*BSD News Article 31886


Return to BSD News archive

Newsgroups: comp.os.386bsd.questions
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!agate!usenet.ins.cwru.edu!lerc.nasa.gov!purdue!mozo.cc.purdue.edu!staff.cc.purdue.edu!bj
From: bj@staff.cc.purdue.edu (Ben Jackson)
Subject: Re: How to get better SCSI performance? [FreeBSD 1.1R/AHA1542Cf]
Sender: news@mozo.cc.purdue.edu (USENET News)
Message-ID: <CrqHA5.Mo2@mozo.cc.purdue.edu>
Date: Tue, 21 Jun 1994 06:13:17 GMT
References: <CrKBr2.8oJ@mozo.cc.purdue.edu> <Z2zQYYR.dysonj@delphi.com> <RSANDERS.94Jun19200651@hrothgar.mindspring.com> <x4wyIGv.dysonj@delphi.com>
Organization: Purdue University
Lines: 30

In article <x4wyIGv.dysonj@delphi.com>, John Dyson  <dysonj@delphi.com> wrote:
>Robert Sanders <rsanders@mindspring.com> writes:
> 
>>Um, well, if he's getting 1/2 disk throughput with his filesystem,
>>he's not getting his money's worth.  [...]

As the one with the slow drives, I agree!

>Versions of *BSD w/o clustering will probably give only 1/2 of disk
>performance on writes.  I *normally* get .95 or better of my drivers
>performance thru the filesystem on FreeBSD on both reads and writes.

That's one of the things that makes me suspicious -- I get the same speed
reading and writing, which suggests that I'm not even approaching the max
possible throughput.  With the card doing DMA at 8M/s, and drives capable
of 5M/s sustained transfers, I see no reason why I shouldn't get more than
650k/s.

>Apparently there is a difference in his drives configuration.

What is *your* configuration?  I believe you asked earlier what adapter
I was using.  As the subject says, it's an AHA1542Cf.  The original
message had more details about the system and SCSI chain, but the
important info is:  486/66 w/16M, AHA1542Cf with a SCSI1 Maxtor, SCSI2
Quantum and SCSI2 NEC CDROM drive.  On transfers that aren't all in the
buffer cache, I get about 650K/s.

Still soliciting suggestions,

--Ben Jackson