Re: some linux tuning (fs benchmarks)

On Mon, 28 Jun 1999, Doug Renner wrote:

> For simple paradigms, I just use bonnie
> (ftp://metalab.unc.edu/pub/Linux/system/benchmark/bonnie.tar.gz) and
> execute it in a little loop. This just runs the FS for 1MB, 50MB, and
> 100MB objects:
>
> for x in 1 1 1 50 50 50 100 100 100
> do
> ./bonnie -s ${x} | grep [0-9] >> /tmp/bonnie.out
> done
>
> I imagine that for our purposes, it'd be intelligent to use smaller
> writes.

bonnie runs that fit entirely in memory are virtually meaningless as
benchmarks.

              -------Sequential Output-------- ---Sequential Input-- --Random--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
            1 4993 92.7 19994 97.6 7332 93.1 4367 98.1 29177 114.0 3864.1 199.0

That't from a dual P120, 64mb, 9gb SCSI disk. The SCSI bus is incapable of
19mb/s or 29mb/s, and the disk couldn't possibly do 3864 seeks per second.

              -------Sequential Output-------- ---Sequential Input-- --Random--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
          128 4373 90.1 4793 31.3 1864 26.5 3829 76.6 4035 25.5 78.8 4.7

Running the test with the file size 2x memory, gives much more realistic
numbers for this old disk.
  Vendor: MICROP Model: 1991-27 1128RV Rev: 28RV

----don't waste your cpu, crack rc5...www.distributed.net team enzo---
 Jon Lewis *jlewis@lewis.org*| Spammers will be winnuked or
 System Administrator | nestea'd...whatever it takes
 Atlantic Net | to get the job done.
_________http://www.lewis.org/~jlewis/pgp for PGP public key__________
Received on Mon Jun 28 1999 - 11:08:06 MDT