[Balloon] IO performance numbers

I ran bonnie++ on the armel machine on flash and USb drive (lacie 250GB)

bonnie++ -d /tmp -n 3 -m balloon-armel -u wookey

(/tmp is flash not ram on this box, currently)

here is what I got:
Version  1.03       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
balloon-armel  496M   816  99   706  99   621  99  1384  99  1960  99 269.4  99
                    ------Sequential Create------ --------Random Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                  3    56 100  3675  99    30  99    55  99 +++++ +++    56  99
balloon-armel,496M,816,99,706,99,621,99,1384,99,1960,99,269.4,99,3,56,100,3675,99,30,99,55,99,+++++,+++,56,9

bonnie++ -d /mnt/tmp -n 3 -m balloon-armel -u wookey

Version  1.03       ------Sequential Output------ --Sequential Input-
--Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
balloon-armel  496M   849  26   854   3   444   1   939  25   947   1  50.3   2
                    ------Sequential Create------ --------Random Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                  3  5241  98 +++++ +++ +++++ +++  5580  99 +++++ +++ +++++ +++
balloon-armel,496M,849,26,854,3,444,1,939,25,947,1,50.3,2,3,5241,98,+++++,+++,+++++,+++,5580,99,+++++,+++,+++++,+++

bonnie does IO tests for throughput (top half) (database type
activity), and then test on lots of small files (bottom half)
(mail/compile type activity)

Decoding this lot:
testing IO throughput with 496MB files. (i.e approx twice size of memory: 256MB on
this box), do files tests with 1024*3 files (-n 3).

%CP is %CPU usage. So looks like flash acces is CPU limited as it's
jammed at 99-100% on all tests. USB access on the other hand has
_much_ lower CPU usage (for IO, but not the file tests) and similar
speed for ouptu, but about half speed for input.

+++++ on the file read tests shows it took less than 500ms. I'll re-do
with more files...

See http://www.penguin-soft.com/penguin/man/8/bonnie++.html
for details

Wookey
--
Principal hats: Balloonz - Toby Churchill - Aleph One - Debian
http://wookware.org/