Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!sri-unix!ctnews!pyramid!decwrl!decvax!ucbvax!CITHEX.CALTECH.EDU!carl
From: carl@CITHEX.CALTECH.EDU
Newsgroups: comp.os.vms
Subject: Two Questions
Message-ID: <870424122836.006@CitHex.Caltech.Edu>
Date: Fri, 24-Apr-87 14:28:36 EDT
Article-I.D.: CitHex.870424122836.006
Posted: Fri Apr 24 14:28:36 1987
Date-Received: Sun, 26-Apr-87 22:39:20 EDT
Sender: daemon@ucbvax.BERKELEY.EDU
Distribution: world
Organization: The ARPA Internet
Lines: 19

Two unrelated questions (gee, it feels strange to be asking instead of
answering):
	1)  I came in this morning to find my 780 wouldn't respond to a
	    carriage return at my terminal by starting up loginout.  In
	    fact, it didn't seem to do anything.  On any terminal.  So,
	    I went to the CONSOLE, and repeated the sequenc ^P, H<CR>,
	    C<CR> several times, and found that the system seemed to be
	    looping at address 80008B1F.  I then forced a crash, ran
	    diagnostics (the system passed all of them), and rebooted. 
	    When I began analyzing the crash dump, the first thing I did,
	    of course, was a "SHOW SYMBOL/ALL", and lo and behold, I found
	    that 80008B1F is EXE$NULLPROC.  My question is:  Is EXE$NULLPROC
	    what I think it is, namely the place where VMS goes when nobody
	    wants to use the CPU?
	2)  I've got a user who wants to invert huge matrices (on the order
	    of 800x800, and who is worried that the algorithms used to invert
	    smaller matrices might be susceptible to cumulative errors in
	    these larger matrices.  Can anybody out there send me an algorithm
	    known to be robust in such situations?  Thanks in advance.