<?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 matrix distribution  in pdgemm and pdgesvx in Intel® oneAPI Math Kernel Library</title>
    <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/matrix-distribution-in-pdgemm-and-pdgesvx/m-p/869256#M8317</link>
    <description>&lt;P&gt;&lt;/P&gt;
&lt;P&gt;I want to solve linear equation system like&lt;/P&gt;
&lt;P&gt;A*X=B*C;&lt;/P&gt;
&lt;P&gt;sizeof A=(MxM) sizeof X=(MxN)&lt;/P&gt;
&lt;P&gt;sizeof B=(MxM) sizeof C=(MxN)&lt;/P&gt;
&lt;P&gt;I set grid as (pxq) and block size is same in both direction as nb.&lt;/P&gt;
&lt;P&gt;I can calculate B*C from pdgemm_ routine.It uses block cyclic distribution.I can distribute the matrices as I want.&lt;/P&gt;
&lt;P&gt;A*X=D=B*C can be solved from pdgesv_ routine. but this routine use block distribution.&lt;/P&gt;
&lt;P&gt;for example when M=3, N=3, p=3,q=3,nb=3 it work correctly&lt;/P&gt;
&lt;P&gt;on the ohter and,&lt;/P&gt;
&lt;P&gt;when M=18,N=18,p=3,q=3,nb=3 it fails, you have to give nb=18/3=9&lt;/P&gt;
&lt;P&gt;So you can't distribute the matrices as you want. you always have to arrange block size or matrix size to provide M%p=0 &amp;amp;&amp;amp; N%q=0. if your matrix is to be prime number or etc. it is unsolvable with scalapack routines. is there any way to solve system, according to pdgemm distribution.&lt;/P&gt;
&lt;P&gt;Thank,&lt;/P&gt;
&lt;P&gt;Regards&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;</description>
    <pubDate>Tue, 16 Mar 2010 14:14:00 GMT</pubDate>
    <dc:creator>oguren</dc:creator>
    <dc:date>2010-03-16T14:14:00Z</dc:date>
    <item>
      <title>matrix distribution  in pdgemm and pdgesvx</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/matrix-distribution-in-pdgemm-and-pdgesvx/m-p/869256#M8317</link>
      <description>&lt;P&gt;&lt;/P&gt;
&lt;P&gt;I want to solve linear equation system like&lt;/P&gt;
&lt;P&gt;A*X=B*C;&lt;/P&gt;
&lt;P&gt;sizeof A=(MxM) sizeof X=(MxN)&lt;/P&gt;
&lt;P&gt;sizeof B=(MxM) sizeof C=(MxN)&lt;/P&gt;
&lt;P&gt;I set grid as (pxq) and block size is same in both direction as nb.&lt;/P&gt;
&lt;P&gt;I can calculate B*C from pdgemm_ routine.It uses block cyclic distribution.I can distribute the matrices as I want.&lt;/P&gt;
&lt;P&gt;A*X=D=B*C can be solved from pdgesv_ routine. but this routine use block distribution.&lt;/P&gt;
&lt;P&gt;for example when M=3, N=3, p=3,q=3,nb=3 it work correctly&lt;/P&gt;
&lt;P&gt;on the ohter and,&lt;/P&gt;
&lt;P&gt;when M=18,N=18,p=3,q=3,nb=3 it fails, you have to give nb=18/3=9&lt;/P&gt;
&lt;P&gt;So you can't distribute the matrices as you want. you always have to arrange block size or matrix size to provide M%p=0 &amp;amp;&amp;amp; N%q=0. if your matrix is to be prime number or etc. it is unsolvable with scalapack routines. is there any way to solve system, according to pdgemm distribution.&lt;/P&gt;
&lt;P&gt;Thank,&lt;/P&gt;
&lt;P&gt;Regards&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;
&lt;P&gt;&lt;/P&gt;</description>
      <pubDate>Tue, 16 Mar 2010 14:14:00 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/matrix-distribution-in-pdgemm-and-pdgesvx/m-p/869256#M8317</guid>
      <dc:creator>oguren</dc:creator>
      <dc:date>2010-03-16T14:14:00Z</dc:date>
    </item>
    <item>
      <title>matrix distribution  in pdgemm and pdgesvx</title>
      <link>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/matrix-distribution-in-pdgemm-and-pdgesvx/m-p/869257#M8318</link>
      <description>Hello, oguren,&lt;BR /&gt;&lt;BR /&gt;Could you please narrow down how exactly do you do 2D-block cyclic distribution?&lt;BR /&gt;&lt;BR /&gt;Thank you,&lt;BR /&gt;Alexander</description>
      <pubDate>Wed, 24 Mar 2010 08:56:03 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/matrix-distribution-in-pdgemm-and-pdgesvx/m-p/869257#M8318</guid>
      <dc:creator>Alex_Kosenkov</dc:creator>
      <dc:date>2010-03-24T08:56:03Z</dc:date>
    </item>
  </channel>
</rss>

