Set 1 is composed by various standard test files: Calgary and Canterbury corpora, enwik8, Generic compression benchmark, Waterloo sets, containing 134 MB (141506533 B) of data in 71 files and 10 folders.
It contains well known reference files widely used for compression benchmarks, representative of different data structures.
This is a classic compression benchmark, meant to synthetically evaluate how fast and efficient can be the compression and extraction of files representing various typical data structures.
Classic benchmark of speed (in the graph, represented by blue square points) and compression ratio (histogram) over few large files representing various typical data structures.
All archivers completed the test without errors.
ZIP format provided low compression ratio with good speed, but RAR and ARC reached excellent results in terms of speed (comparable with better ZIP compressors) providing far better compression ratio than ZIP format.
In the ZIP group PeaZip provided significantly better compression ratio than other applications, at comparable speed.
IZArc achieved the slowest ZIP compression, with compression ratio comparable with WinZip and WinRar; WinRar was the fastest ZIP compressor.
RAR format provided a better compression ratio tahn ZIP, but worse than other non-ZIP formats, and achieved a good speed result, that was comparable with better ZIP compressors.
ZIPX format significantly outperformed RAR compression ratio, but compression was twice slower; it even slightly surpassed 7Z compression ratio with twice faster compression time.
ARC (an experimental format introduced by FreeArc) showed impressive results both in terms of speed and compression, providing highest compression slightly outperforming both ZIPX and 7Z in compression ratio, and showing compression times faster than RAR and comparable to fastest ZIP compressors.
In the 7Z group IZArc reached better compression ratio than PeaZip, at cost of slower compression.
Set 1 Extraction
Classic extraction speed comparison benchmark over few large files representing various typical data structures.
All archivers completed the test without errors
ZIP and RAR formats achieved comparably very good extraction times, with exception of IZArc for ZIP format and, to a lessen extent, WinZip for RAR format, that were significantly slower than the average competitors for each group.
ZIPX format showed a strong drawback being very slow in extraction, completely outside the target time for all other formats, with PeaZip being faster than WinZip in ZIPX extraction.
To a lessen extent, also ARC format showed a drawback in terms of extraction speed, being twice slower than 7Z format, but 4-5 times faster than ZIPX.
7Z format extraction was significantly slower than extraction of RAR and ZIP formats on set 1, but faster than extraction of ARC and ZIPX formats, that in compression test had obtained slightly better compression ratio results with faster compression times.
CompositionSet 2 is composed by PortableApps.com Suite Light 1.6.1, installed, containing 125 MB (131464076 bytes) of data in 2277 files and 343 folders.
This set contains various applications for Win32, most common file types are: executables, resource files (icons, graphic), and various types of documentation files; applications are from different developers, so different strategies for space optimization and for resource bundling are used.
This benchmark is meant to tests the ability to efficiently backup and restore a complex nested structure populated by thousands files.
Also, it evaluates the efficiency (in terms of speed and achiveable compression ratio) in dealing with already partially optimized executables, documentation and resources.
In this test, the ability of efficiently handling I/O is fundamental for achieving good speed results, both for compression and extraction.
Set 2 Compression
Benchmark of speed (in the graph, represented by blue square points) and compression ratio (histogram) over a complex nested structure populated by thousands files, including partially optimized executables, documentation and resources.
The ability of efficiently handling I/O is fundamental for achieving good speed results in this test, while the ability of efficiently dealing with partially optimized data is fundamental to improve compression ratio.
IZArc failed creating archives for benchmark set 1, to complete the test as possible with IZArc the directory structure was flattened extracting all files and directories into the same directory.
For WinZip it was necessary to set inclusion of hidden and system files for benchmark set 1 in order to have all the files archived.
ZIP format provided fast compression but lower compression ratio than other archive formats.
IZArc was significantly slower than other compressors in creating a ZIP archive, also WinZip was significantly slower than PeaZip and WinRar (the fastest in ZIP creation); PeaZip provided better compression ratio for ZIP format, with a smaller margin than in set 1.
RAR and ZIPX formats provided a comparable compression ratio on set 2.
RAR compression was faster than ZIPX compression, but not as faster as in set 1.
ZIPX did not show the impressive compression performances reached on set 1: for both RAR and ZIPX compression ratio on set 2 was better than ZIP but worse than ARC and 7Z formats.
ARC compression achieved an impressive result in providing a file marginally larger than the smaller 7Z archive but with a compression speed comparable with fast ZIP compression.
7Z provided strongest compression, IZArc was twice slower than PeaZip in creating a 7Z archive and reached a worse compression level.
Set 2 Extraction
Benchmark of extraction speed of a complex nested structure populated by thousands files: the ability of efficiently handling I/O is fundamental for achieving good results in this test.
All archivers completed the test without errors
Set 2 extraction was significantly slower than set 1 due to the heavy I/O communication with system to create more than 2000 output files.
Average extraction times for various formats are more uniform than in set 1 extraction test, as in this test for the extraction software the efficiency in handling I/O communication is more important than optimization of extraction speed itself, so the system's responsivity equalizes many of the extraction speed differences between various formats and various utilities noticed on set 1.
In this test RAR format, especially with WinRar, shows generally lower extraction times than other formats including ZIP.
By archive format
ZIP format provides lower compression in all sets, but often shows an important speed advantage on most of other formats (when evaluating fastest applications) providing the fastest compression.
Optimized ZIP compression in PeaZip (provided by 7z backend) consistently achieves a significant advantage in compression ratio mantaining a competitive speed.
RAR format shows a clear and consistent advantage in terms of compression ratio over ZIP, with a moderate decrease of speed.
ZIPX format is slower than RAR in all sets. In terms of compression ratio, in ideal conditions (set 1), it can be a good match even for 7Z and ARC, but in other conditions (set 2) it provides a compression ratio comparable to RAR.
7Z format and the experimental ARC format provide overall the best compression ratio, with ARC worthing being noticed for excellent compression speed comparable with ZIP format.
IZArc achieves slowest speed in ZIP format, providing an average compression level. In 7Z format IZArc was slower than PeaZip in both sets, providing better compression ratio in set 1 and worse in set 2.
WinRar scores fastest ZIP compression, and with RAR format compression it reaches improved compression ratio at a reasonable cost in terms of speed.
WinZip performs on average in terms of speed and compression ratio for ZIP format. With ZIPX format WinZip provides a slower but in some cases better compressor than RAR.
PeaZip provides best ZIP compression level preserving an average speed. With 7Z format PeaZip achieves higher compression ratio than RAR and ZIPX, at cost of lower speed. The experimental ARC format is very promising as it provides both high compression ratio and high compression speed.
By archive format
RAR format shows to be a good match even for ZIP format in terms of fastest extraction speed, an excellent result because RAR consistently provides a better compression ratio than ZIP.
7Z format extraction is slower than RAR and ZIP ones, but this factor is balanced by the decisive advantages of 7Z over ZIP and RAR in terms of compression ratio.
ARC format is impressive for its compression efficiency, both in terms of speed and ratio, but extraction is significantly slower than 7Z format on set 1 where pure extraction performances are more relevant (difference is lower on set 2).
In the same situation, in set 1 ZIPX format extraction shows to be extremely slow, while being only marginally slower than 7Z on set 2.
IZArc provides significantly slower ZIP extraction than the average of other competitors, but achieves average extraction times for RAR and 7Z formats, especially on set 2.
WinRar is consistently amongst fastest extractors, with noticeably good results on ZIP and RAR format in set 2.
WinZip shows good extraction speed for 7Z format, but is consistently slower than PeaZip extracting ZIPX format.
It is also significantly slower than the average of competitors extracting RAR format in set 1, and ZIP format in set 2.
PeaZip is consistently faster than WinZip on ZIPX extraction, and on average for other formats, with proportionally better results in set 1 (where pure extraction speed is more relevant) rather than in set 2.