> -------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
> 1024 4799 23.8 5053 4.9 3611 9.6 12344 50.5 14878 22.8 186.6 2.0
> 1900 5194 25.8 5048 4.9 3506 9.4 12595 51.7 14682 22.7 134.5 1.5
>
> This is in a RAID-5 config with 5 9gb UW drives in a dual PII-450 with
> 512mb RAM running kernel 2.2.2-smp.
disk benchmarks with files merely 2x ram are marginally useful;
I prefer to stick to 3x ram.
while there are good reasons for raid5, these throughput numbers are sad.
a single $16/GB UDMA disk can sustain around 16 MB/s... (Maxtor 4320,
PII/266, bx, 4K ext2).
> Have others noticed that increasing the file size with bonnie seriously
> decreases the random seeks/s measurement? I've seen the same thing
the only bonnie numbers you should ever pay attention to are block IO.
per-char numbers are measuring your libc implementation, compiler and BUFSIZ.
the seeks are amusing, but mainly a function of unuseful things like
how the FS happens to lay out the file, blocksize, readahead, etc.
regards, mark hahn.
--
operator may differ from spokesperson. [EMAIL PROTECTED]
http://java.mcmaster.ca/~hahn
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
