From owner-freebsd-hackers Wed Mar 6 10:16:34 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id KAA10304 for hackers-outgoing; Wed, 6 Mar 1996 10:16:34 -0800 (PST) Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id KAA10298 for ; Wed, 6 Mar 1996 10:16:29 -0800 (PST) Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.12/8.6.9) id FAA32574; Thu, 7 Mar 1996 05:11:56 +1100 Date: Thu, 7 Mar 1996 05:11:56 +1100 From: Bruce Evans Message-Id: <199603061811.FAA32574@godzilla.zeta.org.au> To: hasty@rah.star-gate.com, lehey.pad@sni.de Subject: Re: Triton-II support... when? Cc: hackers@freebsd.org Sender: owner-hackers@freebsd.org Precedence: bulk >>> > Who uses ISA cards? >>> >>> How many people have really 0 ISA boards in their machine? >>> Considering that ISA places such a load on the bus, it seems a good >>> idea to me. >>> >> >> Oh, don't pay attention to Terry's complete dislike for ISA cards. >Well, I don't like them either. I was just curious about how many >people had got round to running systems with none. I have a Triton on I have one system with no ISA cards. However, the on-board i/o is mostly as slow as ISA: Times in usec for inb() from selected ports on ASUS P55TP4XE: min av max speed important for FreeBSD? ----- ----- ----- ---------------------------- 0x21 (pic0 mask) .448 .449 .451 yes 0x40 (timer counter 0) .702 .703 .705 yes (2.1), no (current) 0x43 (timer mode) 1.180 1.180 1.182 yes (2.1), no (current) 0x1f4 (fdc0 status) 1.180 1.180 1.182 no (< other fd slowness) 0x3f8 (sio0 data) 1.180 1.180 1.182 only if you use sio a lot Also, the on-board i/o (UMC8669F) is very buggy. The fdc interrupts at the wrong time (this at best causes busy waiting followed by timeouts to recover under FreeBSD) and the 16550 compatibles aren't (they lose sync when the speed is set or the fifo is changed while data is arriving. The problem is most noticeable for uccp because it does a lot of tcsetattr()s and tcsetattr() sets the speed even when it hasn't changed). The on-board i/o will have to be replaced by an ISA card unless/until acceptable workarounds are found :-(. Bruce