Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!rutgers!mit-eddie!uw-beaver!tektronix!cae780!leadsv!eps2!jon
From: jon@eps2.UUCP
Newsgroups: comp.unix.questions
Subject: Re: Apple Mac II
Message-ID: <81@eps2.UUCP>
Date: Wed, 15-Apr-87 02:39:41 EST
Article-I.D.: eps2.81
Posted: Wed Apr 15 02:39:41 1987
Date-Received: Fri, 17-Apr-87 00:15:53 EST
References: <173@perdix.lu.se> <3423@osu-eddie.UUCP>
Organization: Scumtronics Inc.
Lines: 40
Keywords: UNIX, virtual memory?
Summary: corrections?

In article <3423@osu-eddie.UUCP>, bob@osu-eddie.UUCP (Bob Sutterfield) writes:
> Apple has already done it.  The word is that it will be  on the street

My understanding was that Unisoft did it.

> speak).  They include signals, networking, and "selected utilities".

Yup.  I was told that they added all the BSD IPC stuff, with the sockets
interface.  I'm impressed.  Good job, Unisoft.

> It still has the SV file system - not the Berkeley  fast one, nor with
> long   names.   Expect  SVr3 streams,  but  (for now)  no  RFS nor mpx
> streams.

Yeah, 1K blocks.  Some customers have already expressed an interest in putting
the McKusick file system in A/UX.
 
> To run  A/UX (or Oreo or whatever  it's called today) you must install
> the  optional PMMU chip   - their own  Programmable Memory  Management
> Unit.

I believe that is a Motorola 68851, the *Paged* Memory Management Unit.

Regarding SCSIs without DMA, it would definitely suck if you had multiple
users and/or multiple processes running.  But, on a single-user system
with just one process running, it might be acceptable, because the only
program that could run is sleeping waiting for I/O to complete anyway;
so you could tie up the CPU polling for data from the SCSI host adapter.
Plessey had a 1K FIFO on one of their CPU boards that has an NCR5386
SCSI host adapter.  It generated interrupts when it was half full, then
the CPU would come along and read 512 bytes out.  I've never used it, but
it's possible that the disk controller could transfer continuously into
the FIFO for many consecutive disk sectors, because the CPU would start
to empty it quickly when it got half full.  The penalty you pay is moving
bytes around with the CPU, but with interrupts from the FIFO you don't
waste time polling for data.



Jonathan Hue	DuPont Design Technologies/Via Visuals		leadsv!eps2!jon