From owner-freebsd-hardware Fri Aug 23 15:36:46 1996 Return-Path: owner-hardware Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id PAA21199 for hardware-outgoing; Fri, 23 Aug 1996 15:36:46 -0700 (PDT) Received: from FileServ1.MI.Uni-Koeln.DE (FileServ1.MI.Uni-Koeln.DE [134.95.212.1]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id PAA21190 for ; Fri, 23 Aug 1996 15:36:42 -0700 (PDT) Received: from x14.mi.uni-koeln.de (annexr3-16.slip.Uni-Koeln.DE) by FileServ1.MI.Uni-Koeln.DE with SMTP id AA08798 (5.67b/IDA-1.5 for ); Sat, 24 Aug 1996 00:35:51 +0200 Received: (from se@localhost) by x14.mi.uni-koeln.de (8.7.5/8.6.9) id AAA22927; Sat, 24 Aug 1996 00:34:51 +0200 (MET DST) Date: Sat, 24 Aug 1996 00:34:51 +0200 (MET DST) Message-Id: <199608232234.AAA22927@x14.mi.uni-koeln.de> From: Stefan Esser To: Michael Smith Cc: ambrisko@tcsi.com (Douglas Ambrisko), gibbs@freefall.freebsd.org, richard@pegasus.com, freebsd-hardware@freebsd.org Subject: Re: SCSI again: Asus SC200 vs. Adaptec 2940 ... In-Reply-To: <199608230117.KAA17650@genesis.atrad.adelaide.edu.au> References: <199608221630.JAA10862@cozumel.tcs.com> <199608230117.KAA17650@genesis.atrad.adelaide.edu.au> Sender: owner-hardware@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > Douglas Ambrisko stands accused of saying: > > > > | > > | It does have the entire program in ram as do some of the newer NCR (or > > | should I say symbios?) cards. > > > > Does the driver have to change to take advantage of the new NCR cards with > > on-board ram? If it does has it been done? > > No, and yes respectively. No, sorry. It's actually "Yes and No" ... The driver detects the PCI IDs of the newer cards. By now I've got the data books for the 53c810A, 825A, 860 and the 875, and I've actually read them :) Each of them supports a different combination of features, and the only way to support all of them is to completely redesign the way the SCRIPTS code that is executed by the NCR is supplied with the driver. A command might only be available as a sequence of instructions in the old chips, and the best way to deal with this is to have the NCR code be generated at boot time from a compact higher level description (a byte code). The design of all this is done, and the current "firmware" converted to the new syntax, but I have not yet finished the code generator and the dynamic linker that combines multiple independent fragments into a running NCR program. I could enable instruction prefetch and a few other features based on the information in the data books, but do not intend to do so, as long as I had no way to test the effects on my system. For this reason I do not intend to release any code, before I had a chance to see it run on my still to arrive new system, which will come with a 53c875 ... I'm thinking about buying another 53c810A based card in order to have both types (enhanced vs. enhanced + ULTRA + WIDE + RAM) for testing in one machine. But there has been an announcement of a driver for the new Tekram SCSI cards based on the 53c875. I had no chance to study the driver sources, but the author claims that the current NCR driver in FreeBSD does not correctly initialize that card. I suppose there is a problem with the (unknown to me) clock oscillator frequency used on that card, and it will be easy to fix. I can't tell by now, how good that driver is, but I'll have a look at the sources soon ... Regards, STefan