CompressionRatings.Com

Download

CompressionRatings.com corpus
Gauntlet corpus
Lossless Photo Compression Benchmark corpus
Squeeze Chart corpus

CompressionRatings.com corpus

Application1

OpenOffice.Org 3.0.1-Win32. Installed and precomped with precomp headers removed.

mirror: app1.tar.7z

Application2

MinGW64 compiler with files copied into a single file in random order.

mirror: app2.7z

Application3

http://sourceforge.net/project/downloading.php?groupname=portableapps&filename=PortableApps.com_Suite_Light_Setup_1.1.exe

execute to unpack

mirror: app3.tar.7z

Application4

Firefox 3.6.3, Inkscape 0.4.7.3, Thunderbird 3.0.4 and VLC 1.0.5 binaries. Precomped with its headers removed.

mirror: app4.tar.7z

Audio1

Audio1 is a diverse collection of CD-quality music. This is an original corpus that attempts to sample extreme range of different kinds of music. It is made for CompressionRatings.Com.

Currently a mini version of the corpus is being used:

audio1-mini.tar.7z (88,0 MB, contains wav-files)
audio1-mini.zip (58,5 MB, contains flac-compressed files)
audio1-mini.exe (55,9 MB, self-extracting nz archive containing wav-files)

The full version is available here:

audio1.wav.zip (187,6 MB, contains wav-files)
audio1.flac.zip (118,1 MB, contains flac-compressed files)
audio1.wav.exe (112,7 MB, self-extracting nz archive containing wav-files)

Game1

http://dcemulation.org/files/homebrew/nxdoom/law56ker-nxdoom-collection.rar

content of 'law56kers doom collection disc 1/doom' directory

mirror: game1.tar.7z

Game2

http://www.gamershell.com/download_21853.shtml

delete files: 'GamersHell.url' and 'GH3D.txt'

mirror: game2.tar.7z

Image1

http://www.imagecompression.info/test_images/

8-bit RGB files: 'artificial', 'deer', 'leaves_iso_200', 'nightshot_iso_100'

mirror: img1.tar.7z

Image2

http://www.imagecompression.info/test_images/

8-bit grayscale files: 'nightshot_iso_1600', 'spider_web'

mirror: img2.tar.7z

Text1

English language books from the Project Gutenberg etext00-02 archives. Manually selected. For each file 16384 bytes from the beginning and end of the file removed (to remove P.G. headers and other redundant information).

txt1.bz2 (26,9 MB)
txt1.exe (19,8 MB, self-extracting nz archive)

Text2

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

the 'enwik8' -file

mirror: txt2.7z

OS1

http://download.linhost.info/vmware/ubuntu904alpha2.7z

create file 'ubuntu500m' by taking the first 500000000 bytes of the image file 'Ubuntu 9.04 Desktop Alpha 2.vmdk'

mirror: os1.7z

Source1

ftp://ftp.irisa.fr/pub/mirrors/gcc.gnu.org/gcc/releases/gcc-4.2.0/gcc-core-4.2.0.tar.bz2

the content of 'gcc-4.2.0' directory

mirror: src1.tar.7z

Qualifying1

take the file 'gcc/libexec/gcc/mingw32/4.3.0/cc1.exe' from Application2 test

mirror: q1.7z

Qualifying2

create file 'txt5m' by taking the first 5000000 bytes of Text1 test

mirror: q2.7z

Database1

http://ftp.freedb.org/pub/freedb/freedb-complete-20080601.tar.bz2

create a file 'freedb100m' by taking the first 100000000 bytes of the extracted tar content (7z x -so freedb-complete-20080601.tar > freedb)

mirror: db1.7z

FP1

http://www.csl.cornell.edu/~burtscher/research/FPC/
http://www.csl.cornell.edu/~burtscher/research/FPC/datasets.html

decompress the 'obs_temp.trace' file

mirror: fp1.7z

Pgn1

http://www.top-5000.nl/dl/million.rar

create a file 'pgn100m' by taking the first 100000000 bytes from the file

mirror: pgn1.7z

Pitches1

http://pizzachili.dcc.uchile.cl/texts/music/

the full 'pitches' -file

mirror: pit1.7z

Medical1

http://www.data-compression.info/Corpora/lukas_2d_8_tif.zip

mirror: med1.tar.7z

Medical2

http://www.data-compression.info/Corpora/lukas_2d_16_tif.zip

mirror: med2.tar.7z

Special

Reference files

Files used in the "Reference files" test:

reference_files.zip

zhwik8

100 MB sample of Chinese Wikipedia that is used in the "Reference files" test.

zhwik8.bz2

enwik9

1000 MB sample of English Wikipedia that is used in the "BWT comparison".

enwik9.bz2 (242,2 MB)