Path: utzoo!mnetor!uunet!husc6!cmcl2!nrl-cmf!ames!claris!apple!baum
From: baum@apple.UUCP (Allen J. Baum)
Newsgroups: comp.arch
Subject: Re: The WM Machine
Message-ID: <7552@apple.UUCP>
Date: 5 May 88 16:15:38 GMT
References: <5339@aw.sei.cmu.edu> <6762@lll-winken.llnl.gov>
Reply-To: baum@apple.UUCP (Allen Baum)
Organization: Apple Computer, Inc.
Lines: 26

--------
[]
>In article <6762 (Eugene D. Brooks III) writes:
>I thought I would add a little technical support to a couple of the points.

Me, too.

It seems to me that it is entirely likely that a stream operation will get
interrupted, by a page fault, if nothing else. Saving and restoring is a
pain, and unlikely to be done without some very special instructions, or
a bunch of 'CISC' like "save the state of the machine on the ?stack?"
operations. Furthermore, if I have a two operand stream operation (both
fifo's busy), then I'm locked out of memory until the streams are finished!

Overall, I see very little that a VLIW machine couldn't, including the
stream operations, which would probably be trivial in a VLIW. The WM has,
as far as I can see, only the virtue of being compact, and it isn't very
clear that all the extra instructions you need to solve some of its problems
won't expand the instruction stream to make that marginal as well.

Since a lot of the criticisms are fairly obvious, it could be that we just
don't know all the details. Wulf has been around a while, and must have
considered some of these. Maybe we'll hear from him, and find out what the
real story is.

--
{decwrl,hplabs,ihnp4}!nsc!apple!baum		(408)973-3385