And results are fairly good : Zhuff takes first place at the Speed Oriented ranking, leading the chart by a sizable margin.
Looking at the detailed benchmark chart, the lead over previous king (the excellent 4x4 from Bulat Ziganshin) is especially telling, with +65% compression speed and +45% decoding speed. The difference would be even larger if bandwidth was not effectively limited by the RAM Drive speed. Finally, compression ratios remain fairly close, with a 4% advantage to 4x4.
As a sidenote, don't be fooled by the illusion that Zhuff and LZ4 are "about the same speed" : this result is entirely related to RAM Drive I/O saturation. Without this limitation, LZ4 would fare much better, as can be guessed from its CPU occupation ratio (only 160% to reach saturation, which means 100% for the RAM Drive driver, and "just" 60% (less than one core!) for the decoding algorithm).
Which lead us to the next question : how can we know which compressor is best suited for which speed ?
Using CompressionRatings data, we can build an interesting graph, using a simple file transmission scenario. In this scenario, file is compressed, then sent over a limited-bandwidth pipe, and then decoded. We take the simplistic assumption that there is no other delay in between. Total time is calculated and compared
(click on graph for larger display)
Looking at the graph, Zhuff is the better choice for pipes ranging from 5 MB/s to 200 MB/s. Above that point, LZ4 takes the lead (difference would be much more significantly if there was no RAM Drive I/O limitation issues, but well, that's the data we've got). Below that point, compression power matters more, with 4x4 becoming competitive at -3 level, then FreeArc and Nanozip getting into the fray.
It is also worth noting that more mainstream compressors, such as winrar, gzip, 7zip and even the parallel version pigz, do not reach top position in this graph.
You can download the latest version of Zhuff at its homepage.