<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic Re: PGO and warning # 11503 in Intel® Fortran Compiler</title>
    <link>https://community.intel.com/t5/Intel-Fortran-Compiler/PGO-and-warning-11503/m-p/903572#M81510</link>
    <description>Thank you! I thought that PGO would improve the performances of this application, but unfortunately this application seems hard to predict. So in this case applying PGO results in no effects on execution time.&lt;BR /&gt;Fortunately the other optimizations based on architecture family work great for this application!&lt;BR /&gt;I go a little OT: I'm trying to run the exe file with "prof-run", but the exe file needs some arguments and possibly a redirect. For example to start the execution of the application I double-click on a file .bat which contains:&lt;BR /&gt;&lt;BLOCKQUOTE&gt;set var1=.&lt;BR /&gt;set var2=.&lt;BR /&gt;application.exe -macro file.rsl &amp;gt; file.log&lt;BR /&gt;&lt;/BLOCKQUOTE&gt;So I thought that the right way to run prof-run could be a .bat file like this:&lt;BR /&gt;&lt;BLOCKQUOTE&gt;set var1=.&lt;BR /&gt;set var2=.&lt;BR /&gt;profrun -branch -dcache -icache application.exe -macro file.rsl &amp;gt; file.log&lt;BR /&gt;&lt;/BLOCKQUOTE&gt;Thank you another time!&lt;BR /&gt;</description>
    <pubDate>Wed, 16 Apr 2008 13:55:43 GMT</pubDate>
    <dc:creator>desmo3re</dc:creator>
    <dc:date>2008-04-16T13:55:43Z</dc:date>
    <item>
      <title>PGO and warning # 11503</title>
      <link>https://community.intel.com/t5/Intel-Fortran-Compiler/PGO-and-warning-11503/m-p/903570#M81508</link>
      <description>Hi to all,&lt;BR /&gt;I'm tryng to maximize the performances of a Fortran application. I'm trying to use PGO with IPO, but I'm receiving a lot of warnings like this:&lt;BR /&gt;&lt;BLOCKQUOTE&gt;warning #11503: Total routines 12, routines w/o profile info: 2 ...&lt;BR /&gt;&lt;/BLOCKQUOTE&gt;I think that PGO hasn't profile informations for some routines, but I haven't found several info in Intel Compiler Help.&lt;BR /&gt;I would know some issues to eliminate this warnings, because at the moment the application doesn't go faster after PGO and IPO optimizations and I think that these warnings would be a possible cause.&lt;BR /&gt;</description>
      <pubDate>Wed, 16 Apr 2008 11:24:33 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Fortran-Compiler/PGO-and-warning-11503/m-p/903570#M81508</guid>
      <dc:creator>desmo3re</dc:creator>
      <dc:date>2008-04-16T11:24:33Z</dc:date>
    </item>
    <item>
      <title>Re: PGO and warning # 11503</title>
      <link>https://community.intel.com/t5/Intel-Fortran-Compiler/PGO-and-warning-11503/m-p/903571#M81509</link>
      <description>The warning would mean that you didn't use those routines in your training runs, or didn't compile them with /Qprof-gen.  if you don't waste any time there, that would not be a cause for ineffectiveness of PGO.  
&lt;BR /&gt;The main goal of PGO is to improve branch prediction.  It's hard enough to guess whether it is likely to show a gain when we can see the source code.  Ifort doesn't depend as much on PGO as it did some years ago.  An exception, where PGO should help, is a SELECT CASE where few of the cases are frequently used, but the cases aren't sorted by frequency.</description>
      <pubDate>Wed, 16 Apr 2008 12:43:49 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Fortran-Compiler/PGO-and-warning-11503/m-p/903571#M81509</guid>
      <dc:creator>TimP</dc:creator>
      <dc:date>2008-04-16T12:43:49Z</dc:date>
    </item>
    <item>
      <title>Re: PGO and warning # 11503</title>
      <link>https://community.intel.com/t5/Intel-Fortran-Compiler/PGO-and-warning-11503/m-p/903572#M81510</link>
      <description>Thank you! I thought that PGO would improve the performances of this application, but unfortunately this application seems hard to predict. So in this case applying PGO results in no effects on execution time.&lt;BR /&gt;Fortunately the other optimizations based on architecture family work great for this application!&lt;BR /&gt;I go a little OT: I'm trying to run the exe file with "prof-run", but the exe file needs some arguments and possibly a redirect. For example to start the execution of the application I double-click on a file .bat which contains:&lt;BR /&gt;&lt;BLOCKQUOTE&gt;set var1=.&lt;BR /&gt;set var2=.&lt;BR /&gt;application.exe -macro file.rsl &amp;gt; file.log&lt;BR /&gt;&lt;/BLOCKQUOTE&gt;So I thought that the right way to run prof-run could be a .bat file like this:&lt;BR /&gt;&lt;BLOCKQUOTE&gt;set var1=.&lt;BR /&gt;set var2=.&lt;BR /&gt;profrun -branch -dcache -icache application.exe -macro file.rsl &amp;gt; file.log&lt;BR /&gt;&lt;/BLOCKQUOTE&gt;Thank you another time!&lt;BR /&gt;</description>
      <pubDate>Wed, 16 Apr 2008 13:55:43 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Fortran-Compiler/PGO-and-warning-11503/m-p/903572#M81510</guid>
      <dc:creator>desmo3re</dc:creator>
      <dc:date>2008-04-16T13:55:43Z</dc:date>
    </item>
    <item>
      <title>Re: PGO and warning # 11503</title>
      <link>https://community.intel.com/t5/Intel-Fortran-Compiler/PGO-and-warning-11503/m-p/903573#M81511</link>
      <description>Do you have an example of Fortran application which runs faster after IPO/PGO optimizations?&lt;BR /&gt;I ask for this because I'm working on my thesis and I think that it would be usefull to write an example where PGO/IPO optimizations work well.&lt;BR /&gt;Thanks in advance!&lt;BR /&gt;</description>
      <pubDate>Mon, 21 Apr 2008 11:06:45 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Fortran-Compiler/PGO-and-warning-11503/m-p/903573#M81511</guid>
      <dc:creator>desmo3re</dc:creator>
      <dc:date>2008-04-21T11:06:45Z</dc:date>
    </item>
  </channel>
</rss>

