At the moment I can offer you the Bonnie results on a single
Seagate Cheetah drive attached to an integrated AIC-7890 LVD
controller.
I think the possible differences between 512 o 2000 Mb file size,
should be shown in a similar way through the results on this single
disk.
512Mb file size:
-------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* 512 5812 76.9 21659 23.3 6179 16.1 7498 94.6 19780 21.6
859.2 9.2
2000Mb file size:
-------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*2000 6148 79.1 18639 22.3 6176 15.9 7038 88.3 19431 20.8
107.4 2.2
As you can see, the results are pretty close, except for the random
seeks.
Anyway, when available I also will offer you the results using the
Mylex ExtremeRaid and the 4 cheetahs in a raid0 configuration.
Please note that pure cache reads in my machine are about 200 Mb/sec,
and that
I use two LVD channels, each with two drives.
Regards
--
------------------------------------------------------------------
Oscar Fernández Cantero <[EMAIL PROTECTED]>
CEIFE - Centro Español de Investigación Farmacoepidemiológica
Almirante 28 2º http://www.ceife.es
28004 MADRID, SPAIN
Phone: +34 91 531 7975
Fax: +34 91 531 2871
-------------------------------------------------------------------
Oscar Fernández Cantero wrote:
>
> Hello,
>
> I have currently returned the card because of something wrong with
> the battery backup memory.
>
> As soon as I receive the replacement (one or two weeks),
> I'll perform the test you tell, and send the results to the list.
>
> Bye
>
> James Manning wrote:
> >
> > [ Friday, October 8, 1999 ] Oscar Fernández Cantero wrote:
> > > The machine has 512 Mb RAM and the Bonnie file size 512 Mb.
> > > No cache advantage involved.
> >
> > I don't think many would agree with you on this :)
> >
> > Could you print out runs with -s 512 and -s 2000, please?
> >
> > James
> > --
> > Miscellaneous Engineer --- IBM Netfinity Performance Development
>
