<?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: MKL Parallelism problem in Intel® oneAPI Math Kernel Library</title>
    <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Parallelism-problem/m-p/906501#M11753</link>
    <description>&lt;DIV style="margin:0px;"&gt;
&lt;DIV id="quote_reply" style="margin-top: 5px; width: 100%;"&gt;
&lt;DIV style="margin-left:2px;margin-right:2px;"&gt;Quoting - &lt;A href="https://community.intel.com/en-us/profile/325207"&gt;sspine&lt;/A&gt;&lt;/DIV&gt;
&lt;DIV style="background-color:#E5E5E5; padding:5px;border: 1px; border-style: inset;margin-left:2px;margin-right:2px;"&gt;&lt;EM&gt;
&lt;P&gt;No one can answer this question?&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;
&lt;P&gt;MKL user manual tells me that the default value of OMP_NUM_THREADS=2. But the user manul does not tell me the default value of MKL_NUM_THREADS. I guess MKL_NUM_THREADS=1 by default. Since MKL_NUM_THREADS takes precesence over OMP_NUM_THREADS, the multi-core support is disabled by default. If this is the truth, it can be approved by my experiment.&lt;/P&gt;
&lt;P&gt;Another question is function set_mkl_num_threads() will depend on what libraries? Linux always tell me that some functions unavailable, but I do not know where can I find them.&lt;/P&gt;
&lt;/EM&gt;&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;P&gt;&lt;BR /&gt;Hi,&lt;/P&gt;
&lt;P&gt;MKL uses "Number of physical cores" threads by the default. If you has set MKL_NUM_THREADS or OMP_NUM_THREADS to the positive value then this value is used instead of the default.&lt;/P&gt;
&lt;P&gt;Please, how did you link MKL?&lt;/P&gt;
&lt;P&gt;Thanks,&lt;BR /&gt;-Vladimir&lt;/P&gt;</description>
    <pubDate>Tue, 02 Dec 2008 16:01:43 GMT</pubDate>
    <dc:creator>Vladimir_Lunev</dc:creator>
    <dc:date>2008-12-02T16:01:43Z</dc:date>
    <item>
      <title>MKL Parallelism problem</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Parallelism-problem/m-p/906499#M11751</link>
      <description>&lt;P&gt;I'm trying MKL10 with OpenMP on dual-core linux machine.&lt;/P&gt;
&lt;P&gt;The MKL user's guide told me:&lt;/P&gt;
&lt;P&gt;&lt;EM&gt;In Intel MKL 10.0, the OpenMP* software determines the default&lt;BR /&gt;number of threads. The default number of threads is equal to the number&lt;BR /&gt;of logical processors in your system for Intel OpenMP* libraries.&lt;/EM&gt;&lt;/P&gt;
&lt;P&gt;But this is not the truth for me. I have tried cblas_dgemm function (matrix multiply) using the default settings, MKL only uses one core for the calculation. But after I set the environment variable MKL_NUM_THREADS to 2, the MKL can utilize both cores, and the calculation time is about 1/2 of the first case.&lt;/P&gt;
&lt;P&gt;The following are my settings:&lt;/P&gt;
&lt;P&gt;MKL: 10.0.4.023 with customized shared library&lt;/P&gt;
&lt;P&gt;OS: SuSE Linux 10.2, 32-bit&lt;/P&gt;
&lt;P&gt;CPU: Intel Core2 Duo&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;
&lt;P&gt;Can someone tell me what on earth the default behaviour of MKL as for MKL_NUM_THREADS?&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;
&lt;P&gt;Thanks!&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;</description>
      <pubDate>Thu, 13 Nov 2008 07:51:10 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Parallelism-problem/m-p/906499#M11751</guid>
      <dc:creator>sspine</dc:creator>
      <dc:date>2008-11-13T07:51:10Z</dc:date>
    </item>
    <item>
      <title>Re: MKL Parallelism problem</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Parallelism-problem/m-p/906500#M11752</link>
      <description>&lt;P&gt;No one can answer this question?&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;
&lt;P&gt;MKL user manual tells me that the default value of OMP_NUM_THREADS=2. But the user manul does not tell me the default value of MKL_NUM_THREADS. I guess MKL_NUM_THREADS=1 by default. Since MKL_NUM_THREADS takes precesence over OMP_NUM_THREADS, the multi-core support is disabled by default. If this is the truth, it can be approved by my experiment.&lt;/P&gt;
&lt;P&gt;Another question is function set_mkl_num_threads() will depend on what libraries? Linux always tell me that some functions unavailable, but I do not know where can I find them.&lt;/P&gt;</description>
      <pubDate>Wed, 19 Nov 2008 03:30:53 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Parallelism-problem/m-p/906500#M11752</guid>
      <dc:creator>sspine</dc:creator>
      <dc:date>2008-11-19T03:30:53Z</dc:date>
    </item>
    <item>
      <title>Re: MKL Parallelism problem</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Parallelism-problem/m-p/906501#M11753</link>
      <description>&lt;DIV style="margin:0px;"&gt;
&lt;DIV id="quote_reply" style="margin-top: 5px; width: 100%;"&gt;
&lt;DIV style="margin-left:2px;margin-right:2px;"&gt;Quoting - &lt;A href="https://community.intel.com/en-us/profile/325207"&gt;sspine&lt;/A&gt;&lt;/DIV&gt;
&lt;DIV style="background-color:#E5E5E5; padding:5px;border: 1px; border-style: inset;margin-left:2px;margin-right:2px;"&gt;&lt;EM&gt;
&lt;P&gt;No one can answer this question?&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;
&lt;P&gt;MKL user manual tells me that the default value of OMP_NUM_THREADS=2. But the user manul does not tell me the default value of MKL_NUM_THREADS. I guess MKL_NUM_THREADS=1 by default. Since MKL_NUM_THREADS takes precesence over OMP_NUM_THREADS, the multi-core support is disabled by default. If this is the truth, it can be approved by my experiment.&lt;/P&gt;
&lt;P&gt;Another question is function set_mkl_num_threads() will depend on what libraries? Linux always tell me that some functions unavailable, but I do not know where can I find them.&lt;/P&gt;
&lt;/EM&gt;&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;/DIV&gt;
&lt;P&gt;&lt;BR /&gt;Hi,&lt;/P&gt;
&lt;P&gt;MKL uses "Number of physical cores" threads by the default. If you has set MKL_NUM_THREADS or OMP_NUM_THREADS to the positive value then this value is used instead of the default.&lt;/P&gt;
&lt;P&gt;Please, how did you link MKL?&lt;/P&gt;
&lt;P&gt;Thanks,&lt;BR /&gt;-Vladimir&lt;/P&gt;</description>
      <pubDate>Tue, 02 Dec 2008 16:01:43 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/MKL-Parallelism-problem/m-p/906501#M11753</guid>
      <dc:creator>Vladimir_Lunev</dc:creator>
      <dc:date>2008-12-02T16:01:43Z</dc:date>
    </item>
  </channel>
</rss>

