Return to BSD News archive
Newsgroups: comp.os.386bsd.questions Path: sserve!newshost.anu.edu.au!munnari.oz.au!metro!ipso!runxtsa!bde From: bde@runx.oz.au (Bruce Evans) Subject: Re: Info on writing disktab entries Message-ID: <1993Mar24.061030.16280@runx.oz.au> Keywords: disktab bad144 Organization: RUNX Un*x Timeshare. Sydney, Australia. References: <1993Mar23.090648.3203@news.cs.indiana.edu> <1993Mar23.160453.4278@coe.montana.edu> Date: Wed, 24 Mar 93 06:10:30 GMT Lines: 20 In article <1993Mar23.160453.4278@coe.montana.edu> osyjm@cs.montana.edu (Jaye Mathisen) writes: >In article <1993Mar23.090648.3203@news.cs.indiana.edu> "Michael Squires" <mikes@cs.indiana.edu> writes: >> :pg#585120:og#122430:tg=4.2BSD:bg#4096:fg#512:\ >> :ph#599430:oh#707550:th=4.2BSD:bh#4096:fh#512: > >With 4k/512byte, you will get abysmal performance, you might want to run some >benchmarks and get the fastest numbers you can (in conjunction with tunefs). Probably not. The 0.1 wd (ESDI and MFM) driver tends to give equally abysmal performance for all block sizes. On my system (486/33, WD1007V-SE/1 controller; drive 0 with 35 sectors/track (transfer rate 10Mb/sec); drive 1 with 54 sectors/track (transfer rate 15Mb/sec); drive 1 has to be formatted at 2:1 interleave for the controller to keep up), the driver is too slow at issueing successive i/o's for tunefs to be of any use. Best results are given by the default tuning `tunefs -a 1 -d 4'. Rotatational delays of 1 to 3 msec are equivalent to 4 msec. When the block size is doubled, the gap between sucessive blocks has to be doubled so the speed is not improved. -- Bruce Evans bde@runx.oz.au