Squeeze Chart 2007

**Stephan Busch** · 2nd May 2007, 19:17

I am very curious about TTA in FreeARC

Will it also detect waveforms within the sound modules (.s3m, .mod..)?

**encode** · 2nd May 2007, 19:29

Originally Posted by Bulat Ziganshin

i think it is just too slow for my users

701774 -> 842468 (6.6640 bpc) in 14.45 sec (58.290 KB/sec)

58 KB/sec is indeed too slow...

**Bulat Ziganshin** · 2nd May 2007, 19:36

i just conducted fast test (you can do the same - http://www.haskell.org/bz/tta.rar includes both mmdet and tta compressor). results are: s3m/mod diles are easily detected as MM ones but compression is too bad. probably, by splitting them into MM part and the rest we can got better results, but just now it's menaningless to use TTA on them

current freearc architecture doesn't allow to compress parts of file with different algorithms, i plan to make such incompatible change in archives' structure a few months later

**Black_Fox1** · 2nd May 2007, 19:56

Originally Posted by Stephan Busch

Do you think I should also move the architecture columns to the right?

I think architecture could be divided to windows/linux only (is vista different from xp?) and indicated by some windings letter (like smilie meaning open source)

**Bulat Ziganshin** · 2nd May 2007, 20:56

knowing that compressor is 64bit or supports multi-threading is also important these days

**Black_Fox1** · 2nd May 2007, 21:00

bit depth and core number have its own separate columns, I mean that writing "Vista" is just wasting pixels...

**Bulat Ziganshin** · 2nd May 2007, 21:07

afaik, some gui archivers support vista-specific features. just look at power archiver site

**Black_Fox1** · 2nd May 2007, 21:19

these changes are only gui oriented, that could be written somewhere into "notes" section... what could change testing under vista or under xp, when CCM comes to question?

**Bulat Ziganshin** · 2nd May 2007, 21:30

it seems that this column is rather meanuingless, it contains vista for some newer cmdline archivers (like tornado), but not for other ones (like quad). ccm 1.20 is xp, while 1.21 is vista

may be it is a testing platform?

**Black_Fox1** · 2nd May 2007, 21:34

It indeed is! But what does this mean for common user? Nothing

EDIT: When still showing what is WIN32 and what is Linux... and when not counting the GUI improvements in Vista...

**Bulat Ziganshin** · 2nd May 2007, 21:41

other notes: charc is from russia, ha probably from japan (harry hirvola looks like japan name), exp is my program but is is just shell around bzip2 code. ain is from St-Petersburg, which was never english territory

tornado file extension is .tor. last column seems badly broken

ppmonstr sources aren't available (otherwise it was be included in my program

)

it will be also interesting to know cache size of your system. 1mb per core, i guess?

**encode** · 2nd May 2007, 21:43

Originally Posted by Bulat Ziganshin

harry hirvola looks like japan name

Nope!

Looks like from Finland!

(Горячий Финский парень!

)

**Black_Fox1** · 2nd May 2007, 22:47

Originally Posted by encode

Горячий Финский парень!

Oh my god

But youre right, found his mail, which is (was some time ago) hosted at some Finnish site

**Stephan Busch** · 2nd May 2007, 22:56

updated main.html is online now.
It contains results for FreeARC using -m7 switch

**Bulat Ziganshin** · 2nd May 2007, 23:28

thank you! can you please also delete results of previous testing? if you are interested in testing faster modes then -m4 (default) will be a good variant

**Stephan Busch** · 3rd May 2007, 00:25

thanks for pointing me some errors. I will fix the errors asap.

Each of my cpu's has 1MB cache which can be easily read over
at amd.com. The architecture column is just to inform readers about
bit version, used processors and the OS I ran it on. I can change that XP/Vista info by writing Win or I caould combine bit depth and os into one column by writing "win32", "win64". But I wouldn't replace
them by wingding symbols - there aren't some acceptable ones.
And I decided to move achitecture to the right, but next update has to wait some days - at least a week because I am busy with real life.
besides that the small testset benchmarks are faster and more information-packed. do we really need squeezechart anymore?

Cheers

**Black_Fox1** · 3rd May 2007, 00:39

Originally Posted by Stephan Busch

besides that the small testset benchmarks are faster and more information-packed. do we really need squeezechart anymore?

Yes we do!

Smaller benchmarks dont wield so much information as squeezechart, they cant be more information-packed than SC...

**Bulat Ziganshin** · 3rd May 2007, 08:56

Originally Posted by Stephan Busch

besides that the small testset benchmarks are faster and more information-packed

where at your site are such benchmarks? or you mean that you want to do it instead of current tests?

for me, testsets should be large enough in order to show freearcs advantages

**encode** · 3rd May 2007, 11:56

Originally Posted by Bulat Ziganshin

for me, testsets should be large enough in order to show freearcs advantages

**LovePimple** · 4th May 2007, 15:34

Originally Posted by Stephan Busch

besides that the small testset benchmarks are faster and more information-packed. do we really need squeezechart anymore?

I think so!

**Bulat Ziganshin** · 14th May 2007, 16:02

how i can decode http://www.kellyindustries.com/downloads/48khz_ac3 .zip ?

can you provide mirror of http://www.q-tuner.com/Aspie96.wav (which isn't available)?

**Stephan Busch** · 14th May 2007, 23:09

use nero waveeditor for that; version 7 comes with support for
multiple channel and high resolution waveforms (so called extented waveforms). Aspie96 is not available anymore, but I could send you the wavpack'ed variant by email if you want.
Use wavewizard 0.54b afterwards to check wav header information.

**Bulat Ziganshin** · 14th May 2007, 23:32

Originally Posted by Stephan Busch

Aspie96 is not available anymore, but I could send you the wavpacked

and how large is it?

**Stephan Busch** · 15th May 2007, 00:28

round about 31 MB; I may split it into 5 MB parts because most providers have attachment size limits.

It's really cool - you're the first one that implements high res. and multiple channel audio - only one commerical competitor (WinZIP11).
I hope some more authors follow this important step, because there are many audiophiles and music studio composers that use 24 or 32 bit at up to 384 khz. Cinemas use up to 16 channels, so 65000 is way too seldom.

**Black_Fox1** · 15th May 2007, 00:43

Originally Posted by Stephan Busch

WinZIP11

Well... AFAIK it uses WavPack anyway

**Vacon** · 15th May 2007, 01:00

Hello everyone,

Originally Posted by Black_Fox

Well... AFAIK it uses WavPack anyway

that seems to be true.
IIRC there was athread on maximumcompression that lead to a forum where a screenshot from a beta-version was shown.
In its title one could read ... WavPack...

So you see how leaders are made...

Best regards!

**Black_Fox1** · 15th May 2007, 01:09

hydrogenaudio.org

**Stephan Busch** · 15th May 2007, 01:31

Of course it is wavpack, but with David Bryant's allowance.
I know the person that suggested Corel to use Wavpack.

**Bulat Ziganshin** · 15th May 2007, 10:45

Originally Posted by Bulat Ziganshin

and how large is it?

it was dumbest question on the forum - all sizes are in your table

too large for me, instead i will make such files using wave editors (yes, inet is very expensive and soft is very cheap here

)

Originally Posted by Stephan Busch

youre the first one that implements high res. and multiple channel audio

i just use TTA library and afaik it doesnt have any problem with number of channels. sampling rate also doesnt have any effect on compression with only one exception - TTA codec makes one block for each 1 second of sound, and 1 second may include too many samples. i definitely know that TTA library doesnt work with 10^6 samples per block but i dont know lower bound. i suggested authors to somethat limit nu,ber of samples, but they dont have too much time now. in freearc each block contains just 128k samples, independent on sampling rate, and this fixes any problems (at least in my own tests

)

**Bulat Ziganshin** · 15th May 2007, 10:47

Originally Posted by Stephan Busch

there are many audiophiles and music studio composers that use 24 or 32 bit at up to 384 khz. Cinemas use up to 16 channels, so 65000 is way too seldom.

btw, can you help me by answering question at http://www.encode.su/forums/index.php?action=vthre ad&forum=1&topic=436 ?

Thread: Squeeze Chart 2007

Thread Tools

Search Thread

Display

Similar Threads

Squeeze Chart

Squeeze Chart 2006 - 02 April/13 May

Posting Permissions