

                 
                     
                               
                              
                              
                             

                            
                               
                               
                            

         
                               
                           
                    




              -=    Battle of the Compressors 1993    =-

        The following is a text file describing tests I did on the
        most popular compressors today.  I matched them head-to-head
        in an attempt to determine which did the best job.
        I saved my opinion and closing comments for the last section.
        So for now, let's see the results:













       ۻ   ۻ  ۻ   ۻ ۻ   ۻ ۻ           ۻ
       ۻ ۻ ۺ   ۺ ۻ  ۺ ۻ         ۺ
       ɼ ۺ   ۺ ۺ   ۺ ۻ ۺ ۺ  ۺ         ۺ
       ۻ ۺ   ۺ ۺ   ۺ ۺۻۺ ۺ  ۺ          ۺ
       ۺ  ۺ ɼ ɼ ۺ ۺ ɼ          ۺ
       ͼ  ͼ  ͼ   ͼ  ͼ  ͼ ͼ           ͼ





        I compressed nine single files, in three groups.

        LRG, MED and SML EXE were single executable files.
         LARGE  EXE FILE  = TC.EXE       from the Turbo C++ 3.0
         MEDIUM EXE FILE  = SETUP.EXE    from Windows 3.1
         SMALL  EXE FILE  = USER.EXE     from Windows 3.1

        LRG, MED and SML DATA were single data files.
         LARGE  DATA FILE = EDAM20.DAT   from Writer's Toolkit 2.0
         MEDIUM DATA FILE = GR_AM.LEX    from Word for Windows 2.0
         SMALL  DATA FILE = GRAMMAR.DLL  from Word for Windows 2.0

        LRG, MED and SML DOC were single document files.
         LARGE  DOC FILE  = TCXL55.TXT   from the CXL Library
         MEDIUM DOC FILE  = ARJ.DOC      from ARJ 2.39c
         SMALL  DOC FILE  = CLASSLIB.DOC from Turbo C++ 3.0

                      Here are the results:



  File     Original     PKZIP       ARJ      SQUEEZE      HAP       LHA
  Type    Size        2.04g     2.39d    1.08.3      3.0      2.13
͸
LRG  EXE 1,263,536  525,841   530,002   527,242   531,890   543,427  
MED  EXE   422,080  191,497   193,495   192,613   196,941   198,108  
SML  EXE   264,016  159,823   160,767   159,604   160,498   165,074  
                                                                     
͵
LRG DATA 1,935,360  1,745,098 1,750,275 1,743,423 1,869,035 1,764,467
MED DATA   780,688    463,108   467,266   463,090   469,209   487,099
SML DATA   311,808    142,571   143,939   142,642   139,994   147,207
                                                                     
͵
LRG  DOC   444,693     88,183    89,817    89,252    86,014    96,117
MED  DOC   143,835     45,323    46,146    45,802    37,637    49,809
SML  DOC    55,734     13,138    13,297    13,137    12,133    13,835
                                                                     
;


        The compression was timed using TYME 2.0 by Bob Eyer.  Here
        are the results:


             File     PKZIP    ARJ    SQEEZ    HAP    LHA
             Type    2.04g  2.39d  1.083   3.0   2.13
           ͸
           LRG  EXE 36.15  41.59  39.50  167.03 35.65 
           MED  EXE 13.02  14.72  14.39   55.60 13.02 
           SML  EXE  9.06   9.06   9.94   43.29  8.73 
           ͵
           LRG DATA 68.02  60.43  66.53  508.62 53.79 
           MED DATA 25.54  28.57  25.54  123.84 23.90 
           SML DATA  9.56  10.93  10.32   41.92  9.67 
           ͵
           LRG  DOC 10.54  16.09  11.75  29.61  13.07 
           MED  DOC  5.54   4.34   5.93   9.61   5.10 
           SML  DOC  1.75   2.47   2.30   4.50   2.30 
           ;



 EXES   PKZIP did the best, followed by SQUEEZE.  LHA dragged in last
        place, although did excellent in the time column.
        Overall, everybody was real close to each other, and no one
        stood out prominently.  Timewise, HAP was embarassing, taking
        five times longer than everyone else.

 DATA   SQUEEZE took the prize here, followed by PKZIP.  Again, LHA
        dragged in last.  Everybody still almost tied; no big differences
        in performance.  HAP was incredibly slow, however.

 DOC    HAP beat the pants off every other compressor.  No one could even
        touch HAP, although it took twice as long to compress.  Yet again,
        LHA dragged in last.













        ۻ   ۻ  ۻ   ۻ ۻ   ۻ ۻ          ۻ
        ۻ ۻ ۺ   ۺ ۻ  ۺ ۻ         ۻ
        ɼ ۺ   ۺ ۺ   ۺ ۻ ۺ ۺ  ۺ          ɼ
        ۻ ۺ   ۺ ۺ   ۺ ۺۻۺ ۺ  ۺ         ͼ
        ۺ  ۺ ɼ ɼ ۺ ۺ ɼ         ۻ
        ͼ  ͼ  ͼ   ͼ  ͼ  ͼ ͼ          ͼ

        I compressed a small, medium and large mixture of files in
        the first test.

                   SML MIX consisted of my root directory (9 files)
                    of mixed .bat, .sys and other files.

                   MED MIX consisted of my C:\TASM directory, a mix
                    of 17 executable, .com and .asm files.

                   LRG MIX consisted of my C:\WINDOWS directory, a
                    mix of 109 files, the usual Windows 3.1 type.



        Then I compressed a small, medium and large number of files.

                   SML NUM consisted of 69 C++ files from
                    Turbo C++ 3.0.

                   MED NUM consisted of 97 C++ files.

                   LRG NUM consisted of 490 C files from 15
                    different directories.




        My third test was kind of odd.  I decided to compress a
        218,864 byte executable file using this method:
                I compressed it with ARC
                I compressed the ARC'd file with SQUEEZE
                I then compressed the Squeeze'd file with PKZIP
                I compressed the PKZIP'd file with ARJ
                I finally compressed the ARJ'd file with LHA

        I did this "layering of compressors" to see if another
        compressor would shrink an ALREADY compressed file.  My first
        reaction was "it will probably expand the file", and it did.
        But the results were interesting...

        Here are the results:






  File     Original     PKZIP       ARJ      SQUEEZE      HAP       LHA
  Type    Size        2.04g     2.39d    1.08.3      3.0      2.13
͸
SML MIX    138,040     33,381   33,486     32,770   33,131     33,277
MED MIX    481,262    226,915  227,631    226,170  220,965    234,527
LRG MIX  3,960,328  2,043,9192,062,029  2,043,6162,084,052  2,070,513
                                                                     
͵
SML NUM     59,109     34,514   31,244     30,642   28,798     30,150
MED NUM    652,562    193,848  190,791    188,933  179,072    191,078
LRG NUM  1,394,165    521,876  503,549    496,659  466,637    497,158
                                                                     
͵
                                                                     
ALREADY    136,426    136,534  136,623    136,468  136,574    136,459
COMPRESS                                                             
                                                                     
;



              File     PKZIP    ARJ    SQEEZ    HAP    LHA
              Type    2.04g  2.39d  1.083   3.0   2.13
            ͸
            SML  MIX  2.25    3.35  3.20   15.05  3.29 
            MED  MIX 13.62   13.90 16.59   73.40 16.64 
            LRG  MIX106.70  173.24132.91  763.46122.91 
            ͵
            SML  NUM  3.40   10.65 10.32   67.58  9.89 
            MED  NUM 14.17   36.70 38.46  150.65 40.54 
            LRG  NUM 88.18  205.93206.42 1063.46211.81 
            ͵
            ALREADY   5.32    5.38  5.76   43.13  5.16 
            COMPRESS                                   
            ;

 MIX    SQUEEZE won two awards, and HAP got one.  Everyone was extrememly
        close in the race, although HAP continued to embarass itself time
        wise.  HAP lugged the compression SEVEN times longer than any other
        program.  PKZIP was definitely the fastest, and differed little in
        its superb compression.  I thought the -EX switch would bring it
        to a crawl, but it clearly proved itself here!

 NUM    HAP took all the awards by a huge margin.  Yet again, at a cost.
        SQUEEZE came in second, and PKZIP last.

 ALREADY COMPRESSED
        LHA won, but most likely because the file was LHA'd before the test
        began.  SQUEEZE came in second, but most of the archivers performed
        equally well.












        ۻ   ۻ  ۻ   ۻ ۻ   ۻ ۻ          ۻ
        ۻ ۻ ۺ   ۺ ۻ  ۺ ۻ         ۻ
        ɼ ۺ   ۺ ۺ   ۺ ۻ ۺ ۺ  ۺ          ɼ
        ۻ ۺ   ۺ ۺ   ۺ ۺۻۺ ۺ  ۺ          ۻ
        ۺ  ۺ ɼ ɼ ۺ ۺ ɼ         ɼ
        ͼ  ͼ  ͼ   ͼ  ͼ  ͼ ͼ           ͼ

        I compressed an absolutely HUGE file (12 megabytes) that was created
        using ARC, and storing a two meg file five times.  ARC, of course
        actually expanded the file to almost twelve meg.
                (I was really stretching my 210 meg hard drive...)


        The second test consisted of 100 font files from All Fonts program
        that were split up into small, medium and large chunks.

        For small, I used one font file.
        For medium, I used 9 font files (all starting with 'G')
        For large, I used all 100 fonts.


        Here are the results:


  File     Original     PKZIP       ARJ      SQUEEZE      HAP       LHA
  Type    Size        2.04g     2.39d    1.08.3      3.0      2.13
͸
  HUGE  11,612,491 10,469,20410,500,87011,612,54311,214,26410,586,545
                                                                     
SML FONT    66,160     14,394    14,481    14,386    15,862    14,353
MED FONT   280,319     61,780    61,839    62,063    64,020    61,792
LRG FONT 2,841,294    803,066   804,027   802,669   847,240   800,412
;



          The compression was timed using TYME 2.0 by Bob Eyer.  Here
          are the results:


                 File     PKZIP    ARJ    SQEEZ    HAP    LHA
                 Type    2.04g  2.39d  1.083   3.0   2.13
               ͸
                 HUGE   415.98 355.16 51.04 3051.15 331.75
                                                          
               SML FONT   5.65  12.03  3.40    4.89   2.80
               MED FONT  11.53  36.37 10.38   25.05   9.28
               LRG FONT 140.21 370.00109.39  323.13  88.40
               ;

 HUGE   PKZIP won this award, followed by ARJ.  The huge twelve meg.
        file was actually expanded by SQUEEZE.
        Notice that HAP took almost an HOUR to compress it.
        Geeeeeeeeeeeeeeeeeeeeeeez!  I went to Taco Bell, got a burrito,
        went to my girlfriend's house and played video games, and it
        STILL wasn't done!  That's 7.5 times longer than the other
        competitors.

 FONTS  LHA took the awards for fonts.  It compressed them faster and
        tighter than anyone else (PKZIP won the MDFONT though).
        ARJ was the absolute slowest font compressor.




















============= THE ULTIMATE SQUISHY SQUISHY ==================================

        I have received many letters stating "You didn't use the correct
        switch for the compressor!", and claimed that the compressor
        was capable of MUCH MORE than I was testing.

        For the ultimate SQUISHY SQUISHY, I used the ABSOLUTE MAXIMUM
        compression switches, and threw out the time variable.  I didn't
        care if it took four hours, I wanted the absolute best compression.

        The following switches were used:

        PKZIP   -ex
        ARJ   a -jm -jt -jh16384
        SQZ   a /q
        HAP   a
        LHA   a


        I then chose to compress the "ultimate CLIP ART, DOC and EXE
        files", from Word for Windows.

        ULT CLIP       were 20 clip art files
        ULT  DOC       were all the DOC files in Winword
        ULT  EXE       was Winword.exe


        Here are the results:


  File     Original     PKZIP       ARJ      SQUEEZE      HAP       LHA
  Type    Size        2.04g     2.39d    1.08.3      3.0      2.13
͸
                                                                     
ULT CLIP  104,624    47,906    47,324    46,605     48,229    46,526 
ULT  DOC  380,164   109,329   109,944   109,675    107,114   114,092 
ULT  EXE1,268,224   875,710   879,854   876,108    905,691   890,870 
                                                                     
;

 Everybody did about the same with clipart, but LHA won by a hair.  HAP
 and PKZIP embarassed themselves here.  With DOC files, HAP continues
 to remain untouched.  Absolutely incredible, but there was a wait...

 PKZIP continues to compress executables the best (barely behind SQUEEZE).
 Everybody made a good showing with EXES.



















                          
                             
                                 
                              

                  
                          
                          
               




        So who won?

  *  SPEED *Ŀ
                                                     
          PKZIP          10 awards out of 22         
          LHA            10 awards out of 22         
          ARJ             1 award  out of 22         
          SQUEEZE         1 award  out of 22         
                                                     
  

  * COMPRESSON *Ŀ
                                                     
          HAP             9 awards out of 23         
          PKZIP           5 awards out of 23         
          SQUEEZE         5 awards out of 23         
          LHA             4 awards out of 23         
                                                     
  



  PKZIP =       This is great for speed, especially on mixtures
                of files, and large number of files.  If you are
                in a hurry, use PKZIP.  I used the -EX switch,
                which probably slowed it down a bit.  If you just
                use PKZIP A, I am sure you'll find a big improvement
                on speed.  For a general compressor with good speed,
                good squishing....  try PKZIP.  It is also the
                most popular compressor on BBS' and with Shareware
                distributors.

  HAP =         HAP was extremely slow, and made an idiot of itself
                with the HUGE twelve meg file.  If you're in a hurry,
                don't use HAP.
                On the flipside, it's great for compressing text files.
                It beat the PANTS off everybody else!
                If you have a 486-50 or better and lots of text to
                compress, use HAP.
                It does not have damaged file recovery, so if you have
                a disk error, you're screwed.
                I think the Author got a little too excited about his
                program, though.  He states in the FILE_ID.DIZ,
                "Compresses up to 30% better than the NEW Pkzip", which
                is definitely a false statement.

  SQUEEZE =     This thing impressed me a LOT.  It's fantastic for
                data, mixtures of files and large number of files.
                It's extremely slow, though.  But lately, with 486-66
                and the new 586 coming out, I'm sure you won't notice
                the difference.  If you want damn good compression,
                try this out.  It has absolutely no features though,
                so beware.  A disk error will send you searching
                for a backup!

  ARJ =         Great for features, but slow.  The compression is average.
                There's a whole lot of ARJ supporters out there, and I
                understand why.  If you have the time to read up on
                ARJ.DOC, you'll find a wealth of goodies.
                If you don't like PKZIP, and want a reliable, average
                compressor, use this.

  LHA =         Extremely fast, but not so good compression.  It compresses
                graphic files great, but chokes hard on single files,
                executables and general mixtures of files.

  ARC =         No comment.  It doesn't even compress a data file, no less
                rating 30-60% worse compression than everyone else.  I
                sincerely hope this compression program falls by the side
                QUICKLY on major BBS's.










 Ŀ Ŀ  Ŀ Ŀ  Ŀ Ŀ    Ŀ Ŀ Ŀ    Ŀ Ŀ  Ŀ
          Ŀ             Ŀ               Ĵ        Ŀ
                                   


        All tests were done on a 386-40 with a Western Digital 210 meg
        hard drive using DOS 5.0 and a 64 built-in cache.
        I used DOS' himem.sys, and DOSKEY installed.

        To do the actual tests, I created a batch file consisting of:

        tyme  pkzip -ex     %1 %2     >1zip
        tyme  arj a -e -jm  %1 %2     >1arj
        tyme  sqz a         %1 %2     >1sqz
        tyme  hap a         %1 %2     >1hap
        tyme  lha a         %1 %2     >1lhz


        NOTE: ARJ users --> According to the Author's documentation,
              "ARJ a -e -jm" is supposed to be used for testing
              purposes.  I quote: "The ARJ -jm compression is intended
              to demonstrate the best that ARJ can do ..."


        and then used: DO LGEXE C:\ND\ND.EXE, etc...
        This allowed me to time the programs, and eliminated a lot
        of unnecessary typing.

        All of the compression programs were in the current directory,
        and were downloaded from a local BBS.


Timing done by:         TYME 2.0  from Bob Eyer

LARGE  EXE FILE  = TC.EXE       from Turbo C++ 3.0
MEDIUM EXE FILE  = SETUP.EXE    from Windows 3.1
SMALL  EXE FILE  = USER.EXE     from Windows 3.1

LARGE  DATA FILE = EDAM20.DAT   from Writer's Toolkit 2.0
MEDIUM DATA FILE = GR_AM.LEX    from Word for Windows 2.0
SMALL  DATA FILE = GRAMMAR.DLL  from Word for Windows 2.0

LARGE  DOC FILE  = TCXL55.TXT   from the CXL Library
MEDIUM DOC FILE  = ARJ.DOC      from ARJ 2.39c
SMALL  DOC FILE  = CLASSLIB.DOC from Turbo C++ 3.0





==================== FINAL COMMENTS ==========================================

        WARNING:  This is my opinion, so if you don't want to hear it,
                  don't read it:




        Although SQEEZE did quite well in all the tests, it is not
        supported by many BBS', and does not have a lot of support.

        IE: If you have a damaged SQUEEZE or HAP file, you probably
            cannot recover any data from it.  But PKZIP has
            PKZIPFIX.EXE, and would be able to recover something.
            The SQUEEZE file is also twice as large as PKZIP,
            and therefore would use more space on a hard drive.

            I am impressed at SQUEEZE and HAP's compression, but not
            its overall program.  I am also impressed at ARJ's options
            (all 6 pages of options), but not its compression and speed.

            Overall, I probably will stick with PKZIP, just because of its
            reliability and speed.  If I have a SQUEEZE file that loses
            a couple of bytes because of disk errors, then I am screwed.

            Hey, I also program in C++, and have some stuff available.
            Run the CATALOG.EXE file for details on ordering and what's
            new.

                                 Write to:

                                David Smith
                               1104 Mason Dr.
                               Hurst TX 76053






   ۻ ۻ   ۻ ۻ  ۻ   ۻ  ۻ  ۻ ۻ
   ͼ ۺ   ۺ ۻ ۻ ۻ ۻ ͼ ۻ
ۻ ۺ   ۺ ɼ ɼ ۺ   ۺ ɼ    ۺ      ķ
  ۺ ۺ   ۺ ͼ  ͼ  ۺ   ۺ ۻ    ۺ      ͼ  
  ۺ ɼ ۺ      ۺ      ɼ ۺ  ۺ    ۺ      ۻ    
  ͼ  ͼ  ͼ      ͼ       ͼ  ͼ  ͼ    ͼ      ͼ    
                                                                             
                                                                             
      Support for my programs is available in the following ways:            
                                                                             
                                                                             
          Ŀ Ŀ Ŀ Ŀ     Ŀ Ŀ Ŀ    Ŀ                 
                         Ŀ     ӷ                    
                                             
          User ID:  71441,2723                                               
          I usually log on every other day and check my mail.                
                                                                             
                                                                             
                                                          
                                                             
                                                             
                                                        
          Drop me a line on the MotherBoard II Bulletin Board                
                     David Smith   (214) 642-1287                            
                                                                             
                                                                             
                                                     
                                              
                                              
                                                 
                                             
                                                
                               Write to:                                     
                                                                             
                              David Smith                                    
                            1104 Mason Dr.                                   
                            Hurst TX 76053                                   
                                                                             
Ľ





