Results 1 to 24 of 24

Thread: PAQ8HP8

  1. #1
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    PAQ8HP8 is available from Alexander Rhatushnyak's site.

    P.S. It's either extremely optimized for enwik tests or I'm doing something wrong. 776,660 bytes for TXT1 when using -6 is just unbelievable...

  2. #2
    Moderator

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

  3. #3
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    If it were possible to incorporate the text compression technology from PAQ8HPx into PAQ8Jx, the compression could be truly amazing.

    Is this possible?

  4. #4
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    It should be possible, because it's just one of compressing models (which are removed from HPx versions)

    EDIT: pht compressed to 2,933,118 bytes...

  5. #5
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    PAQ8HP8 was updated today for some reason.


    Original version: 11th Jan 2007, 152,668 bytes

    Updated version: 17th Jan 2007, 152,692 bytes


    EDIT: The updated version of PAQ8HP8(-7) compresses your PSD test file to 888,681 bytes.

  6. #6
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    Quote Originally Posted by LovePimple
    PAQ8HP8 was updated today for some reason.
    Aaah, I thought I was faster

    Extremely looking forward to testing it -> tried compressing enwik8 with original version with very dissapointing results, the updated one should be powerful

  7. #7
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    The updated version of PAQ8HP8(-7) compresses World95.txt to 330,625 bytes.

  8. #8
    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
    Extremely looking forward to testing it
    What was the resulting compressed size of ENWIK8?

  9. #9
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    Quote Originally Posted by LovePimple
    What was the resulting compressed size of ENWIK8?
    I am currently testing its performance on my testset - updated version gets few hundreds bytes closer to Durilca 0.5 Hutters result on TXT1 (English book) - now loses by 275 bytes... one more version and PAQ8HPx will gain lead at this file
    PAQ8HP8 compared to its predecessors - HP7(a) and HP6 - gets worse results at all my testfiles, except for BMP and TXT1

    EDIT: PAQHP8 - 11,167,578 bytes.

    I will post -6 result of ENWIK8 here at 6:45 CET tomorrow.

  10. #10
    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
    EDIT: PAQHP8 - 11,167,578 bytes.
    Is this at -7?

    Quote Originally Posted by LovePimple
    I will post -6 result of ENWIK8 here at 6:45 CET tomorrow.
    Thanks!

  11. #11
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    Quote Originally Posted by LovePimple
    Is this at -7?
    No, only at -6.

  12. #12
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    I only asked because I wanted to compare it with your HP7(-7) results.

  13. #13
    Moderator

    Join Date
    May 2008
    Location
    Tristan da Cunha
    Posts
    2,034
    Thanks
    0
    Thanked 4 Times in 4 Posts
    How many people do you know personally that have 2GB or more of RAM?

  14. #14
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    Quote Originally Posted by LovePimple
    How many people do you know personally that have 2GB or more of RAM?
    I cannot remind me of anyone People around usually have 1GB or 512MB...

  15. #15
    Moderator

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

    I was hoping to find someone that could test Emilcont 0.2(-9) on your PSD file.

    I think it would require at least 3GB of RAM and the /3GB Startup Switch would need to be added to XP's (32 bit ver) boot.ini file.

  16. #16
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    I think in Emilcont 0.3 or some other Emilcont there are approximate requirements and it's around 3,2GB

    The school is giving me a headache in last weeks, but at least today it's positive that I stayed awake so long that PAQ8HP8 enwik compression finished However, I will have to test HP7 -6 because all results are -7/-8 only...

  17. #17
    Member
    Join Date
    Dec 2006
    Posts
    611
    Thanks
    0
    Thanked 1 Time in 1 Post
    enwik8-6.paq8hp7 -16,833,465
    enwik8-6.paq8hp8 -16,776,152

  18. #18
    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
    I think in Emilcont 0.3 or some other Emilcont there are approximate requirements and its around 3,2GB
    I this is true then it must be impossible to use Emilcont 0.2 with setting -9 under 32 bit Windows XP.

    By default Windows XP (32 bit) allocates a maximum of 2GB (3GB with the /3GB startup switch) address space for apps.

    In modern operating systems, including Windows, application programs and many system processes always reference memory using virtual memory addresses which are automatically translated to real (RAM) addresses by the hardware. Only core parts of the operating system kernel bypass this address translation and use real memory addresses directly. All processes (e.g. application executables) running under 32 bit Windows gets virtual memory addresses (a Virtual Address Space) going from 0 to 4,294,967,295 (2*32-1 = 4 GB), no matter how much RAM is actually installed on the computer. In the default Windows OS configuration, 2 GB of this virtual address space are designated for each process private use and the other 2 GB are shared between all processes and the operating system. RAM is a limited resource, whereas virtual memory is, for most practical purposes, unlimited. There can be a large number of processes each with its own 2 GB of private virtual address space. When the memory in use by all the existing processes exceeds the amount of RAM available, the operating system will move pages (4 KB pieces) of one or more virtual address spaces to the computers hard disk, thus freeing that RAM frame for other uses. In Windows systems, these "paged out" pages are stored in one or more files called pagefile.sys in the root of a partition. Virtual Memory is always in use, even when the memory required by all running processes does not exceed the amount of RAM installed on the system.
    Microsoft Help and Support


    Limitations exist for 32 bit apps even with 64 bit Windows XP.

    Windows XP Professional x64 Edition

  19. #19
    Moderator

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

    The current release of PAQ8HP8(-7) compresses ENWIK8 to 16,528,353 bytes.

  20. #20
    The Founder encode's Avatar
    Join Date
    May 2006
    Location
    Moscow, Russia
    Posts
    4,000
    Thanks
    387
    Thanked 365 Times in 145 Posts
    Matt explored a bug in ASM code... All PAQ compressors since PAQ7 have such bug!

  21. #21

  22. #22
    Guest
    > Matt explored a bug in ASM code... All PAQ compressors since PAQ7 have such bug!

    No, there is no problem with 32 bit versions. Only the 64 bit ASM code recently written by Matthew Fite didn't work. It was used to make Win64 paq8f and paq8jc executables at http://ilovemyking.googlepages.com/paqpage which probably do not work. I fixed the 64 bit ASM code which is now available for 64 bit Linux versions of paq8f and paq8jd at http://cs.fit.edu/~mmahoney/compression/#paq8
    The new assembler code should work for all paq versions but I have tested it only in Linux because I don't have a 64 bit version of Windows. To compile it, replace paq7asm.asm with paq7asm-x86_64.asm with no changes to the .cpp files. You need yasm to assemble the asm code.

    Also, I have tested paq8hp8
    http://cs.fit.edu/~mmahoney/compression/text.html
    There was an older version that compressed poorly (about 22 MB for enwik so I did not post any results then. That was fixed yesterday.

  23. #23
    Moderator

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

    Summary:

    -6 > 16,776,152 bytes

    -7 > 16,528,353 bytes

    -8 > 16,372,960 bytes

  24. #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
    I think in Emilcont 0.3 or some other Emilcont there are approximate requirements and its around 3,2GB
    PAQ8A also has the same problem as Emilcont 0.2. PAQ8A at -9 setting needs ~ 4GB memory.

    It is impossible to use Emilcont 0.2 or PAQ8A with the -9 setting under 32 bit Windows.

Posting Permissions

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