Page 1 of 7 123 ... LastLast
Results 1 to 30 of 185

Thread: Black Fox benchmark updated

  1. #1
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    Table 3 is now available.
    http://blackfox.wz.cz/pcman/benchmark.htm

    I will appreciate any comments regarding layout and contents
    Is there any compressor you would like to see here (in my to-do list there currently are some versions of upack, uda and aspack)?

  2. #2
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Thanks!

    I'm surprised that OpenDark-A outperforms Dark versions 0.41 - 0.51 and PX in your tests.

  3. #3
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    It's mainly because OpenDark handles PSD file much better...

  4. #4
    Tester

    Join Date
    May 2008
    Location
    St-Petersburg, Russia
    Posts
    182
    Thanks
    3
    Thanked 0 Times in 0 Posts
    It will be better if you add in future decompression time (or speed like compression speed) in your table...

  5. #5
    Tester

    Join Date
    May 2008
    Location
    St-Petersburg, Russia
    Posts
    182
    Thanks
    3
    Thanked 0 Times in 0 Posts
    I am glad to see that both Quad 1.02a and Dark 0.51 improved compression ratio
    But why you don't test WinACE 2.65 and WinZIP 11? (one of the most popular archivers in the world)

  6. #6
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    Quote Originally Posted by Squxe
    It will be better if you add in future decompression time
    Im trying to tweak the table to put in decompression speed AND to squeeze the table into my webpage design when done, I will add them surely.
    Quote Originally Posted by Squxe
    But why you dont test WinACE 2.65 and WinZIP 11?
    Well WinAce didnt work when recently tried... and about the newest WinZIP I just totally forgot I will add them during next update.

  7. #7
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    WinAce 2.65 and WinZIP 11 build 7313 have been tested...

    And kvark, here is the PSD file in 7-zip archive... extract and rename to .psd
    Here is a down-sized preview.

  8. Thanks:

    encode (8th May 2016)

  9. #8
    Programmer kvark's Avatar
    Join Date
    Aug 2006
    Location
    Toronto, Canada
    Posts
    74
    Thanks
    1
    Thanked 1 Time in 1 Post
    Thanks a lot! Maybe this situation will open new improving possibilites...

    BTW, You are that RSK on compression.ru forum?

  10. #9
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    Quote Originally Posted by kvark
    You are that RSK on compression.ru forum?
    No Im not, Im not even a registered member and was there today for the first time

  11. #10
    Programmer kvark's Avatar
    Join Date
    Aug 2006
    Location
    Toronto, Canada
    Posts
    74
    Thanks
    1
    Thanked 1 Time in 1 Post
    Please explain the blocksize you used in compression with dark (-b138m and -b117m). What size was used with OpenDark?

  12. #11
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    The blocksizes are chosen so that my ~700MB of free RAM is enough, for earlier versions it was 6N (~117MB), for newer 5N (~138MB). Some time ago I had 2GB ram instead of 1GB which I have now, so I could use more. OpenDark was the only one tested earlier - with N=300MB.

  13. #12
    Programmer kvark's Avatar
    Join Date
    Aug 2006
    Location
    Toronto, Canada
    Posts
    74
    Thanks
    1
    Thanked 1 Time in 1 Post
    But as far as I see you total uncompressed size is 30Mb. Maximum file size is lower than 7Mb so you have no need to use more than 7Mb block for BWT compressors.

    It's a pity this will not affect compression ratio...

    BTW: I can't decompress pht.7z: unsufficient memory

    And a question: did you try to unpack OpenDark's file?

  14. #13
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Quote Originally Posted by kvark
    BTW: I cant decompress pht.7z: unsufficient memory
    You should be able to decompress this version.

    Download PSD File

    The archive size is slightly larger but it needs only 34 MB to decompress.

  15. #14
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    Umm... Reuploaded easier decompressible version, too.

    Yup, more memory than 7MB doesn't affect compression ratio in this case, but if it did, I wouldn't be able to unpack OpenDark's archives now I tried decompressing and had no problems.

  16. #15
    Programmer kvark's Avatar
    Join Date
    Aug 2006
    Location
    Toronto, Canada
    Posts
    74
    Thanks
    1
    Thanked 1 Time in 1 Post
    Correct decompression confirmed.
    I just couldnt believe this diference in compression is possible...
    New possibilities, anyway
    Quote Originally Posted by LovePimple
    You should be able to decompress this version.
    Download PSD File
    The archive size is slightly larger but it needs only 34 MB to decompress.
    Thanks, LovePimple!

  17. #16
    The Founder encode's Avatar
    Join Date
    May 2006
    Location
    Moscow, Russia
    Posts
    3,985
    Thanks
    377
    Thanked 353 Times in 141 Posts
    Additional mirror:
    pht.psd.quad (1,172 KB)

    Quote Originally Posted by kvark
    I just couldnt believe this diference in compression is possible...
    Probably, you can combine the power sides of both making Dark even more powerful!


  18. #17
    Programmer kvark's Avatar
    Join Date
    Aug 2006
    Location
    Toronto, Canada
    Posts
    74
    Thanks
    1
    Thanked 1 Time in 1 Post
    To tell the truth, OpenDark has very simplyfied contexts. It's closer to ybs then to dark... So this file may help understanding ybs compression model.

  19. #18
    Guest
    I'm not sure the contexts in ybs are so simple, though

  20. #19
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Quote Originally Posted by Black_Fox
    And kvark, here is the PSD file in 7-zip archive... extract and rename to .psd
    The Intel compile of Emilcont 0.2 (-7) compresses this file to 915480 bytes in less than 8 minutes on my machine. This almost rivals the compression results from PAQ8H (915384) but *MUCH* faster.

    It would be interesting to see just how small Emilcont 0.2 can compress this file. Unfortunately I dont have enough free memory to try settings higher than -7 on any file.

    Could someone with more RAM please try the same test at settings -8 and -9 and post the results here.

    BTW: Does anyone know exactly how much free memory Emilcont 0.2 requires for each of the settings from -0 to -9?

    Thanks!

  21. #20
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    Quote Originally Posted by LovePimple
    Emilcont 0.2 (-7) compresses this file to 915480 bytes
    I have this version in my benchmark as EMILCONT 0.2fast (30kB/s = five times faster than PAQ... it is interesting, that it provided with psd file produced result of 915,471 bytes. Are you able to decompress file you made? I tried decompressing -7 archive of psd i made some weeks ago and it failed at 0,5%, then tried compressing the psd again now and then decompressing it, but again - decompression failed very early. What CPU do you have?

  22. #21
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Emilcont 0.3a (-6) compresses it to 913,618 bytes in 27 minutes 49 seconds.

    It would be interesting to try Emilcont 0.3a on this file at -7 to -9 if anyone has enough RAM fitted.

    Can anyone help with this?

    Quote Originally Posted by Black_Fox
    I have this version in my benchmark as EMILCONT 0.2fast (30kB/s = five times faster than PAQ... it is interesting, that it provided with psd file produced result of 915,471 bytes. Are you able to decompress file you made? I tried decompressing -7 archive of psd i made some weeks ago and it failed at 0,5%, then tried compressing the psd again now and then decompressing it, but again - decompression failed very early.
    Will try a decompression test and post my results later.

    Quote Originally Posted by Black_Fox
    What CPU do you have?
    AMD Sempron 2400+ (32 bit).

  23. #22
    The Founder encode's Avatar
    Join Date
    May 2006
    Location
    Moscow, Russia
    Posts
    3,985
    Thanks
    377
    Thanked 353 Times in 141 Posts
    PIMPLE is able to compress this file to:

    Maximum: 992,361 bytes
    Extreme: 990,044 bytes

    Keep in mind:
    + With additional tuning it is possible to achieve even higher compression
    + PIMPLE compresses this file FAST (in 12 seconds)!


  24. #23
    The Founder encode's Avatar
    Join Date
    May 2006
    Location
    Moscow, Russia
    Posts
    3,985
    Thanks
    377
    Thanked 353 Times in 141 Posts
    Quote Originally Posted by LovePimple
    It would be interesting to see just how small Emilcont 0.2 can compress this file. Unfortunately I dont have enough free memory to try settings higher than -7 on any file.
    Same thing here - with 1 GB of RAM emilcont throws "Out of memory"!

  25. #24
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Quote Originally Posted by Black_Fox
    Are you able to decompress file you made?
    Decompression was successful. Original and decompressed files exactly the same:

    Extracting archive Emilcont -7 testec02-7 ...
    6755163 blackfox.psd: extracted


    Initial Size (Bytes) 6755163
    Compressed Size (Bytes) 915480


    File 1:
    ************************************************** ************************************************
    Filename: Original.psd
    8.3 Path: E:PAQDEC~1Original.psd
    Full path: E:PAQdecompOriginal.psd
    Version: n/a
    File size: 6755163 bytes
    CRC32: 3313987510

    File 2:
    ************************************************** ************************************************
    Filename: Decompressed.psd
    8.3 Path: E:PAQDEC~1DECOMP~1.PSD
    Full path: E:PAQdecompDecompressed.psd
    Version: n/a
    File size: 6755163 bytes
    CRC32: 3313987510


    Comparison status:
    ************************************************** ************************************************
    Filename:
    No match

    Extension:
    100% match

    Size:
    100% match

    CRC32:
    100% match

  26. #25
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Quote Originally Posted by encode
    Same thing here - with 1 GB of RAM emilcont throws "Out of memory"!
    Exactly the same here. I wonder just how much RAM is required?

  27. #26
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Emilcont 0.3 decompression also confirmed. Original and decompressed files exactly the same:

    Extracting archive Emilcont -6 testec03-6 ...
    6755163 blackfox.psd: extracted


    Initial Size (Bytes) 6755163
    Compressed Size (Bytes) 913618

    File 1:
    ************************************************** ************************************************
    Filename: Original.psd
    8.3 Path: E:PAQDEC~1Original.psd
    Full path: E:PAQdecompOriginal.psd
    Version: n/a
    File size: 6755163 bytes
    CRC32: 3313987510

    File 2:
    ************************************************** ************************************************
    Filename: Decompressed.psd
    8.3 Path: E:PAQDEC~1DECOMP~1.PSD
    Full path: E:PAQdecompDecompressed.psd
    Version: n/a
    File size: 6755163 bytes
    CRC32: 3313987510


    Comparison status:
    ************************************************** ************************************************
    Filename:
    No match

    Extension:
    100% match

    Size:
    100% match

    CRC32:
    100% match

  28. #27
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Quote Originally Posted by Black_Fox
    Some time ago I had 2GB ram instead of 1GB which I have now, so I could use more.
    How much RAM did you have when you tested DURILCA 0.5 at -m1800?

    When did you cut your RAM to 1GB?

    What CPU do you have?

  29. #28
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    I had been using two 1GB modules for about a month, table 1 was made almost completely with 2GB RAM (I shut down almost everything in Windows to get 1800MB free RAM that time. ). One of these modules was used for testing and then sold few days before X-mas, hence the RAM cutting.
    I have dual-core Athlon 64 X2 3800+ (64-bit), but there are not many compressors able to take advantage of more than one core under 32-bit Windows.

    Could you please upload that psd you compressed with emilcont0.2? I just want to try if I'm able to decompress at least this one

  30. #29
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Quote Originally Posted by Black_Fox
    Could you please upload that psd you compressed with emilcont0.2? I just want to try if Im able to decompress at least this one
    No Problem!

    Download Emilcont 0.2 test file

  31. #30
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    Thanks I'll try decompressing it in a few hours' time...
    EDIT: Interesting. Now it correctly compresses and decompresses anything.


    Compression benchmark has been updated to table 4.

Page 1 of 7 123 ... LastLast

Similar Threads

  1. Pitch-Shifted Michael Jackson - Black Or White
    By encode in forum The Off-Topic Lounge
    Replies: 7
    Last Post: 12th May 2009, 01:27
  2. Updated PeaZip project: 1.5
    By giorgiotani in forum Forum Archive
    Replies: 8
    Last Post: 21st March 2007, 15:02
  3. Bee archiver updated!
    By LovePimple in forum Forum Archive
    Replies: 7
    Last Post: 18th February 2007, 15:05

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •