Date: Wed, 24 Sep 1997 16:00:22 +0200 From: Stefan Esser <se@FreeBSD.ORG> To: Terry Lambert <tlambert@primenet.com> Cc: dyson@FreeBSD.ORG, syssgm@dtir.qld.gov.au, freebsd-current@FreeBSD.ORG Subject: Re: New timeout capability (was Re: cvs commit:....) Message-ID: <19970924160022.50021@mi.uni-koeln.de> In-Reply-To: <199709240617.XAA04899@usr07.primenet.com>; from Terry Lambert on Wed, Sep 24, 1997 at 06:17:23AM %2B0000 References: <199709230920.EAA00190@dyson.iquest.net> <199709240617.XAA04899@usr07.primenet.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sep 24, Terry Lambert <tlambert@primenet.com> wrote: > > I think that 4K performs pretty darned well anyway though. In the > > real world, I wouldn't think that one would see much of a performance > > difference between 4K and 16K. > > For 8k, there used to be about a 40% improvement over 4k for iozone; I > haven't really tried this for about 5 moths now, though. Hmmm, did you measure accesses to the char device, block device, or a file ??? If you used the character device, then the difference is due to processing overhead in the drive. But because of the cluster code, requests for small data transfers are only sent to the drive if the data is fragmented, both with 4KB and larger block sizes ... Accesses to the block device should be mostly independent of the request size. 2KB requests are sent to the drive in any case ... I'd be very surprised, if your IOZONE results were different by 40% for file accesses (were clustering helps). Regards, STefan
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19970924160022.50021>