📄 autobook_116.html
字号:
<HTML><!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"><!-- Created on September, 12 2004 by texi2html 1.64 --><!-- Written by: Lionel Cons <Lionel.Cons@cern.ch> (original author) Karl Berry <karl@freefriends.org> Olaf Bachmann <obachman@mathematik.uni-kl.de> and many others.Maintained by: Olaf Bachmann <obachman@mathematik.uni-kl.de>Send bugs and suggestions to <texi2html@mathematik.uni-kl.de> --><HEAD><TITLE>Autoconf, Automake, and Libtool: C Floating Point</TITLE><META NAME="description" CONTENT="Autoconf, Automake, and Libtool: C Floating Point"><META NAME="keywords" CONTENT="Autoconf, Automake, and Libtool: C Floating Point"><META NAME="resource-type" CONTENT="document"><META NAME="distribution" CONTENT="global"><META NAME="Generator" CONTENT="texi2html 1.64"><script language="Javascript"><!-- // Check the browser version. function checkVersion() { if (navigator.appVersion.charAt(0)>=3) return true; if (navigator.appVersion.charAt(0)>=4) return true; else return false; } if (checkVersion()) { homeon = new Image(); homeon.src = "homeon.png"; homeoff = new Image(); homeoff.src = "home.png"; tocon = new Image(); tocon.src = "tocon.png"; tocoff = new Image(); tocoff.src = "toc.png"; indexon = new Image(); indexon.src = "indexon.png"; indexoff = new Image(); indexoff.src = "index.png"; helpon = new Image(); helpon.src = "helpon.png"; helpoff = new Image(); helpoff.src = "help.png"; backon = new Image(); backon.src = "backon.png"; backoff = new Image(); backoff.src = "back.png"; forwardon = new Image(); forwardon.src = "forwardon.png"; forwardoff = new Image(); forwardoff.src = "forward.png"; prevon = new Image(); prevon.src = "prevon.png"; prevoff = new Image(); prevoff.src = "prev.png"; nexton = new Image(); nexton.src = "nexton.png"; nextoff = new Image(); nextoff.src = "next.png"; upon = new Image(); upon.src = "upon.png"; upoff = new Image(); upoff.src = "up.png"; } function img_act(imgName) { if (checkVersion()) { imgOn = eval(imgName + "on.src"); document [imgName].src = imgOn; } } function img_inact(imgName) { if (checkVersion()) { imgOff = eval(imgName + "off.src"); document [imgName].src = imgOff; } }// --></SCRIPT></HEAD><BODY LANG="EN" BGCOLOR="#FFFFFF" TEXT="#000000" LINK="#6688AA" VLINK="#336688" ALINK="#808080"><A NAME="SEC116"></A><TABLE BORDER=0 CELLPADDING=0 CELLSPACING=10><TR VALIGN="TOP"><TD ALIGN="MIDDLE" WIDTH=50 BGCOLOR="#e6e6e6"><TABLE CELLPADDING=1 CELLSPACING=1 BORDER=0><TR VALIGN="TOP" ALIGN="LEFT"><TD VALIGN="MIDDLE" ALIGN="LEFT"><A HREF="autobook_115.html#SEC115" onMouseover="img_act('prev')" onMouseout="img_inact('prev')"><IMG SRC="prev.png" BORDER="0" ALT="Back: C Structure Layout" ALIGN="MIDDLE" NAME="prev"></A></TD></TR><TR VALIGN="TOP" ALIGN="LEFT"><TD VALIGN="MIDDLE" ALIGN="LEFT"><A HREF="autobook_117.html#SEC117" onMouseover="img_act('next')" onMouseout="img_inact('next')"><IMG SRC="next.png" BORDER="0" ALT="Forward: GNU cc Extensions" ALIGN="MIDDLE" NAME="next"></A></TD></TR><TR VALIGN="TOP" ALIGN="LEFT"><TD VALIGN="MIDDLE" ALIGN="LEFT"> <TR VALIGN="TOP" ALIGN="LEFT"><TD VALIGN="MIDDLE" ALIGN="LEFT"><A HREF="autobook_117.html#SEC117" onMouseover="img_act('back')" onMouseout="img_inact('back')"><IMG SRC="back.png" BORDER="0" ALT="FastBack: GNU cc Extensions" ALIGN="MIDDLE" NAME="back"></A></TD></TR><TR VALIGN="TOP" ALIGN="LEFT"><TD VALIGN="MIDDLE" ALIGN="LEFT"><A HREF="autobook_111.html#SEC111" onMouseover="img_act('up')" onMouseout="img_inact('up')"><IMG SRC="up.png" BORDER="0" ALT="Up: C Language Portability" ALIGN="MIDDLE" NAME="up"></A></TD></TR><TR VALIGN="TOP" ALIGN="LEFT"><TD VALIGN="MIDDLE" ALIGN="LEFT"><A HREF="autobook_118.html#SEC118" onMouseover="img_act('forward')" onMouseout="img_inact('forward')"><IMG SRC="forward.png" BORDER="0" ALT="FastForward: Cross-Unix Portability" ALIGN="MIDDLE" NAME="forward"></A></TD></TR><TR VALIGN="TOP" ALIGN="LEFT"><TD VALIGN="MIDDLE" ALIGN="LEFT"><A HREF="autobook.html#SEC_Top" onMouseover="img_act('home')" onMouseout="img_inact('home')"><IMG SRC="home.png" BORDER="0" ALT="Top: Autoconf, Automake, and Libtool" ALIGN="MIDDLE" NAME="home"></A></TD></TR><TR VALIGN="TOP" ALIGN="LEFT"><TD VALIGN="MIDDLE" ALIGN="LEFT"><A HREF="autobook_toc.html#SEC_Contents" onMouseover="img_act('toc')" onMouseout="img_inact('toc')"><IMG SRC="toc.png" BORDER="0" ALT="Contents: Table of Contents" ALIGN="MIDDLE" NAME="toc"></A></TD></TR><TR VALIGN="TOP" ALIGN="LEFT"><TD VALIGN="MIDDLE" ALIGN="LEFT"><A HREF="autobook_285.html#SEC285" onMouseover="img_act('index')" onMouseout="img_inact('index')"><IMG SRC="index.png" BORDER="0" ALT="Index: Index" ALIGN="MIDDLE" NAME="index"></A></TD></TR><TR VALIGN="TOP" ALIGN="LEFT"><TD VALIGN="MIDDLE" ALIGN="LEFT"><A HREF="autobook_abt.html#SEC_About" onMouseover="img_act('help')" onMouseout="img_inact('help')"><IMG SRC="help.png" BORDER="0" ALT="About: About this document" ALIGN="MIDDLE" NAME="help"></A></TD></TR></TABLE></TD><TD ALIGN="LEFT"><H3> 15.1.5 C Floating Point </H3><!--docid::SEC116::--><P>Most modern systems handle floating point following the IEEE-695standard. However, there are still portability issues.</P><P>Most processors use 64 bits of precision when computing floating pointvalues. However, the widely used Intel x86 series of processors computetemporary values using 80 bits of precision, as do most instances of theMotorola 68k series. Some other processors, such as the PowerPC,provide fused multiply-add instructions which perform a multiplicationand an addition using high precision for the intermediate value.Optimizing compilers will generate such instructions based on sequencesof C operations.</P><P>For almost all programs, these differences do not matter. However, forprograms which do intensive floating point operations, the differencescan be significant. It is possible to write floating point loops whichterminate on one sort of processor but not on another.</P><P>Unfortunately, there is no rule of thumb that can be used to avoid theseproblems. Most compilers provide an option to disable the use ofextended precision (for GNU cc, the option is<SAMP>`-ffloat-store'</SAMP>). However, on the one hand, this merely shifts theportability problem elsewhere, and, on the other, the extended precisionis often good rather than bad. Although these portability problems cannot be easily avoided, you should at least be aware of them if you writeprograms which require very precise floating point operations.</P><P>The IEEE-695 standard specifies certain flags which the floating pointprocessor should make available (e.g., overflow, underflow, inexact),and specifies that there should be some control over the floating pointrounding mode. Most processors make these flags and controls available;however, there is no portable way to access them. A portable programshould not assume that it will have this degree of control over floatingpoint operations.</P><P><A NAME="GNU cc Extensions"></A></TR></TABLE><BR> <FONT SIZE="-1">This document was generatedby <I>Gary V. Vaughan</I> on <I>September, 12 2004</I>using <A HREF="http://www.mathematik.uni-kl.de/~obachman/Texi2html"><I>texi2html</I></A></BODY></HTML>⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -