Newsgroups: comp.parallel.pvm
From: wkinning@nyx10.cs.du.edu (Warren Kinninger)
Subject: Re: PVM on Windows PCs?
Organization: University of Denver, Math/CS Dept.
Date: 2 Sep 1994 13:01:14 -0600
Message-ID: <347spq$krl@nyx10.cs.du.edu>

dfields@lehman.com (Doug Fields) writes:

>In any event, I don't have much of a point to make other than, "Rather than
>argue, accept as a given that PVM will be ported to Windows. Under this 
>premise, how would you go about doing this? Has anything like this been
>done? What are the possible pitfalls?"

Windows V4 (Chicago), due out early next year has some features that would
make an implementation of PVM easier. It's going to have support separate 
memory space for each process, provide some form of multitasking, and 
provide support for TCP/IP. Device drivers that run in real mode can 
still crash the machine, but standard applications are better protected. 
Programs using the Win32 API are run as threads by the kernel and can 
be preemptively multitasked. Unfortunately, some of the performance 
monitoring tools provided by Windows NT won't be provided with Windows 4.0. 
It would probably be better to port PVM to NT now and then move it to 4.0 later.
Most of this info is from the September Byte magazine.

Warren Kinninger
wkinning@nyx10.cs.du.edu


