1<html> 2<head> 3<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> 4<title>The Overhead in the Number Class Wrapper</title> 5<link rel="stylesheet" href="../../multiprecision.css" type="text/css"> 6<meta name="generator" content="DocBook XSL Stylesheets V1.79.1"> 7<link rel="home" href="../../index.html" title="Chapter 1. Boost.Multiprecision"> 8<link rel="up" href="../perf.html" title="Performance Comparison"> 9<link rel="prev" href="../perf.html" title="Performance Comparison"> 10<link rel="next" href="realworld.html" title="Floating-Point Real World Tests"> 11</head> 12<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF"> 13<table cellpadding="2" width="100%"><tr> 14<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../../boost.png"></td> 15<td align="center"><a href="../../../../../../index.html">Home</a></td> 16<td align="center"><a href="../../../../../../libs/libraries.htm">Libraries</a></td> 17<td align="center"><a href="http://www.boost.org/users/people.html">People</a></td> 18<td align="center"><a href="http://www.boost.org/users/faq.html">FAQ</a></td> 19<td align="center"><a href="../../../../../../more/index.htm">More</a></td> 20</tr></table> 21<hr> 22<div class="spirit-nav"> 23<a accesskey="p" href="../perf.html"><img src="../../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../perf.html"><img src="../../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../../index.html"><img src="../../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="realworld.html"><img src="../../../../../../doc/src/images/next.png" alt="Next"></a> 24</div> 25<div class="section"> 26<div class="titlepage"><div><div><h3 class="title"> 27<a name="boost_multiprecision.perf.overhead"></a><a class="link" href="overhead.html" title="The Overhead in the Number Class Wrapper">The Overhead in the 28 Number Class Wrapper</a> 29</h3></div></div></div> 30<p> 31 Using a simple <a href="http://www.boost.org/doc/libs/release/libs/multiprecision/doc/html/../../performance/arithmetic_backend.hpp" target="_top">backend 32 class</a> that wraps any <a href="https://en.cppreference.com/w/cpp/language/types" target="_top">fundamental 33 (built-in)</a> arithmetic type we can measure the overhead involved in 34 wrapping a type inside the <code class="computeroutput"><span class="identifier">number</span></code> 35 frontend, and the effect that turning on expression templates has. The following 36 table compares the performance between <code class="computeroutput"><span class="keyword">double</span></code> 37 and a <code class="computeroutput"><span class="keyword">double</span></code> wrapped inside 38 class <code class="computeroutput"><span class="identifier">number</span></code>: 39 </p> 40<div class="table"> 41<a name="boost_multiprecision.perf.overhead.bessel_functions_16_digit_precis"></a><p class="title"><b>Table 1.12. Bessel Functions (16 digit precision)</b></p> 42<div class="table-contents"><table class="table" summary="Bessel Functions (16 digit precision)"> 43<colgroup> 44<col> 45<col> 46</colgroup> 47<thead><tr> 48<th> 49 <p> 50 Type 51 </p> 52 </th> 53<th> 54 <p> 55 Time 56 </p> 57 </th> 58</tr></thead> 59<tbody> 60<tr> 61<td> 62 <p> 63 arithmetic_backend<double> 64 </p> 65 </td> 66<td> 67 <p> 68 2.09301 (0.00133409s) 69 </p> 70 </td> 71</tr> 72<tr> 73<td> 74 <p> 75 arithmetic_backend<double> - no expression templates 76 </p> 77 </td> 78<td> 79 <p> 80 1 (0.000637403s) 81 </p> 82 </td> 83</tr> 84<tr> 85<td> 86 <p> 87 double 88 </p> 89 </td> 90<td> 91 <p> 92 1.07956 (0.000688113s) 93 </p> 94 </td> 95</tr> 96</tbody> 97</table></div> 98</div> 99<br class="table-break"><p> 100 As you can see whether or not there is an overhead, and how large it is depends 101 on the actual situation, but the overhead is in any cases small. Expression 102 templates generally add a greater overhead the more complex the expression 103 becomes due to the logic of figuring out how to best unpack and evaluate 104 the expression, but of course this is also the situation where you save more 105 temporaries. For a "trivial" backend like this, saving temporaries 106 has no benefit, but for larger types it becomes a bigger win. 107 </p> 108<p> 109 The following table compares arithmetic using either <code class="computeroutput"><span class="keyword">long</span> 110 <span class="keyword">long</span></code> or <code class="computeroutput"><span class="identifier">number</span><span class="special"><</span><span class="identifier">arithmetic_backend</span><span class="special"><</span><span class="keyword">long</span> <span class="keyword">long</span><span class="special">></span> <span class="special">></span></code> for the <a href="http://www.boost.org/doc/libs/release/libs/multiprecision/doc/html/../../performance/voronoi_performance.cpp" target="_top">voronoi-diagram 111 builder test</a>: 112 </p> 113<div class="informaltable"><table class="table"> 114<colgroup> 115<col> 116<col> 117</colgroup> 118<thead><tr> 119<th> 120 <p> 121 Type 122 </p> 123 </th> 124<th> 125 <p> 126 Relative time 127 </p> 128 </th> 129</tr></thead> 130<tbody> 131<tr> 132<td> 133 <p> 134 <code class="computeroutput"><span class="identifier">int64_t</span></code> 135 </p> 136 </td> 137<td> 138 <p> 139 <span class="bold"><strong>1.0</strong></span>(0.0128646s) 140 </p> 141 </td> 142</tr> 143<tr> 144<td> 145 <p> 146 <code class="computeroutput"><span class="identifier">number</span><span class="special"><</span><span class="identifier">arithmetic_backend</span><span class="special"><</span><span class="identifier">int64_t</span><span class="special">>,</span> 147 <span class="identifier">et_off</span><span class="special">></span></code> 148 </p> 149 </td> 150<td> 151 <p> 152 1.005 (0.0129255s) 153 </p> 154 </td> 155</tr> 156</tbody> 157</table></div> 158<p> 159 This test involves mainly creating a lot of temporaries and performing a 160 small amount of arithmetic on them, with very little difference in performance 161 between the native and "wrapped" types. 162 </p> 163<div class="table"> 164<a name="boost_multiprecision.perf.overhead.platform"></a><p class="title"><b>Table 1.13. Platform Details</b></p> 165<div class="table-contents"><table class="table" summary="Platform Details"> 166<colgroup> 167<col> 168<col> 169</colgroup> 170<thead><tr> 171<th> 172 <p> 173 Platform 174 </p> 175 </th> 176<th> 177 <p> 178 Linux 5.3.0-24-generic, version #26-Ubuntu SMP Thu Nov 14 01:33:18 179 UTC 2019, x86_64 180 </p> 181 </th> 182</tr></thead> 183<tbody> 184<tr> 185<td> 186 <p> 187 Compiler 188 </p> 189 </td> 190<td> 191 <p> 192 GNU C++ version 9.2.1 20191008 193 </p> 194 </td> 195</tr> 196<tr> 197<td> 198 <p> 199 Boost 200 </p> 201 </td> 202<td> 203 <p> 204 1.72.0 205 </p> 206 </td> 207</tr> 208<tr> 209<td> 210 <p> 211 Run date 212 </p> 213 </td> 214<td> 215 <p> 216 Dec 13 2019 217 </p> 218 </td> 219</tr> 220</tbody> 221</table></div> 222</div> 223<br class="table-break"> 224</div> 225<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr> 226<td align="left"></td> 227<td align="right"><div class="copyright-footer">Copyright © 2002-2020 John 228 Maddock and Christopher Kormanyos<p> 229 Distributed under the Boost Software License, Version 1.0. (See accompanying 230 file LICENSE_1_0.txt or copy at <a href="http://www.boost.org/LICENSE_1_0.txt" target="_top">http://www.boost.org/LICENSE_1_0.txt</a>) 231 </p> 232</div></td> 233</tr></table> 234<hr> 235<div class="spirit-nav"> 236<a accesskey="p" href="../perf.html"><img src="../../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../perf.html"><img src="../../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../../index.html"><img src="../../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="realworld.html"><img src="../../../../../../doc/src/images/next.png" alt="Next"></a> 237</div> 238</body> 239</html> 240
