Page 58 of 58 FirstFirst ... 848565758
Results 1,711 to 1,718 of 1718

Thread: paq8px

  1. #1711
    Programmer schnaader's Avatar
    Join Date
    May 2008
    Location
    Hessen, Germany
    Posts
    551
    Thanks
    206
    Thanked 182 Times in 87 Posts
    Results for reymont and FlashMX.pdf PDF files on a Hetzner cloud server (2 vCPU, 4 GB RAM, 64-Bit Ubuntu) to test the PDF part from v182:

    Code:
    				comp. size (bytes)	time (s)	memory (MiB)
    
    reymont (6,627,202 bytes), polish text, uncompressed PDF
      paq8px_v181fix1 -8		771,345			1795.53 	2204
      paq8px_v181fix1 -9		770,107			1770.64 	4028
      paq8px_v181fix1 -9a		765,737			1804.49 	4028
    
      paq8px_v182fix1 -9		758,969			1802.48 	4018
      paq8px_v182fix1 -9a		755,188			1836.77		4018
      
    FlashMX.pdf (4,526,946 bytes), english text, compressed PDF, many images
      paq8px_v181fix1 -8		1,329,386		1898.80		2529
      paq8px_v181fix1 -8a		1,336,691		1920.17		2529
      paq8px_v181fix1 -9		out of memory
    
      paq8px_v182fix1 -8		1,321,460		1844.51		2523
      paq8px_v182fix1 -8a		1,328,893		1868.74		2523
    Last edited by schnaader; 17th September 2019 at 20:25.
    http://schnaader.info
    Damn kids. They're all alike.

  2. The Following User Says Thank You to schnaader For This Useful Post:

    Gotty (17th September 2019)

  3. #1712
    Member
    Join Date
    Dec 2008
    Location
    Poland, Warsaw
    Posts
    940
    Thanks
    558
    Thanked 379 Times in 283 Posts
    Quote Originally Posted by schnaader View Post
    Results for reymont and FlashMX.pdf PDF files on a Hetzner cloud server (2 vCPU, 4 GB RAM, 64-Bit Ubuntu) to test the PDF part from v182:

    Code:
                    comp. size (bytes)    time (s)    memory (MiB)
    
    reymont (6,627,202 bytes), polish text, uncompressed PDF
      paq8px_v181fix1 -8        771,345            1795.53     2204
      paq8px_v181fix1 -9        770,107            1770.64     4028
      paq8px_v181fix1 -9a        765,737            1804.49     4028
    
      paq8px_v182fix1 -9        758,969            1802.48     4018
      paq8px_v182fix1 -9a        755,188            1836.77        4018
      
    FlashMX.pdf (4,526,946 bytes), english text, compressed PDF, many images
      paq8px_v181fix1 -8        1,329,386        1898.80        2529
      paq8px_v181fix1 -8a        1,336,691        1920.17        2529
      paq8px_v181fix1 -9        out of memory
      paq8px_v182fix1 -8        1,321,460        1844.51        2523
      paq8px_v182fix1 -8a        1,328,893        1868.74        2523
    Some my scores for FlashMX.pdf (difference to v182fix1 with option -9et = -8 bytes):
    paq8px_v182fix2 -9 1,320,362
    paq8px_v182fix2 -9a 1,327,992
    paq8px_v182fix2 -9et 1,319,236
    paq8px_v182fix2 -9eta 1,326,909
    Looks like "et" option gives some gain and option "a" hurts compression this time.

  4. The Following User Says Thank You to Darek For This Useful Post:

    Gotty (17th September 2019)

  5. #1713
    Member
    Join Date
    May 2008
    Location
    Estonia
    Posts
    395
    Thanks
    148
    Thanked 225 Times in 123 Posts
    Quote Originally Posted by kaitz View Post
    osdb improvment comes from Wordmodel i think, cant remember what context/check it was.
    px was worse because old wordmodel (v181) had :
    Code:
    if ((c>='a' && c<='z') || c==1 || c==2 || (c>=128 &&(c!=3))) {
    it affected word0. These are wrt Firstupper and Upperword.
    In pxd its part of text0 but not word0. etc.

    Now its almost same compression as pxd.
    KZo


  6. The Following 2 Users Say Thank You to kaitz For This Useful Post:

    Gotty (17th September 2019),LucaBiondi (17th September 2019)

  7. #1714
    Member Gotty's Avatar
    Join Date
    Oct 2017
    Location
    Hungary
    Posts
    378
    Thanks
    259
    Thanked 266 Times in 142 Posts
    Quote Originally Posted by kaitz View Post
    px was worse because old wordmodel (v181) had :
    Code:
    if ((c>='a' && c<='z') || c==1 || c==2 || (c>=128 &&(c!=3))) {
    it affected word0. These are wrt Firstupper and Upperword.
    Wow, I had no idea what that 1 and 2 could have meant. Now I know. A mystery is resolved in my head. Thanx.

  8. The Following User Says Thank You to Gotty For This Useful Post:

    LucaBiondi (17th September 2019)

  9. #1715
    Member Gotty's Avatar
    Join Date
    Oct 2017
    Location
    Hungary
    Posts
    378
    Thanks
    259
    Thanked 266 Times in 142 Posts
    Quote Originally Posted by Darek View Post
    enwik7 test scores with timings - in Excel file. I've made it for all versions that I've had on the laptop.
    And the two charts - scores and timings by version and scores in time (based on files dates).
    I think it could be good also input into Jarek's database.
    Nice! Well done!

    The explanation for the v153 speedup is here (exemodel is not applied to text blocks anymore).
    The explanation for the v179fix1 and v179fix2 speedup is here (removed stuff) and here (divisions were eliminated).

  10. The Following 2 Users Say Thank You to Gotty For This Useful Post:

    Darek (17th September 2019),LucaBiondi (17th September 2019)

  11. #1716
    Member
    Join Date
    Sep 2014
    Location
    Italy
    Posts
    35
    Thanks
    54
    Thanked 22 Times in 13 Posts
    Hi guys!

    We got big improvements from V182 to V182fix2

    JPEG files gain 225 KB (WOW!)
    PDF files gain 11 KB
    DOC files gain 400 bytes
    ISO files gain 1 KB
    other files gain 0 byte

    Click image for larger version. 

Name:	paq8px182fix2.png 
Views:	17 
Size:	40.2 KB 
ID:	6884

    NEW OVERALL RECORD!
    NEW JPEG RECORD
    NEW PDF RECORD
    NEW DOC RECORD
    NEW ISO RECORD

    Click image for larger version. 

Name:	paq8px182fix2_2.png 
Views:	25 
Size:	75.5 KB 
ID:	6885


    Well done Gotty & Kaitz you are a pro!
    I love it when plains come togheter!

    Luca
    follow my cutting edge blog @https://sqlserverperformace.blogspot.com/

  12. The Following 2 Users Say Thank You to LucaBiondi For This Useful Post:

    Darek (18th September 2019),Gotty (18th September 2019)

  13. #1717
    Member
    Join Date
    Dec 2008
    Location
    Poland, Warsaw
    Posts
    940
    Thanks
    558
    Thanked 379 Times in 283 Posts
    Looks like the v182fix change is biggest improvement for JPG files during lots of versions ago...
    @LucaBiondi - could you format yours second table numbers to 0 decimal places and to use 1000 separator? It would be slightly easier to read.

  14. #1718
    Member
    Join Date
    May 2008
    Location
    Estonia
    Posts
    395
    Thanks
    148
    Thanked 225 Times in 123 Posts
    Quote Originally Posted by kaitz View Post

    Code:
    DSCN0791.AVI (30018828 bytes)
                         Compressed  Time    Mem
     paq8pxd_v66.exe -s8 20236432    703.32  2277 MB
     paq8px_v181 -8      20354233    906.67  2378 MB 
     paq8pxd_v64.exe -s8 20558895    668.31  3157 MB
    Code:
    paq8px_v182fix2 -8 20321247 687.30  2372 MB
    KZo


  15. The Following User Says Thank You to kaitz For This Useful Post:

    Gotty (18th September 2019)

Page 58 of 58 FirstFirst ... 848565758

Similar Threads

  1. FrontPAQ - GUI frontend for PAQ8PF and PAQ8PX
    By LovePimple in forum Download Area
    Replies: 26
    Last Post: 17th January 2019, 13:36
  2. Alternative paq8px builds
    By M4ST3R in forum Download Area
    Replies: 20
    Last Post: 25th June 2010, 16:19
  3. Optimized paq7asm.asm code not compatible with paq8px?
    By M4ST3R in forum Data Compression
    Replies: 7
    Last Post: 3rd June 2009, 15:34

Tags for this Thread

Posting Permissions

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