Naive benchmark of compressors

compressors_benchmark.sh

#! /usr/bin/env zsh

# Naive benchmark of compressors
# (C) 2015 Jérémie Roquet <[email protected]>

# Disclaimer: only one run per program to get an idea, do not rely on the results!

# TODO wrapper for lzham because lzhamtest is not pipe-friendly (it needs to know the input size)

file=frwiki-20151002.xml

export TIMEFMT='%U'

get() {
    head -c $1 $file
}

bench() {
    (
	    {
	        get $1 |
	        (
		        time $=2
	        )
	    } > >(
    	    wc -c
    	) > >(
	        (
		        time $=3 > /dev/null
	        )
	    )
    ) 2>&1 | tr '\n' ' '
    echo \| $@[1,2]
}

bench_all() {
    echo --- bench@$1 ---

    bench $1 'lz4 -1' 'lz4 -d'
    bench $1 'lz4 -5' 'lz4 -d'
    bench $1 'lz4 -9' 'lz4 -d'

    bench $1 'gzip -1' 'gunzip'
    bench $1 'gzip -5' 'gunzip'
    bench $1 'gzip -9' 'gunzip'

    bench $1 'bro --quality 1' 'bro --decompress'
    bench $1 'bro --quality 4' 'bro --decompress'
    bench $1 'bro --quality 9' 'bro --decompress'
    bench $1 'bro --quality 11' 'bro --decompress'

    bench $1 'bzip2 -1' 'bunzip2'
    bench $1 'bzip2 -5' 'bunzip2'
    bench $1 'bzip2 -9' 'bunzip2'

    bench $1 'xz -0' 'unxz'
    bench $1 'xz -6' 'unxz'
    bench $1 'xz -9e' 'unxz'
}

for size in 5k 50k 500k 5M 50M; do
    bench_all $size
done