Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!seismo!lll-lcc!ames!ucbcad!ucbvax!roger@seismo.CSS.GOV@celtics.UUCP From: roger@seismo.CSS.GOV@celtics.UUCP Newsgroups: mod.computers.vax Subject: Submission for mod-computers-vax Message-ID: <8701072104.AA08348@seismo.CSS.GOV> Date: Wed, 7-Jan-87 16:01:36 EST Article-I.D.: seismo.8701072104.AA08348 Posted: Wed Jan 7 16:01:36 1987 Date-Received: Thu, 8-Jan-87 00:50:39 EST Sender: daemon@ucbvax.BERKELEY.EDU Organization: The ARPA Internet Lines: 30 Approved: info-vax@sri-kl.arpa Path: celtics!roger From: roger@celtics.UUCP (Roger Klorese) Newsgroups: mod.computers.vax Subject: Re: Null arguments in VMS Fortran Message-ID: <1426@celtics.UUCP> Date: 7 Jan 87 21:01:35 GMT References: <8701051538.AA09457@ucbvax.Berkeley.EDU> Reply-To: roger@celtics.UUCP (Roger Klorese) Organization: Celerity Computing (Eastern Region), Framingham, MA Lines: 19 In article <8701051538.AA09457@ucbvax.Berkeley.EDU> writes: >In the example you give, it is highly unlikely that any optimizer would choose >to put b in a register (or even copy it) - with only a single reference to b >itself, it costs more to load up the register than can be saved by having b >available there. Well, yes and no. On a VAX, you're right. In a system based on a register-to-register orientation, or a system which uses the "sliding-window" memory model, you're wrong. If it can only be accessed through a register anyway, it makes sense to pre-load it. In fact, even on this model (which Celerity systems follow), we've implemented optional parameters. We pass a predefined and illegal address for the omitted arguments. -- ///==\\ (No disclaimer - nobody's listening anyway.) /// Roger B.A. Klorese, Celerity Computing (Northeast Area) \\\ 40 Speen St., Framingham, MA 01701 +1 617 872-1552 \\\==// celtics!roger@seismo.CSS.GOV - seismo!celtics!roger