Improves memchunk benchmark to account for block size.

[cyriltovena]

No major code change except some cleanup.

none_524 kB: 3952.20 MB/s
none_262 kB: 3818.19 MB/s
none_262 kB: 3588.39 MB/s
none_524 kB: 3528.82 MB/s
none_66 kB: 3347.10 MB/s
none_66 kB: 3306.29 MB/s
none_524 kB: 3052.64 MB/s
none_262 kB: 2932.31 MB/s
lz4-1M_66 kB: 2567.87 MB/s
lz4_262 kB: 2467.42 MB/s
none_66 kB: 2453.80 MB/s
lz4-256k_524 kB: 2398.77 MB/s
lz4-1M_262 kB: 2295.78 MB/s
lz4-256k_524 kB: 2289.92 MB/s
lz4-256k_262 kB: 2281.90 MB/s
lz4-1M_524 kB: 2213.51 MB/s
lz4-256k_66 kB: 2187.04 MB/s
lz4_524 kB: 2183.28 MB/s
lz4_524 kB: 2177.57 MB/s
lz4-1M_66 kB: 2148.51 MB/s
lz4_66 kB: 2145.61 MB/s
lz4-1M_262 kB: 2120.39 MB/s
sample_none_66 kB: 2104.96 MB/s
lz4-64k_66 kB: 2098.88 MB/s
lz4-64k_524 kB: 2077.25 MB/s
lz4-64k_524 kB: 2073.97 MB/s
lz4-256k_66 kB: 2070.37 MB/s
lz4_262 kB: 2052.62 MB/s
lz4-256k_262 kB: 2030.32 MB/s
lz4_66 kB: 2013.14 MB/s
lz4-64k_262 kB: 2009.75 MB/s
lz4_262 kB: 1998.54 MB/s
lz4-64k_66 kB: 1992.59 MB/s
lz4-64k_262 kB: 1942.55 MB/s
lz4-1M_524 kB: 1914.17 MB/s
lz4-256k_262 kB: 1865.59 MB/s
snappy_524 kB: 1859.09 MB/s

Interesting to see lz4 beating snappy now.

I'm guessing this is due to the an upgrade of the vendor code.

As for size lz4 has a compression ratio of 13 vs 9 for snappy.

Signed-off-by: Cyril Tovena cyril.tovena@gmail.com

[sandeepsukhani]

LGTM