benchmarkbenchmark results 1

 

1 billion byte text file

A text file with the size of 1,000,000,000 bytes [2] is compressed using FastLZ-based 6pack compressor [3] and the de-facto standard zip utility [4]. The time for compression and decompression [5, 6] and the size of the compressed file are compared, giving the following result (shorter is better):

It is obvious that zip can compress better since FastLZ is not optimized for compression ratio, but rather the speed. Although the compressed size of zip is 77% compared to that of FastLZ, FastLZ compresses almost 4x faster and decompressed almost 2x faster than zip, reaching the remarkable compression speed of over 50 MB/s for the above test case.

Notes

[1] Please take into account that benchmark results always depend on the test data. Real-world performance will vary based on system configuration, working data, and other factors.

[2] It is the first one billion characters of English Wikipedia dump on Mar. 3, 2006. See also test data for large text compression benchmark from Matt Mahoney.

[3] See more info on 6pack.

[4] zip command line archiver is taken from Info-ZIP's zip 3.23. It is executed in the fastest mode, i.e. zip -1.

[5] Test was performed on Intel Pentium 4-M 1.8 GHz, 512 MB RAM running Windows XP.

[6] Execution time was measured using timer 3.01 utility from Igor Pavlov. Only process time (kernel time + user time) is considered, thereby excluding disk I/O time.