Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!yarrina.connect.com.au!munnari.OZ.AU!news.hawaii.edu!ames!usenet.kornet.nm.kr!news.kreonet.re.kr!usenet.seri.re.kr!news.imnet.ad.jp!lab!wsclark!hu-eos-news!hiroshi From: hiroshi@teine.chem2.hokudai.ac.jp (Hiroshi Murakami) Newsgroups: comp.unix.bsd.freebsd.misc Subject: Aligned or not aligned that is the problem. Date: 19 Nov 1995 00:47:26 GMT Organization: Hokkaido Univ., Sapporo, Japan. Lines: 30 Message-ID: <48luqu$c2f@nyx.eos.hokudai.ac.jp> NNTP-Posting-Host: teine.chem2.hokudai.ac.jp I wish to know how the double 8 byte data will be allocated for different installation of versions of FreeBSS with the combination of the versions of gccs. Let try to compile the following sample source code % gcc -v a.c And run it to show if the double variable is out of 8 byte aligned or not. For pentium, it seems every load/store/fp-operation of double data about twice slower than the 8 byte aligned case. When you wish upon the Pentium so that your computation will be fast than not just come true, you must align you double or you waste your machine. /*-----------------------*/ double a[100]; static double b[100]; main() { double c[100]; int i; double d[100]; printf("Octal address: a:%o, b:%o, c:%o, d:%o\n", (unsigned)a,(unsigned)b,(unsigned)c,(unsigned)d); } /*-----------------------*/