<?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 MKL multi-threading in Intel® Moderncode for Parallel Architectures</title>
    <link>https://community.intel.com/t5/Intel-Moderncode-for-Parallel/MKL-multi-threading/m-p/980063#M5711</link>
    <description>Hi:&lt;BR /&gt;I'm wondering how do I enable the multi-threaded execution of BLAS operations using MKL. The website says it's no effort work, do I need to set some flags when compiling or linking? Does MKL apply some thresholding on the size to determine when to MT and when not?&lt;BR /&gt;Thanks&lt;BR /&gt;XW</description>
    <pubDate>Thu, 20 Nov 2003 07:48:17 GMT</pubDate>
    <dc:creator>Deleted_U_Intel</dc:creator>
    <dc:date>2003-11-20T07:48:17Z</dc:date>
    <item>
      <title>MKL multi-threading</title>
      <link>https://community.intel.com/t5/Intel-Moderncode-for-Parallel/MKL-multi-threading/m-p/980063#M5711</link>
      <description>Hi:&lt;BR /&gt;I'm wondering how do I enable the multi-threaded execution of BLAS operations using MKL. The website says it's no effort work, do I need to set some flags when compiling or linking? Does MKL apply some thresholding on the size to determine when to MT and when not?&lt;BR /&gt;Thanks&lt;BR /&gt;XW</description>
      <pubDate>Thu, 20 Nov 2003 07:48:17 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Moderncode-for-Parallel/MKL-multi-threading/m-p/980063#M5711</guid>
      <dc:creator>Deleted_U_Intel</dc:creator>
      <dc:date>2003-11-20T07:48:17Z</dc:date>
    </item>
    <item>
      <title>Re: MKL multi-threading</title>
      <link>https://community.intel.com/t5/Intel-Moderncode-for-Parallel/MKL-multi-threading/m-p/980064#M5712</link>
      <description>&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;Hello,&lt;/DIV&gt;
&lt;DIV&gt;Sorry it took so long to respond to your question. Migrating to the new forum software was bumpier than expected.&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;MKL uses OpenMP for threading. To take advantage of threaded MKL functions, simply set the &lt;STRONG&gt;OMP_NUM_THREADS &lt;/STRONG&gt;environment variable to a value greater than one. It's usually best to set &lt;STRONG&gt;OMP_NUM_THREADS &lt;/STRONG&gt;less than or equal to the number of processors in your system.&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;MKL does have internal thresholds to determine whether to perform a computation in parallel or serial. You can also apply your own thresholdsusing the OpenMP API, specifically the&lt;STRONG&gt;omp_set_num_threads &lt;/STRONG&gt;function. The online technical notes havesome additional information about &lt;A href="http://www.intel.com/software/products/mkl/docs/mklusel.htm#Using%20MKL%20Parallelism" target="_blank"&gt;MKL parallelism&lt;/A&gt;.&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;Please let us know if you have any other questions.&lt;/DIV&gt;
&lt;DIV&gt;&lt;/DIV&gt;
&lt;DIV&gt;Best regards,&lt;/DIV&gt;
&lt;DIV&gt;Henry&lt;/DIV&gt;&lt;P&gt;Message Edited by MADhagabb on &lt;SPAN class="date_text"&gt;12-11-2003&lt;/SPAN&gt; &lt;SPAN class="time_text"&gt;02:44 PM&lt;/SPAN&gt;&lt;/P&gt;</description>
      <pubDate>Fri, 12 Dec 2003 06:37:19 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Moderncode-for-Parallel/MKL-multi-threading/m-p/980064#M5712</guid>
      <dc:creator>Henry_G_Intel</dc:creator>
      <dc:date>2003-12-12T06:37:19Z</dc:date>
    </item>
  </channel>
</rss>

