

The latest version I used was 1.30a and included CCM, CCMX, a readme and even Linux versions. I attached it to this post so you can have a look at it.

EDIT: This link from lovepimple.110mb.com is 1.30c, so it's newer, but I'll leave the other one for the included Linux versions.

[Bulat Ziganshin]

avitar, most files compressed are binary ones, so i said abouth them. for text files, ppmd may be better. there are also lot of other conditions on which compression ratio depends, there is decompression speed which you don't measured

[avitar]

Still looking for v1.0.5/6 - quite a few people from this site seemed to have them at one time from Christian, Werner, Nania, founder, black fox, Stephan so I'm hoping someone will make them available.

Bulat- agreed, ccm isn't really a developed program, I only spent < 10 mins doing the tests, just to see where ccm fitted wrt the other compressors. The decompress times were between 7-10s.

John

[Shelwien]

http://encode.su/threads/280-M1-benc...ull=1#post5096

[avitar]

Shelwien, thanks - whew, I only thought the versions went as far as 1.0.6 or so because I thought Christian gave up after that. Can you point me at any emails etc describing how the later versions are different/better?

Do you (or anyone) have a short overview of the versions'

John

[Shelwien]

There's my benchmark further in that thread.

[avitar]

I've looked at those... but some guidance re Christians ideas would be good - it is very difficult from the benchmarks to decide what's the best version to use.

Any remembered thoughts (or old emails) from those who were in the discussion & testing would be useful. Of course normally one would use the latest version of any app... John

[Shelwien]

> I've looked at those... but some guidance re Christians ideas would be good -

So what do you want to do? Compress something with the best ccm version?
Or write a better coder of your own?

As to ideas, there's nothing special, its the density of ideas that matters.
Basically you can take that - http://encode.su/threads/1153-Simple...angecoder-demo ,
compute hash functions of a few contexts (like o1-o5) and look up counters
(fsm counters from paq would do) in a hashtable. Then mix the predictions according to your taste.
It should be enough to compress better than ccm at least in some cases.
There're also: match model, sparse contexts, text detection, preprocessing.

> it is very difficult from the benchmarks to decide what's the best version to use.

The last one. Also ccmx has better compression but worse overall metric.

And I meant that - http://ctxmodel.net/files/ccm_sfc.htm
There you can see compression/decompression times and compressed size - total and per file.
There're also values of "distribution metric" - time required to upload the archive to 10 users;
ie version/mode with the smallest metric is the best.

[avitar]

Thanks for the info to all.

btw, I can't make multiple files or wildcards work- I'm guessing that Christian didn't implement this, ie only works for a single file.

Can anyone confirm? or tell me syntax for multiple files encode/decode?

TIA
John

[Shelwien]

I'd suggest making an archive first with 7z a -mx=0 -mhc=off archive path
Or rar -m0. Or tar.
Anyway, most experimental compressors don't support multiple files, there're archivers for that.

[joerg]

@avitar
you asked how ccm works
---
here is an old link, where christian martelock wrote something about this
---
http://encode.su/threads/738-CCM-file-compressor
---
and here is are two older programs,
where trying to use ccm for archv-purposes:
---
http://sourceforge.net/projects/coffeearc/
---
download:
---
http://sourceforge.net/projects/coff...b.zip/download
---
and the other program download:
---
http://lovepimple.110mb.com/myfreeware/tarccm_v1.1.zip
---
@avitar:
if someone other know how this ccm works - it should be Shelwien
- if i remember good - years ago he does a reengineering of the ccm 1.30

best regards