Squashfs benchmarks

24

August
2012

Before I choose the right block size and compression method for Slax filesystems, I always need to do some benchmarks. Today, I took a 100MB filesystem consisting of linux kernel (already compressed) and driver modules (not compressed), which is IMHO ideal sample for benchmarks.

I will write another blog post with conclusions based on the results mentioned below.

format	block size	uncompressed size	compressed size	ratio
gz	64	109 189 120	36 491 264	33,42%
gz	128	109 189 120	36 159 488	33,12%
gz	256	109 189 120	36 036 608	33,00%
gz	512	109 189 120	35 995 648	32,97%
gz	1024	109 189 120	35 966 976	32,94%
lzma2	64	109 189 120	31 473 664	28,82%
lzma2	128	109 189 120	30 445 568	27,88%
lzma2	256	109 189 120	29 712 384	27,21%
lzma2	512	109 189 120	29 089 792	26,64%
lzma2	1024	109 189 120	28 549 120	26,15%

format	block size	time for reading
gz	64	16.698
gz	128	16.608
gz	256	16.526
gz	512	16.507
gz	1 024	16.416
lzma2	64	18.638
lzma2	128	18.558
lzma2	256	18.453
lzma2	512	18.282
lzma2	1 024	18.261

format	block size	RAM for mount (KB)
gz	64	792
gz	128	2 224
gz	256	1 292
gz	512	2 196
gz	1 024	5 204
lzma2	64	1 028
lzma2	128	1 436
lzma2	256	2 256
lzma2	512	2 864
lzma2	1 024	6 336

Slax author's Blog (RSS)

24

Squashfs benchmarks

User comments