|
|
<!DOCTYPE HTML>
<html lang="en" class="light sidebar-visible" dir="ltr">
<head>
<!-- Book generated using mdBook -->
<meta charset="UTF-8">
<title>pgo - Notes</title>
<!-- Custom HTML head -->
<meta name="description" content="">
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta name="theme-color" content="#ffffff">
<link rel="icon" href="../favicon.svg">
<link rel="shortcut icon" href="../favicon.png">
<link rel="stylesheet" href="../css/variables.css">
<link rel="stylesheet" href="../css/general.css">
<link rel="stylesheet" href="../css/chrome.css">
<link rel="stylesheet" href="../css/print.css" media="print">
<!-- Fonts -->
<link rel="stylesheet" href="../FontAwesome/css/font-awesome.css">
<link rel="stylesheet" href="../fonts/fonts.css">
<!-- Highlight.js Stylesheets -->
<link rel="stylesheet" href="../highlight.css">
<link rel="stylesheet" href="../tomorrow-night.css">
<link rel="stylesheet" href="../ayu-highlight.css">
<!-- Custom theme stylesheets -->
<!-- Provide site root to javascript -->
<script>
var path_to_root = "../";
var default_theme = window.matchMedia("(prefers-color-scheme: dark)").matches ? "navy" : "light";
</script>
<!-- Start loading toc.js asap -->
<script src="../toc.js"></script>
</head>
<body>
<div id="body-container">
<!-- Work around some values being stored in localStorage wrapped in quotes -->
<script>
try {
var theme = localStorage.getItem('mdbook-theme');
var sidebar = localStorage.getItem('mdbook-sidebar');
if (theme.startsWith('"') && theme.endsWith('"')) {
localStorage.setItem('mdbook-theme', theme.slice(1, theme.length - 1));
}
if (sidebar.startsWith('"') && sidebar.endsWith('"')) {
localStorage.setItem('mdbook-sidebar', sidebar.slice(1, sidebar.length - 1));
}
} catch (e) { }
</script>
<!-- Set the theme before any content is loaded, prevents flash -->
<script>
var theme;
try { theme = localStorage.getItem('mdbook-theme'); } catch(e) { }
if (theme === null || theme === undefined) { theme = default_theme; }
const html = document.documentElement;
html.classList.remove('light')
html.classList.add(theme);
html.classList.add("js");
</script>
<input type="checkbox" id="sidebar-toggle-anchor" class="hidden">
<!-- Hide / unhide sidebar before it is displayed -->
<script>
var sidebar = null;
var sidebar_toggle = document.getElementById("sidebar-toggle-anchor");
if (document.body.clientWidth >= 1080) {
try { sidebar = localStorage.getItem('mdbook-sidebar'); } catch(e) { }
sidebar = sidebar || 'visible';
} else {
sidebar = 'hidden';
}
sidebar_toggle.checked = sidebar === 'visible';
html.classList.remove('sidebar-visible');
html.classList.add("sidebar-" + sidebar);
</script>
<nav id="sidebar" class="sidebar" aria-label="Table of contents">
<!-- populated by js -->
<mdbook-sidebar-scrollbox class="sidebar-scrollbox"></mdbook-sidebar-scrollbox>
<noscript>
<iframe class="sidebar-iframe-outer" src="../toc.html"></iframe>
</noscript>
<div id="sidebar-resize-handle" class="sidebar-resize-handle">
<div class="sidebar-resize-indicator"></div>
</div>
</nav>
<div id="page-wrapper" class="page-wrapper">
<div class="page">
<div id="menu-bar-hover-placeholder"></div>
<div id="menu-bar" class="menu-bar sticky">
<div class="left-buttons">
<label id="sidebar-toggle" class="icon-button" for="sidebar-toggle-anchor" title="Toggle Table of Contents" aria-label="Toggle Table of Contents" aria-controls="sidebar">
<i class="fa fa-bars"></i>
</label>
<button id="theme-toggle" class="icon-button" type="button" title="Change theme" aria-label="Change theme" aria-haspopup="true" aria-expanded="false" aria-controls="theme-list">
<i class="fa fa-paint-brush"></i>
</button>
<ul id="theme-list" class="theme-popup" aria-label="Themes" role="menu">
<li role="none"><button role="menuitem" class="theme" id="light">Light</button></li>
<li role="none"><button role="menuitem" class="theme" id="rust">Rust</button></li>
<li role="none"><button role="menuitem" class="theme" id="coal">Coal</button></li>
<li role="none"><button role="menuitem" class="theme" id="navy">Navy</button></li>
<li role="none"><button role="menuitem" class="theme" id="ayu">Ayu</button></li>
</ul>
<button id="search-toggle" class="icon-button" type="button" title="Search. (Shortkey: s)" aria-label="Toggle Searchbar" aria-expanded="false" aria-keyshortcuts="S" aria-controls="searchbar">
<i class="fa fa-search"></i>
</button>
</div>
<h1 class="menu-title">Notes</h1>
<div class="right-buttons">
<a href="../print.html" title="Print this book" aria-label="Print this book">
<i id="print-button" class="fa fa-print"></i>
</a>
<a href="https://github.com/johannst/notes" title="Git repository" aria-label="Git repository">
<i id="git-repository-button" class="fa fa-github"></i>
</a>
</div>
</div>
<div id="search-wrapper" class="hidden">
<form id="searchbar-outer" class="searchbar-outer">
<input type="search" id="searchbar" name="searchbar" placeholder="Search this book ..." aria-controls="searchresults-outer" aria-describedby="searchresults-header">
</form>
<div id="searchresults-outer" class="searchresults-outer hidden">
<div id="searchresults-header" class="searchresults-header"></div>
<ul id="searchresults">
</ul>
</div>
</div>
<!-- Apply ARIA attributes after the sidebar and the sidebar toggle button are added to the DOM -->
<script>
document.getElementById('sidebar-toggle').setAttribute('aria-expanded', sidebar === 'visible');
document.getElementById('sidebar').setAttribute('aria-hidden', sidebar !== 'visible');
Array.from(document.querySelectorAll('#sidebar a')).forEach(function(link) {
link.setAttribute('tabIndex', sidebar === 'visible' ? 0 : -1);
});
</script>
<div id="content" class="content">
<main>
<h1 id="profile-guided-optimization-pgo"><a class="header" href="#profile-guided-optimization-pgo">Profile guided optimization (pgo)</a></h1>
<p><code>pgo</code> is an optimization technique to optimize a program for its usual
workload.</p>
<p>It is applied in two phases:</p>
<ol>
<li>Collect profiling data (best with representative benchmarks).</li>
<li>Optimize program based on collected profiling data.</li>
</ol>
<p>The following simple program is used as demonstrator.</p>
<pre><code class="language-c">#include <stdio.h>
#define NOINLINE __attribute__((noinline))
NOINLINE void foo() { puts("foo()"); }
NOINLINE void bar() { puts("bar()"); }
int main(int argc, char *argv[]) {
if (argc == 2) {
foo();
} else {
bar();
}
}
</code></pre>
<h2 id="clang"><a class="header" href="#clang">clang</a></h2>
<p>On the actual machine with <code>clang 15.0.7</code>, the following code is generated for
the <code>main()</code> function.</p>
<pre><code class="language-x86asm"># clang -o test test.c -O3
0000000000001160 <main>:
1160: 50 push rax
; Jump if argc != 2.
1161: 83 ff 02 cmp edi,0x2
1164: 75 09 jne 116f <main+0xf>
; foor() is on the hot path (fall-through).
1166: e8 d5 ff ff ff call 1140 <_Z3foov>
116b: 31 c0 xor eax,eax
116d: 59 pop rcx
116e: c3 ret
; bar() is on the cold path (branch).
116f: e8 dc ff ff ff call 1150 <_Z3barv>
1174: 31 c0 xor eax,eax
1176: 59 pop rcx
1177: c3 ret
</code></pre>
<p>The following shows how to compile with profiling instrumentation and how to
optimize the final program with the collected profiling data (<a href="https://clang.llvm.org/docs/UsersManual.html#profile-guided-optimization">llvm
pgo</a>).</p>
<p>The arguments to <code>./test</code> are chosen such that <code>9/10</code> runs call <code>bar()</code>, which
is currently on the <code>cold path</code>.</p>
<pre><code class="language-bash"># Compile test program with profiling instrumentation.
clang -o test test.cc -O3 -fprofile-instr-generate
# Collect profiling data from multiple runs.
for i in {0..10}; do
LLVM_PROFILE_FILE="prof.clang/%p.profraw" ./test $(seq 0 $i)
done
# Merge raw profiling data into single profile data.
llvm-profdata merge -o pgo.profdata prof.clang/*.profraw
# Optimize test program with profiling data.
clang -o test test.cc -O3 -fprofile-use=pgo.profdata
</code></pre>
<blockquote>
<p>NOTE: If <code>LLVM_PROFILE_FILE</code> is not given the profile data is written to
<code>default.profraw</code> which is re-written on each run. If the <code>LLVM_PROFILE_FILE</code>
contains a <code>%m</code> in the filename, a unique integer will be generated and
consecutive runs will update the same generated profraw file,
<code>LLVM_PROFILE_FILE</code> can specify a new file every time, however that requires
more storage in general.</p>
</blockquote>
<p>After optimizing the program with the profiling data, the <code>main()</code> function
looks as follows.</p>
<pre><code class="language-x86asm">0000000000001060 <main>:
1060: 50 push rax
; Jump if argc == 2.
1061: 83 ff 02 cmp edi,0x2
1064: 74 09 je 106f <main+0xf>
; bar() is on the hot path (fall-through).
1066: e8 e5 ff ff ff call 1050 <_Z3barv>
106b: 31 c0 xor eax,eax
106d: 59 pop rcx
106e: c3 ret
; foo() is on the cold path (branch).
106f: e8 cc ff ff ff call 1040 <_Z3foov>
1074: 31 c0 xor eax,eax
1076: 59 pop rcx
1077: c3 ret
</code></pre>
<h2 id="gcc"><a class="header" href="#gcc">gcc</a></h2>
<p>With <code>gcc 13.2.1</code> on the current machine, the optimizer puts <code>bar()</code> on the
<code>hot path</code> by default.</p>
<pre><code class="language-x86asm">0000000000001040 <main>:
1040: 48 83 ec 08 sub rsp,0x8
; Jump if argc == 2.
1044: 83 ff 02 cmp edi,0x2
1047: 74 0c je 1055 <main+0x15>
; bar () is on the hot path (fall-through).
1049: e8 22 01 00 00 call 1170 <_Z3barv>
104e: 31 c0 xor eax,eax
1050: 48 83 c4 08 add rsp,0x8
1054: c3 ret
; foo() is on the cold path (branch).
1055: e8 06 01 00 00 call 1160 <_Z3foov>
105a: eb f2 jmp 104e <main+0xe>
105c: 0f 1f 40 00 nop DWORD PTR [rax+0x0]
</code></pre>
<p>The following shows how to compile with profiling instrumentation and how to
optimize the final program with the collected profiling data.</p>
<p>The arguments to <code>./test</code> are chosen such that <code>2/3</code> runs call <code>foo()</code>, which
is currently on the <code>cold path</code>.</p>
<pre><code class="language-bash">gcc -o test test.cc -O3 -fprofile-generate
./test 1
./test 1
./test 2 2
gcc -o test test.cc -O3 -fprofile-use
</code></pre>
<blockquote>
<p>NOTE: Consecutive runs update the generated <code>test.gcda</code> profile data file
rather than re-write it.</p>
</blockquote>
<p>After optimizing the program with the profiling data, the <code>main()</code> function</p>
<pre><code class="language-x86asm">0000000000001040 <main.cold>:
; bar() is on the cold path (branch).
1040: e8 05 00 00 00 call 104a <_Z3barv>
1045: e9 25 00 00 00 jmp 106f <main+0xf>
0000000000001060 <main>:
1060: 51 push rcx
; Jump if argc != 2.
1061: 83 ff 02 cmp edi,0x2
1064: 0f 85 d6 ff ff ff jne 1040 <main.cold>
; for() is on the hot path (fall-through).
106a: e8 11 01 00 00 call 1180 <_Z3foov>
106f: 31 c0 xor eax,eax
1071: 5a pop rdx
1072: c3 ret
</code></pre>
</main>
<nav class="nav-wrapper" aria-label="Page navigation">
<!-- Mobile navigation buttons -->
<a rel="prev" href="../development/gcov.html" class="mobile-nav-chapters previous" title="Previous chapter" aria-label="Previous chapter" aria-keyshortcuts="Left">
<i class="fa fa-angle-left"></i>
</a>
<a rel="next prefetch" href="../linux/index.html" class="mobile-nav-chapters next" title="Next chapter" aria-label="Next chapter" aria-keyshortcuts="Right">
<i class="fa fa-angle-right"></i>
</a>
<div style="clear: both"></div>
</nav>
</div>
</div>
<nav class="nav-wide-wrapper" aria-label="Page navigation">
<a rel="prev" href="../development/gcov.html" class="nav-chapters previous" title="Previous chapter" aria-label="Previous chapter" aria-keyshortcuts="Left">
<i class="fa fa-angle-left"></i>
</a>
<a rel="next prefetch" href="../linux/index.html" class="nav-chapters next" title="Next chapter" aria-label="Next chapter" aria-keyshortcuts="Right">
<i class="fa fa-angle-right"></i>
</a>
</nav>
</div>
<script>
window.playground_copyable = true;
</script>
<script src="../elasticlunr.min.js"></script>
<script src="../mark.min.js"></script>
<script src="../searcher.js"></script>
<script src="../clipboard.min.js"></script>
<script src="../highlight.js"></script>
<script src="../book.js"></script>
<!-- Custom JS scripts -->
</div>
</body>
</html>
|
