Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1 6/24/83 (MC840302); site turing.UUCP Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!godot!harvard!seismo!mcvax!turing!dik From: dik@turing.UUCP Newsgroups: net.lang.c Subject: Re: C language hacking Message-ID: <227@turing.UUCP> Date: Fri, 30-Nov-84 13:43:19 EST Article-I.D.: turing.227 Posted: Fri Nov 30 13:43:19 1984 Date-Received: Sun, 2-Dec-84 04:05:52 EST References: <5942@brl-tgr.ARPA> Organization: CWI, Amsterdam Lines: 20 Apparently-To: rnews@mcvax.LOCAL >The reason why this matters should be pretty obvious to anyone who >does graphics for a living. In that case it is simple. You do not need large parts of the fancy stuff executed by standard sine and cosine routines; e.g. you will not need a 16 digit precision (as delivered by the standard Vax Unix sine and cosine). Moreover, you may make quite sure that the argument to these functions is within a range like [-pi,pi], so range reduction is also much simpler to do. It is in general a pain that all standard functions in C deliver a result in double precision (what?, oh yes, the attempt is made) although you need it in single precision, which results in a waste of time. -- dik t. winter centrum voor wiskunde en informatica postbus 4079 1009 AB amsterdam nederland +31 20 592 4102 (polish your dutch) UUCP: decvax!mcvax!turing!dik