QuadCEM/QuadBlog

Summary

All in all, the choice you make really depends on a variety of factors.  If you’re looking for maximum compression, regardless of portability, processing time and power, then clearly PeaZip has a lot to offer.  With PAQ8O, PeaZip managed to smash a highly mixed set of files down to 58% of its original size, though it took over 4.5 hours to do.  In just under 6 minutes, PeaZip can also compress these files down to 60% with LPAQ8 – still excellent and much better on time.  Along those same lines, NanoZip also shows a lot of potential;  its CM algorithm compressed to 60% and in slightly less time than PeaZip’s LPAQ8 algorithm.

One level higher in terms of portability is the 7Z archive format.  Using the LZMA algorithm, PeaZip and 7-Zip both compressed my test files down to 61-62% of their original — all in under a minute.  The only thing that could come close to that was the RAR format (at 63%).  Although the RAR format is more popular and widely spread than 7Z, it is also proprietary (so no one but WinRAR can really create RAR files — and WinRAR costs $$$).

Finally we make it to the most popular formats on the web — ZIP, GZ and BZ2.  Based on what I could tell, these formats were all very closely matched (the differences were almost negligible).  They could all compress my test files to 65-66% with most of the programs, usually in less than a minute.  The WinRAR implementation had the worst ZIP results (maxed to a 67% ratio) — which is one more reason not to use it!  The best programs in the category were, once again, PeaZip and 7-Zip.

After all is said and done, I have no choice but to recommend using PeaZip, followed closely by 7-Zip.  The great thing about PeaZip is that it gives you the option of using either (incredibly powerful) niche algorithms or popular archive formats (ZIP, GZ, BZ2, etc.), and it does an exceptional job with compression.  7-Zip also gives you the option of compressing to the popular formats, but typically not as well as PeaZip.  Both of these programs typically did a better job at compression than Tar, WinRAR and WinZip.  NanoZip may someday get a recommendation as it gets developed further.

One last important thing to note is that compression is a very finicky thing.  Some algorithms work better on certain data sets than others, and sometimes don’t work as well on other data sets.  One thing especially true is with text data sets vs multimedia vs binary data (such as object files).  Perhaps the next article will try each of these programs out with text-only data, multimedia files, and binary data separately (instead of mixed together nicely).