*BSD News Article 34333


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!agate!howland.reston.ans.net!pipex!sunic!news.kth.se!news.kth.se!d87-mal
From: d87-mal@byse.nada.kth.se (Mats Löfkvist)
Newsgroups: comp.os.386bsd.questions
Subject: What performance is normal with bt445s scsi controller and fast disk?
Date: 15 Aug 94 00:22:24
Organization: Royal Institute of Technology, Stockholm, Sweden
Lines: 22
Distribution: comp
Message-ID: <D87-MAL.94Aug15002224@byse.nada.kth.se>
NNTP-Posting-Host: byse.nada.kth.se
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

I have a bt445s (rev. C, firmware 3.36) and a fast disk (Fujitsu 2694esa,
capable of at least 3MB/s sustained) in a 486dx2/66 vlb box. I'm running
FreeBSD 1.1.5.1. Everything works like a charm, but I'm a bit disappointed
with the performance I get: iozone results are around 600kB/s write, 800kB/s
read with the kernel configured for aha scsi. With a bt kernel I get worse
write performance and better read performance (~4-500kB/s and ~1MB/s).

The bt driver says my card is isa ("bt54x-ISA(24bit) bus" instead of the
expected "PCI/EISA/VLB(32bit) bus"), is this a hint to a configuration problem?

I have tried changing the undocumented switches on the card, switch 2-10 in
the not default position makes the bt driver add "eisa dma" (*) and makes the
aha kernel not work. Switch 2-9 doesn't seem to make a difference.

      _
Mats Lofkvist
d87-mal@nada.kth.se


(*)  In the 32-bit bt driver (bt742a_32.c) this message is changed to
     "busmastering" which seems more right. Didn't make any noticeable
     difference in iozone results though.