Path: utzoo!attcan!uunet!nuchat!sugar!peter
From: peter@sugar.uu.net (Peter da Silva)
Newsgroups: comp.sys.amiga.tech
Subject: Re: IRQ virus (and a personal note to Steve)
Message-ID: <3206@sugar.uu.net>
Date: 4 Jan 89 01:03:24 GMT
References: <5601@cbmvax.UUCP> <5602@cbmvax.UUCP> <10788@s.ms.uky.edu> <10800@s.ms.uky.edu>
Organization: Sugar Land Unix - Houston, TX
Lines: 28

In article <10800@s.ms.uky.edu>, sean@ms.uky.edu (Sean Casey) writes:
> Real-time work not compatable with multiuser VM Systems? Tell it to Masscomp.

Masscomp uses an exotic dual-processor system to achieve real-time. Their
solution is not a general solution to the problem of real-time work under
UNIX. The same thing is true of DEC's real-time VAX software. On the other
hand you can do real realtime on a simple PDP-11 or 68000 if you don't take
on the load of memory management.

> I've been using Real Time Unix for three years. Tell it to NeXT. The NeXT
> box can multitask and DSP too. It's almost the nineteen nineties. Computers
> can walk and chew gum at the same time now.

Again, it's a dual processor, on a totally unloaded system. For 10 grand (the
cost of a useful NeXT system, if you can get one) I'd expect that much. Give
it a real task before you claim it's a realtime system.

CMU (who developed Mach) do not claim that it's a real-time system, though
it's got the right framework... we're evaluating it for real-time enhancement.
The two advantages that Mach has are the lightweight processes and the user
paging. I suspect that real-time tasks will have to be lightweight processes
and be locked into RAM. No fancy memory management for them, let alone VM.

Note that the cost of mach threads is about the same as the cost of Amiga
processes.
-- 
Peter "Have you hugged your wolf today" da Silva  `-_-'  Hackercorp.
...texbell!sugar!peter, or peter@sugar.uu.net      'U`