Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 6 Feb 2001 06:57:48 +0000 (GMT)
From:      Terry Lambert <tlambert@primenet.com>
To:        dillon@earth.backplane.com (Matt Dillon)
Cc:        tlambert@primenet.com (Terry Lambert), phk@critter.freebsd.dk (Poul-Henning Kamp), gibbs@scsiguy.com (Justin T. Gibbs), bright@wintelcom.net (Alfred Perlstein), rjesup@wgate.com (Randell Jesup), mjacob@feral.com (Matthew Jacob), msmith@FreeBSD.ORG (Mike Smith), des@ofug.org (Dag-Erling Smorgrav), dnelson@emsphone.com (Dan Nelson), tanimura@r.dl.itc.u-tokyo.ac.jp (Seigo Tanimura), arch@FreeBSD.ORG
Subject:   Re: Bumping up {MAX,DFLT}*PHYS (was Re: Bumping up {MAX,DFL}*SIZ in i386)
Message-ID:  <200102060657.XAA12458@usr08.primenet.com>
In-Reply-To: <200102060546.f165kHq58398@earth.backplane.com> from "Matt Dillon" at Feb 05, 2001 09:46:17 PM

next in thread | previous in thread | raw e-mail | index | archive | help
>    I think the idea Poul originally articulated -- having simple information
>    like recommended I/O size, recommended cluster size, and/or maximum I/O
>    size, is the correct solution.  Getting fancy might buy us a percent
>    or two... it isn't worth the effort.

I thought Poul had discarded that idea as unworkable, after having
tried to make it work; I got the impression that he still liked
the idea, but that he didn't have a way to make it practical (Poul,
please correct me if I am misinterpreting your last post).

I can't see hints being much more useful than the seek optimization
code, which was disabled as a pessimization for most ZBR drives,
where the track boundaries were unknown back in the early fictional
geometry days (predating SCSI II, where it could be fixed again).

I would think that you would want your optimization to work at
least 51% of the time for it to be worthwhile, or at least "mostly
harmless", and I really have doubts that "hints" would be able to
do that.

You really don't want to end up with something that makes a
microbenchmark run fast, at the expense of real system loads,
like some of the stuff that happened in the buffer cache code,
historically.


					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-arch" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200102060657.XAA12458>