Results 1 to 10 of 10

Thread: PAQ8j

  1. #1
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Bill Pettis has released PAQ8j.

    http://ilovemyking.googlepages.com/paqpage

    Bill claims that "paq8j -4 beats paq8i -7 in the DLL compression test".

    This information was originally posted by Bill Pettis in the maximumcompression.com guestbook.


    PAQ8j always crashes during compression on my AMD Sempron machine.

  2. #2
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    The problem appears to have been fixed.


    Quick test (PAQ8j -7)...

    a10.jpg > 699,694

    mso97.dll > 1,291,576

    ohs.doc > 548,683

    rafale.bmp > 635,634

    vcfiu.hlp > 407,681

    world95.txt > 362,250

  3. #3
    The Founder encode's Avatar
    Join Date
    May 2006
    Location
    Moscow, Russia
    Posts
    3,954
    Thanks
    359
    Thanked 332 Times in 131 Posts
    The performance is amazing!



    Also I have a good news about TC. Currently I theoretically motivate the entire algorithm and after some additional calculations I will continue improving this promising compressor. I divide this engine to two parts:

    <u>1. LZ or ROLZ part.</u>
    It can be slightly improved compared to current dev5 version. But also compression speed can be affected and at the same time decompression speed will be untouched or can be even faster, though.

    <u>2. CM or PAQ part. </u>
    The most expensive part. Currently I spend all my development time to research, test and evaluate the new variations. Also, currently I'm asking Malcolm Taylor for some advices with this part.

    Like I said, I already have the unreleased 'dev6' with SSE - a slower version, witch is stable on random data and have higher compresison on analog data such as WAV-files. But since eventually I want to add this engine to the PIM or other GUI archiver, I keep in mind - "haste makes a waste". So better I spend more time and get something ultra efficient and world's best.


  4. #4
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Also I have a good news about TC. Currently I theoretically motivate the entire algorithm and after some additional calculations I will continue improving this promising compressor. I divide this engine to two parts:

    1. LZ or ROLZ part.
    It can be slightly improved compared to current dev5 version. But also compression speed can be affected and at the same time decompression speed will be untouched or can be even faster, though.

    2. CM or PAQ part.
    The most expensive part. Currently I spend all my development time to research, test and evaluate the new variations. Also, currently I'm asking Malcolm Taylor for some advices with this part.

    Like I said, I already have the unreleased 'dev6' with SSE - a slower version, witch is stable on random data and have higher compresison on analog data such as WAV-files. But since eventually I want to add this engine to the PIM or other GUI archiver, I keep in mind - "haste makes a waste". So better I spend more time and get something ultra efficient and world's best.


    This is the best news for some time! Looks like TC will become the very best compressor available.

  5. #5
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Another slightly longer test...

    Most of the settings used in this test were copied from the PAQ8i results at maximumcompression.com.

    A10.jpg (- > 699,687

    AcroRd32.exe (-7) > 923,247

    english.dic (-6) > 390,472

    FlashMX.pdf (-6) > 3,568,866

    FP.LOG (-7) > 280,949

    MSO97.DLL (-7) > 1,291,576

    ohs.doc (-6) > 548,640

    rafale.bmp (-5) > 634,921

    vcfiu.hlp (-7) > 407,681

    world95.txt (-7) > 362,250

  6. #6
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    I have tested PAQ8j with several files of my own and the results have been quite impressive. It now appears to be 100% stable.

    Thanks Bill!

  7. #7
    The Founder encode's Avatar
    Join Date
    May 2006
    Location
    Moscow, Russia
    Posts
    3,954
    Thanks
    359
    Thanked 332 Times in 131 Posts
    This Bill is really funny guy - I saw his videos...


  8. #8
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    I decided to perform a final quick test to see how PAQ8j performs on ENWIK8.

    PAQ8j (-7) > 18,208,284

    Good perfomance considering there are no dictionaries used.

  9. #9
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Serge Osnach has released PAQ8ja.

    http://www.cs.fit.edu/~mmahoney/compression/

    Quick test...

    Most of the settings used in this test were copied from the PAQ8i results at maximumcompression.com.

    A10.jpg (-7) > 699,697

    AcroRd32.exe (-7) > 918,883

    english.dic (-6) > 389,909

    FlashMX.pdf (-6) > 3,567,807

    FP.LOG (-7) > 280,273

    MSO97.DLL (-7) > 1,287,295

    ohs.doc (-6) > 548,210

    rafale.bmp (-5) > 634,927

    vcfiu.hlp (-7) > 405,312

    world95.txt (-7) > 361,339

    Total = 9,093,652 bytes

  10. #10
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    Serge Osnach has release PAQ8jb.

    PAQ8jb (Nov. 21, 2006) adds a distance model, using context of distance back to an anchor character (x00, space, newline, xff) combined with previous characters. Win32 compiled with VS2003.

    http://www.cs.fit.edu/~mmahoney/compression/

Posting Permissions

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