Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site angband.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!genrad!teddy!panda!talcott!harvard!seismo!ut-sally!mordor!angband!sjc
From: sjc@angband.UUCP (Steve Correll)
Newsgroups: net.lang.f77
Subject: Re: Equivalence problem
Message-ID: <44@angband.UUCP>
Date: Thu, 24-Jan-85 21:00:51 EST
Article-I.D.: angband.44
Posted: Thu Jan 24 21:00:51 1985
Date-Received: Sun, 27-Jan-85 05:12:30 EST
Distribution: net
Organization: S-1 Project, LLNL
Lines: 30

Here's an even simpler demonstration of the phenomenon:

	IMPLICIT DOUBLE PRECISION (A-H,O-Z)
	DIMENSION W(10),NW(2)
	EQUIVALENCE (WW,NW(1)),(NQ,NW(2))
C MEMORY MAP OF WW, NW AND NQ
C WORD:		1	2
C WW	 (..............)
C NW     (...1..)(...2..)
C NQ		 (......)
C
	ww=0.0
	NQ=10
	write(*,*)'ww,nw(1),nw(2),nq',ww,nw(1),nw(2),nq
	temp=ww
	ww=temp
	write(*,*)'ww,nw(1),nw(2),nq',ww,nw(1),nw(2),nq
	end

The output is:

  ww,nw(1),nw(2),nq  0.  0  10  10
  ww,nw(1),nw(2),nq  0.  0  0  0

The "-S" option of f77 shows that it uses the "movd" instruction to transfer
"ww" to "temp" and back again. I suspect that this normalizes the floating
point value, bashing your data.
-- 
                                                           --Steve Correll
sjc@s1-c.ARPA, ...!decvax!decwrl!mordor!sjc, or ...!ucbvax!dual!mordor!sjc