Added first published results -> https://github.com/ning/jvm-compressor-benchmark/wiki

[Tatu Saloranta]

Ok finally got enough tests run & figure out how to use github, so
first test results are now linked to from
[https://github.com/ning/jvm-compressor-benchmark/wiki].

Let me know if they make sense; Japex produces graphs that look ok,
but explanations are bit sparse. For each test there are two sets of
numbers: "value" is throughput in megabytes per second (of
uncompressed data), and "valueX" is relative compressed size (so
40.91, for example would mean that resulting data is 40.91% of
original, giving 59.09% compression rate).

There isn't anything too surprising in results; bzip2 gives best
compression ratio typically, but is dead slow to compress and
decompress. LZF is the fastest, with lowest compression ratio; and it
is particularly fast to compress relatively speaking.
QuickLZ gets similar compression in L1 mode with bit less speed; and
bit better compression (but much worse speed) in L3 mode.
JDK/gzip(deflate) gets pretty decent speed with average compression
(better than LZF/QuickLZ, not quite as good as bzip2).

-+ Tatu +-