7zip need better ratio in PA



  • I tried with many options in 7-Zip to improve my compression ratio and I think that it will be good if PA Developers add this options to PA.
    And noone archiver test crush new PA compression ratio in any tests.

    I think that this formula should be use in Ultra mode
    or You need add this formula in Extreme Mode 😉

    And this formula for Parameters bring the best compression ratioand about increase PA ratio about 1-4%, and for me is big diffrence

    [b]x=9 s=on f=on hc=on hcf=on mt=on a=1 mf=bt4 fb=273 mc=1000000000 lc=0 lp=0 pb=0[/b]
    

    and some explanation:

    x=9
    (level of compression)
    options: x=[0 | 1 | 5 | 7 | 9] ,
    higher number=better ratio

    s=on
    (solid mode)
    options: s=[off | on]
    on gives better ratio

    f=on
    (filters for executable files: dll, exe, ocx, sfx, sys)
    options: f=[off | on]
    on gives better ratio

    hc=on
    (archive header compressing)
    options: hc=[off | on]
    on gives better ratio

    hcf=on
    (full archive header compressing)
    options: hcf=[off | on]
    on gives better ratio

    mt=on
    (multithread mode)
    options: mt=[off | on | {N}]
    on gives faster processing time

    **a=1
    **(compressing mode)
    options: a=[0|1]
    higher number=better ratio

    mf=bt4
    (Match Finder for LZMA)
    options: mf={MF_ID}
    bt2 = d9.5 + 4 MB = Binary Tree with 2 bytes hashing.
    bt3 = d
    11.5 + 4 MB = Binary Tree with 3 bytes hashing.
    bt4 = d11.5 + 4 MB = Binary Tree with 4 bytes hashing.
    hc4 = d
    7.5 + 4 MB = Hash Chain with 4 bytes hashing.
    the bt=4 provide the better compression ratio

    fb=273
    (fast bytes for LZMA)
    options: fb={N} where N=<5,273>
    higher number=better ratio

    mc=1000000000
    (cycles (passes) for match finder)
    options: mc={N} where N=<0,1000000000>
    higher number=better ratio

    lc=0
    (literal context bits (high bits of previous literal))
    options: lc={N} where N=<0,8>**
    lower number**=better ratio

    **
    lp=0**
    (literal pos bits (low bits of current position for literals))
    options: lp={N} where N=<0,4>
    lower number=better ratio

    pb=0
    (pos bits (low bits of current position))
    options: pb={N} where N=<0,4>
    lower number=better ratio



  • noone intrested add Extreme mode to 7zip compression to improve ratio 😞



  • I must admit I use 7zip for 90% of all my compression at work due to using Spreadsheets, Databases and word documents.

    If there is an improvement to the compression ratio then its worth looking into.

    However, it’s down to the PA Crew and the time it would take to develop and maintain for on going versions of 7Zip and PA.


  • conexware

    More important - what is speed difference? 50% slower? 🙂



  • @spwolf:

    More important - what is speed difference? 50% slower? 🙂

    Nope :p, but too my surprise it was faster than normal settings in 7-Zip

    And some test on about 350 files (doc, wav, mpeg etc.) in 1 folder:

    uncompressed folder
    335 MB (352.235.162 bytes)

    compressed in PA2007b9 (soild, Ultra mode)
    224 MB (235.877.605 bytes)
    10min 58 sek 😃 PA OWNS!!! 🆒

    compressed in 7-Zip 4.44beta (soild, Ultra mode, Dictionary 48MB and other settings on default)
    224 MB (235.136.186 bytes)
    14min 35sek

    compressed in 7-Zip 4.44beta (soild, Ultra mode, Dictionary 48MB and parameters like x=9 s=on f=on hc=on hcf=on mt=on a=1 mf=bt4 fb=273 mc=1000000000 lc=0 lp=0 pb=0
    202 MB (211.952.354 bytes)
    14min 17sek

    this test was performed on PIII 900Mhz, 768MB RAM

    and I think that settings for 7zip in PA2007 + Yours super mega tweaks improve compression speed = Super PA 2007 with the best compression ratio on the world :p

    You may test normal PA2007 settings (Ultra, Solild) and my setting on PA2007
    and You see what settings are better, because I dont have access to developer console :p
    and I dont know what settings are better, because this test was performed only in 7-Zip


  • conexware

    Here is the problem - if we use same settings as 7zip, speed will be the same :-).

    So it is better to have 2% worse compression and get a lot faster speed….



  • @spwolf:

    Here is the problem - if we use same settings as 7zip, speed will be the same :-).

    So it is better to have 2% worse compression and get a lot faster speed….

    Yep, I agree with You :p

    But I was thinking that, PA use only modified code with some compilation improvement for todays procesors 😉

    Well, this topic should be closed now 🆒



  • @spwolf:

    Here is the problem - if we use same settings as 7zip, speed will be the same :-).

    So it is better to have 2% worse compression and get a lot faster speed….

    Hmm…packing
    Vista x64(Atlon 64 3500+), PA 2007 beta, 7 zip 4.45 beta x64
    PA- 1m24s
    7z 4.45- 1m08s


  • conexware

    there are several things that might go there - like did you try compress each several times? For instance if you try 7z later, it will be faster since Windows cached the file already.

    In our tests, it is 10% faster, and gets 1% less compression, with files that can be compressed (unlike driver installation you tried)… Testing can also vary based on type of file you are testing as well…



  • In my tests in same cases(dif. files and queue) PA did best compression, but ALWAYS bad time…
    I think it may be only for Vista x64… 7 zip - 64bit!!! and PA only 32 bit…


Locked
 

4
Online

9.8k
Users

6.0k
Topics

36.6k
Posts

Copyright © 1998-2018 ConeXware, Inc.
All rights reserved. Privacy Policy