| Daniel Dunbar | 9d284e5 | 2009-06-26 18:14:41 +0000 | [diff] [blame] | 1 | <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" | 
 | 2 |           "http://www.w3.org/TR/html4/strict.dtd"> | 
 | 3 | <html> | 
 | 4 | <head> | 
 | 5 |   <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1" /> | 
 | 6 |   <title>Clang - Performance</title> | 
 | 7 |   <link type="text/css" rel="stylesheet" href="menu.css" /> | 
 | 8 |   <link type="text/css" rel="stylesheet" href="content.css" /> | 
 | 9 |   <style type="text/css"> | 
 | 10 | </style> | 
 | 11 | </head> | 
 | 12 | <body> | 
 | 13 |  | 
 | 14 | <!--#include virtual="menu.html.incl"--> | 
 | 15 |  | 
 | 16 | <div id="content"> | 
 | 17 |  | 
 | 18 | <!--*************************************************************************--> | 
 | 19 | <h1>Clang - Performance</h1> | 
 | 20 | <!--*************************************************************************--> | 
 | 21 |  | 
 | 22 | <p>This page shows the compile time performance of Clang on two | 
 | 23 | interesting benchmarks: | 
 | 24 | <ul> | 
 | 25 |   <li><i>Sketch</i>: The Objective-C example application shipped on | 
 | 26 |     Mac OS X as part of Xcode. <i>Sketch</i> is indicative of a | 
 | 27 |     "typical" Objective-C app. The source itself has a relatively | 
 | 28 |     small amount of code (~7,500 lines of source code), but it relies | 
 | 29 |     on the extensive Cocoa APIs to build its functionality. Like many | 
 | 30 |     Objective-C applications, it includes <tt>Cocoa/Cocoa.h</tt> in | 
 | 31 |     all of its source files, which represents a significant stress | 
 | 32 |     test of the front-end's performance on lexing, preprocessing, | 
 | 33 |     parsing, and syntax analysis.</li> | 
 | 34 |   <li><i>176.gcc</i>: This is the gcc-2.7.2.2 code base as present in | 
 | 35 |     SPECINT 2000. In contrast to Sketch, <i>176.gcc</i> consists of a | 
 | 36 |     large amount of C source code (~200,000 lines) with few system | 
 | 37 |     dependencies. This stresses the back-end's performance on generating | 
 | 38 |     assembly code and debug information.</li> | 
 | 39 | </ul> | 
 | 40 | </p> | 
 | 41 |  | 
 | 42 | <p> | 
 | 43 | For previous performance numbers, please | 
 | 44 | go <a href="performance-2008-10-31.html">here</a>. | 
 | 45 | </p> | 
 | 46 |  | 
 | 47 | <!--*************************************************************************--> | 
 | 48 | <h2><a name="experiments">Experiments</a></h2> | 
 | 49 | <!--*************************************************************************--> | 
 | 50 |  | 
 | 51 | <p>Measurements are done by running a full build (using xcodebuild or | 
 | 52 | make for Sketch and 176.gcc respectively) using Clang and gcc 4.2 as | 
 | 53 | compilers; gcc is run both with and without the new clang driver (ccc) | 
 | 54 | in order to evaluate the overhead of the driver itself.</p> | 
 | 55 |  | 
 | 56 | <p>In order to track the performance of various subsystems the timings | 
 | 57 | have been broken down into separate stages where possible. This is | 
 | 58 | done by over-riding the CC environment variable used during the build | 
 | 59 | to point to one of a few simple shell scripts which may skip part of | 
 | 60 | the build. | 
 | 61 |  | 
 | 62 | <ul> | 
 | 63 |   <li><tt>non-compiler</tt>: The overhead of the build system itself; | 
 | 64 |     for Sketch this also includes the time to build/copy various | 
 | 65 |     non-source code resource files.</li> | 
 | 66 |   <li><tt>+ driver</tt>: Add execution of the driver, but do not execute any | 
 | 67 |     commands (by using the -### driver option).</li> | 
 | 68 |   <li><tt>+ pch gen</tt>: Add generation of PCH files.</li> | 
 | 69 |   <li><tt>+ cpp</tt>: Add preprocessing of source files (this time is | 
 | 70 |     include in syntax for gcc).</li> | 
 | 71 |   <li><tt>+ parse</tt>: Add parsing of source files (this time is | 
 | 72 |     include in syntax for gcc).</li> | 
 | 73 |   <li><tt>+ syntax</tt>: Add semantic checking of source files (for | 
 | 74 |     gcc, this includes preprocessing and parsing as well).</li> | 
 | 75 |   <li><tt>+ IRgen</tt>: Add generation of LLVM IR (gcc has no | 
 | 76 |     corresponding phase).</li> | 
 | 77 |   <li><tt>+ codegen</tt>: Add generation of assembler files.</li> | 
 | 78 |   <li><tt>+ assembler</tt>: Add assembler time to generate .o files.</li> | 
 | 79 |   <li><tt>+ linker</tt>: Add linker time.</li> | 
 | 80 | </ul> | 
 | 81 | </p> | 
 | 82 |  | 
 | 83 | <p>This set of stages is chosen to be approximately additive, that is | 
 | 84 | each subsequent stage simply adds some additional processing. The | 
 | 85 | timings measure the delta of the given stage from the previous | 
 | 86 | one. For example, the timings for <tt>+ syntax</tt> below show the | 
 | 87 | difference of running with <tt>+ syntax</tt> versus running with <tt>+ | 
 | 88 | parse</tt> (for clang) or <tt>+ driver</tt> with gcc. This amounts to | 
 | 89 | a fairly accurate measure of only the time to perform semantic | 
 | 90 | analysis (and preprocessing/parsing, in the case of gcc).</p> | 
 | 91 |  | 
 | 92 | <!--*************************************************************************--> | 
 | 93 | <h2><a name="timings">Timing Results</a></h2> | 
 | 94 | <!--*************************************************************************--> | 
 | 95 |  | 
 | 96 | <!--=======================================================================--> | 
 | 97 | <h3><a name="2009-03-02">2009-03-02</a></h3> | 
 | 98 | <!--=======================================================================--> | 
 | 99 |  | 
 | 100 | <a href="timing-data/2009-03-02/sketch.pdf"> | 
 | 101 | <img class="img_slide"  | 
 | 102 |      src="timing-data/2009-03-02/sketch.png" alt="Sketch Timings"/> | 
 | 103 | </a> | 
 | 104 |  | 
 | 105 | <a href="timing-data/2009-03-02/176.gcc.pdf"> | 
 | 106 | <img class="img_slide"  | 
 | 107 |      src="timing-data/2009-03-02/176.gcc.png" alt="176.gcc Timings"/> | 
 | 108 | </a> | 
 | 109 |  | 
 | 110 | </div> | 
 | 111 | </body> | 
 | 112 | </html> |