Date: Wed, 15 Mar 1995 10:42:19 -0800 (PST) From: "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com> To: phk@ref.tfs.com (Poul-Henning Kamp) Cc: current@FreeBSD.org Subject: Re: newfs: sectors per cylinder (4096) disagrees with disk label (36) Message-ID: <199503151842.KAA13193@gndrsh.aac.dev.com> In-Reply-To: <199502210226.SAA21140@ref.tfs.com> from "Poul-Henning Kamp" at Feb 20, 95 06:26:18 pm
next in thread | previous in thread | raw e-mail | index | archive | help
>
> > > > This is caused by phk's changes to the behavior of newfs, it now
> > > > uses 1 head/cylinder, 4096 sectors/track by default. This was an
> > > > attempt to increase performace. I have found that it does nothing
> > > > for any disks I use and just leads to problems while newfsing several
> > > > of them.
> > > How did you measure this ?
> >
> > Iozone using auto and 128 8192 and dd to/from /dev/{null,zero}.
> > Note that the newfs code already defeats the rotational delay stuff
> > by -n 1 -d 0 defaults.
> >
> Try this:
>
> newfs /dev/rsdXy
> mount /dev/sdXy /mnt
> (cd /usr/src ; find . -print | cpio -dumpV /mnt)
> time tar cf /dev/null /mnt
> umount /mnt
I have run several iterations of this using the default newfs and
using the -t 0 -u 0 overrides and can find no significant difference
now that the ufs code has been fixed to deal with NRPOS=1.
If no one objects (more specifically if Poul-Henning does not object) I
am going to removed the newfs hacks for 1 track of 4096 cylinders. And
fix src/etc/Makefile to not have the -t 0 -u 0.
--
Rod Grimes rgrimes@gndrsh.aac.dev.com
Accurate Automation Company Custom computers for FreeBSD
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199503151842.KAA13193>