[3865] in linux-scsi channel archive
Re: [BACK ON TOPIC] SCSI & IDE (fwd)
daemon@ATHENA.MIT.EDU (Bill Broadley)
Thu May 7 20:44:44 1998
From: Bill Broadley <bill@math.UCDavis.edu>
To: linux-scsi@vger.rutgers.edu, linux-smp@vger.rutgers.edu
Date: Thu, 7 May 1998 17:41:01 -0700 (PDT)
>I currently have a 4 processor P6/200 machine with UW scsi disks. I
>recently had a dual-processor PII/300, but with IDE disks. We ran an
>application on both that had a lot of computation, plus a lot of file
>I/O. the SCSI system toasts the IDE system pretty badly. But for a
>single-user machine, IDE isn't awful, just not as good as SCSI if you
>have multiple disks...
There are many variables here:
Which parameters were you using with hdparm?
Did you enable busmastering/dma?
Were the block sizes the same on all disks?
Which partition were you using (speed varies with where you
are on the disk).
Were you purely disk bound? Or are you seeing a difference between cpu's?
4 p6-200' are going to be alot faster then a dual pII-300 in alot of
things, thats no surprise.
How much memory on each system?
Where kernel's?
Were all the disks freshly formatted? Otherwise the layout of the files
could largely effect the performance (both position and how contiguous
they are).
Which disks were involved on each? Same rpm? Same brand?
Machines A "toasting" machine B is a useless piece of information I've
experienced the opposite as measured by bonnie. If you could
completely specify all the hardware involved and quantify your
results then maybe we can figure out the performance differences
and why they happened.
--
Bill Broadley Bill@math.ucdavis.edu UCD Math Sys-Admin
Linux is great. http://math.ucdavis.edu/~bill PGP-ok
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
