From owner-freebsd-questions Thu May 18 10:28:41 1995 Return-Path: questions-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id KAA13431 for questions-outgoing; Thu, 18 May 1995 10:28:41 -0700 Received: from cs.weber.edu (cs.weber.edu [137.190.16.16]) by freefall.cdrom.com (8.6.10/8.6.6) with SMTP id KAA13425 for ; Thu, 18 May 1995 10:28:39 -0700 Received: by cs.weber.edu (4.1/SMI-4.1.1) id AA15601; Thu, 18 May 95 11:21:44 MDT From: terry@cs.weber.edu (Terry Lambert) Message-Id: <9505181721.AA15601@cs.weber.edu> Subject: Re: SCSI boards and then some. To: SahagunS@aol.com Date: Thu, 18 May 95 11:21:43 MDT Cc: Questions@FreeBSD.org In-Reply-To: <950518023555_122918958@aol.com> from "SahagunS@aol.com" at May 18, 95 02:35:55 am X-Mailer: ELM [version 2.4dev PL52] Sender: questions-owner@FreeBSD.org Precedence: bulk > I'm glad you addressed the issue of performance of the Sound Blaster SCSI > board. I had been curious as to what one might expect from such a 'do all' > card. The Adaptec controller with built in sound *will* do bus mastering. > When you refered to 'bus mastering', i'm assuming that you were refering > to the SCSI bus and not the PC bus. No, I was referring to the PC bus. A bus mastering controller (like *most* SCSI controllers and unlike *most* IDE/EIDE) offloads some of the processing from the main processer by doing the memory transfers itself. The salient point is that low quality processors and motherboards may have a problem with it, but if they don't, you will get a significant performance boost if your OS can handle async I/O (ie: it is a preemptive multitasker, unlike DOS or Windows pre Win95). > Is it safe to assume that the 'tick' time a task gets is adjustable and > not based on the PC's 18 times a second limit? The amount of time a task gets is based on two things: it's quantum and whether it requests a "slow" operation. A task gives away the processer voluntarily (a "voluntary context switch") when it performs an operation that can take place asynchronously but which it requires the results of to continue processing. Like a disk read. A task has the processer taken away from it (an "involuntary context switch") when the process has used its entire processing "quantum". This "quantum" is what you called "the 'tick' time a task gets". Without RealTime extensions, the tick time is *not* adjustable. However, since the context switch in this case is involuntary, the machine will be able to more fairly divide the CPU among a number of tasks. For instance, in Windows, if a process does not volunarily yield the processor, all other tasks will hang except the task running until it does. > Do you know of any 'visual' X tools, i.e. ones that allow you to > interactively design a window and all its widgets? There are many of them. There is currently a thread on this topic in the news group comp.windows.x.motif, which is probably the best place to ask, although there are people who advocate tcl/tkl based soloutions; I'm not sure about Warner Losh's current opinion, but Object-Interface Builder from PARC Place software will let you build apps as well. > And finally, what is NetBSD? A BSD 4.4 derived OS that runs on several platforms. Terry Lambert terry@cs.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers.