Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/18/84; site harvard.ARPA Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!think!harvard!stew From: stew@harvard.ARPA (Stew Rubenstein) Newsgroups: net.lang.c Subject: Re: pointer alignment on a Vax 780 Message-ID: <214@harvard.ARPA> Date: Wed, 26-Jun-85 02:04:42 EDT Article-I.D.: harvard.214 Posted: Wed Jun 26 02:04:42 1985 Date-Received: Fri, 28-Jun-85 00:17:12 EDT References: <478@crystal.UUCP> Distribution: net Organization: Aiken Computation Laboratory, Harvard Lines: 12 > Are there any alignment constraints on the Vax 780? > Could one, for example, set an integer pointer to > the middle of a machine word? > > If there aren't alignment constraints, is it faster > to compare two word-aligned integers? There are no constraints, but there is definitely a speed penalty for fetching non-aligned data. longwords should be aligned on longword boundaries, quadwords on quadwords, etc., for best performance. I have even heard of differences in benchmark timing based on the alignment of the instruction stream, though this is harder to predict and control.