<?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: IPP on a multicore system. in Intel® Integrated Performance Primitives</title>
    <link>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965456#M20111</link>
    <description>&lt;P&gt;Is there some way to know which of the IPP'sfunctions are multi-threaded and which aren't?&lt;/P&gt;
&lt;P&gt;thanks,&lt;/P&gt;
&lt;P&gt;Nizan.&lt;/P&gt;
&lt;DIV&gt;&lt;/DIV&gt;</description>
    <pubDate>Wed, 02 Nov 2005 04:51:00 GMT</pubDate>
    <dc:creator>nizanh</dc:creator>
    <dc:date>2005-11-02T04:51:00Z</dc:date>
    <item>
      <title>IPP on a multicore system.</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965454#M20109</link>
      <description>&lt;DIV&gt;Hello,&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;I need to know how IPP behave on a multi-core system (say, a Pentium D).&lt;/DIV&gt;
&lt;DIV&gt;When I call IPP function does it run in multi-threads? If so, can I expect significant performance improvement on a dual -core, without rewriting the code?&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;thanks,&lt;/DIV&gt;
&lt;DIV&gt;Nizan.&lt;/DIV&gt;</description>
      <pubDate>Tue, 01 Nov 2005 23:52:20 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965454#M20109</guid>
      <dc:creator>nizanh</dc:creator>
      <dc:date>2005-11-01T23:52:20Z</dc:date>
    </item>
    <item>
      <title>Re: IPP on a multicore system.</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965455#M20110</link>
      <description>&lt;DIV&gt;Hi,&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;Yes, somefunctions in IPP DLLs use OpenMP threading inside, whereas the same functions in static libraries does not usethreading at all. Why only some functions? Because not all agorithms can get good speedup from threading. Other reason - it is better to implement threading above IPP functions, because you know your application specifics better.&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;Regards,&lt;BR /&gt; Vladimir&lt;/DIV&gt;</description>
      <pubDate>Wed, 02 Nov 2005 00:20:06 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965455#M20110</guid>
      <dc:creator>Vladimir_Dudnik</dc:creator>
      <dc:date>2005-11-02T00:20:06Z</dc:date>
    </item>
    <item>
      <title>Re: IPP on a multicore system.</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965456#M20111</link>
      <description>&lt;P&gt;Is there some way to know which of the IPP'sfunctions are multi-threaded and which aren't?&lt;/P&gt;
&lt;P&gt;thanks,&lt;/P&gt;
&lt;P&gt;Nizan.&lt;/P&gt;
&lt;DIV&gt;&lt;/DIV&gt;</description>
      <pubDate>Wed, 02 Nov 2005 04:51:00 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965456#M20111</guid>
      <dc:creator>nizanh</dc:creator>
      <dc:date>2005-11-02T04:51:00Z</dc:date>
    </item>
    <item>
      <title>Re: IPP on a multicore system.</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965457#M20112</link>
      <description>Yes, please document exactly which APIs use multiple threads. My software must never block as I already maintain exactly one runnable thread per logical CPU. Any multi-threading done inside IPP (which would require blocking for things like thread startup/shutdown and memory allocation) will seriosly harm my application's performance.</description>
      <pubDate>Wed, 02 Nov 2005 13:49:02 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965457#M20112</guid>
      <dc:creator>kevin120bl</dc:creator>
      <dc:date>2005-11-02T13:49:02Z</dc:date>
    </item>
    <item>
      <title>Re: IPP on a multicore system.</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965458#M20113</link>
      <description>&lt;P&gt;Hi,&lt;/P&gt;
&lt;P&gt;unfortunately we do not provide exact list of functions which use OpenMP because it is subject of change andvaries on different architectures. Moreover,it also depends on amount of data to process. Thecriteria can be individual for each functionand also depends on architecture. In general, OpenMP threadingis usedmostly incolor-conversion functions. We also provide special function, ippSetNumThreads,(in ippCore library), it gives you possibility to disable internal threading at all. And I want to remind you that IPP static libraries does not contains OpenMP code, it is used only in DLL.&lt;/P&gt;
&lt;DIV&gt;Regards,&lt;BR /&gt; Vladimir&lt;/DIV&gt;</description>
      <pubDate>Fri, 04 Nov 2005 00:56:00 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/IPP-on-a-multicore-system/m-p/965458#M20113</guid>
      <dc:creator>Vladimir_Dudnik</dc:creator>
      <dc:date>2005-11-04T00:56:00Z</dc:date>
    </item>
  </channel>
</rss>

