<?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 Multicore performance drop under Windows x64 in Intel® oneAPI Math Kernel Library</title>
    <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817413#M4419</link>
    <description>&lt;DIV&gt;&lt;P&gt;No, that not correct.
You are using non-threaded libraries. Please try to use,&lt;/P&gt;

&lt;P&gt;mkl_intel_lp64.lib
mkl_intel_thread.lib mkl_core.liblibiomp5md.lib&lt;/P&gt;

&lt;P&gt;as an examplefor static
linking with supporting LP64 interface.&lt;/P&gt;

&lt;P&gt;&lt;/P&gt;

&lt;P&gt;For more info, please
refer to the Users Guide or see here &lt;A href="http://software.intel.com/en-us/articles/intel-mkl-link-line-advisor/"&gt;the Linker Adviser&lt;/A&gt;.&lt;/P&gt;&lt;/DIV&gt;</description>
    <pubDate>Wed, 26 Jan 2011 09:11:17 GMT</pubDate>
    <dc:creator>Gennady_F_Intel</dc:creator>
    <dc:date>2011-01-26T09:11:17Z</dc:date>
    <item>
      <title>Multicore performance drop under Windows x64</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817408#M4414</link>
      <description>I'm using MKL for matrix multiplication (cblas_sgemm function) under 64-bit Windows. This far I'been using 32-bit code, and -- according to the Windows Task Manager -- my routine exploited both my cores to 100%. Now I compiled a 64-bit version from the code, and now Windows Task Manager shows only 50-60% processor use. interestingly, one of the cores is used at about 80%, and the other one about 20% (compared to the earlier 100%-100%). Is there anything I can do to have the same performance with the 64-bit version than with the 32-bit version?&lt;BR /&gt;Thanks a lot,&lt;BR /&gt;</description>
      <pubDate>Mon, 24 Jan 2011 11:53:58 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817408#M4414</guid>
      <dc:creator>lajszlo</dc:creator>
      <dc:date>2011-01-24T11:53:58Z</dc:date>
    </item>
    <item>
      <title>Multicore performance drop under Windows x64</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817409#M4415</link>
      <description>It looks like you're getting just one thread. This would be expected MKL default behavior if you run on a single core HyperThreaded CPU. In case the OS is confusing MKL into thinking you have a single core when you have dual, does MKL_DYNAMIC help?&lt;BR /&gt;Windows 7 is said to have scheduling improvements over earlier versions of Windows X64.</description>
      <pubDate>Mon, 24 Jan 2011 14:25:29 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817409#M4415</guid>
      <dc:creator>TimP</dc:creator>
      <dc:date>2011-01-24T14:25:29Z</dc:date>
    </item>
    <item>
      <title>Multicore performance drop under Windows x64</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817410#M4416</link>
      <description>You are right, the 32-bit executable has 3 threads (according to Task Manager), while the 64-bit version has only 1 thread. I forgot to mention that I have Windows XP 64-bit version.&lt;BR /&gt;I'm absolutely not familiar with threading in MKL. I added the following two lines to my code:&lt;BR /&gt;mkl_set_dynamic(0);&lt;BR /&gt;mkl_set_num_threads(2, MKL_BLAS);&lt;BR /&gt;but nothing has changed, still there is only one thread. Do you mean this or something else by "does MKL_DYNAMIC help"?&lt;BR /&gt;Thanks&lt;BR /&gt;</description>
      <pubDate>Mon, 24 Jan 2011 15:37:07 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817410#M4416</guid>
      <dc:creator>lajszlo</dc:creator>
      <dc:date>2011-01-24T15:37:07Z</dc:date>
    </item>
    <item>
      <title>Multicore performance drop under Windows x64</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817411#M4417</link>
      <description>Which libraries do you use for linking application?</description>
      <pubDate>Tue, 25 Jan 2011 13:10:27 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817411#M4417</guid>
      <dc:creator>Gennady_F_Intel</dc:creator>
      <dc:date>2011-01-25T13:10:27Z</dc:date>
    </item>
    <item>
      <title>Multicore performance drop under Windows x64</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817412#M4418</link>
      <description>I'm using MKL 10.1.1.022. For the 32-bit version (which works fine) I link &lt;B&gt;mkl_c.lib, mkl_sequential.lib, libiomp5md.lib&lt;/B&gt; from the ia32\lib directory. For the x64 version there is no mkl_c.lib in the em64t\lib directory, so I link &lt;B&gt;mkl_core.lib, mkl_intel_lp64.lib, mkl_sequential.lib, libiomp5md.lib. &lt;/B&gt;Is that OK? Thanks.&lt;BR /&gt;</description>
      <pubDate>Wed, 26 Jan 2011 08:38:24 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817412#M4418</guid>
      <dc:creator>lajszlo</dc:creator>
      <dc:date>2011-01-26T08:38:24Z</dc:date>
    </item>
    <item>
      <title>Multicore performance drop under Windows x64</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817413#M4419</link>
      <description>&lt;DIV&gt;&lt;P&gt;No, that not correct.
You are using non-threaded libraries. Please try to use,&lt;/P&gt;

&lt;P&gt;mkl_intel_lp64.lib
mkl_intel_thread.lib mkl_core.liblibiomp5md.lib&lt;/P&gt;

&lt;P&gt;as an examplefor static
linking with supporting LP64 interface.&lt;/P&gt;

&lt;P&gt;&lt;/P&gt;

&lt;P&gt;For more info, please
refer to the Users Guide or see here &lt;A href="http://software.intel.com/en-us/articles/intel-mkl-link-line-advisor/"&gt;the Linker Adviser&lt;/A&gt;.&lt;/P&gt;&lt;/DIV&gt;</description>
      <pubDate>Wed, 26 Jan 2011 09:11:17 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817413#M4419</guid>
      <dc:creator>Gennady_F_Intel</dc:creator>
      <dc:date>2011-01-26T09:11:17Z</dc:date>
    </item>
    <item>
      <title>Multicore performance drop under Windows x64</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817414#M4420</link>
      <description>This solved the problem, thanks a lot!!&lt;BR /&gt;</description>
      <pubDate>Wed, 26 Jan 2011 13:10:42 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Multicore-performance-drop-under-Windows-x64/m-p/817414#M4420</guid>
      <dc:creator>lajszlo</dc:creator>
      <dc:date>2011-01-26T13:10:42Z</dc:date>
    </item>
  </channel>
</rss>

