<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
	<title type="html"><![CDATA[Approximatrix Forums — MS MPI Package in SF]]></title>
	<link rel="self" href="http://forums.approximatrix.com/extern.php?action=feed&amp;tid=498&amp;type=atom" />
	<updated>2015-10-15T13:00:10Z</updated>
	<generator>PunBB</generator>
	<id>http://forums.approximatrix.com/viewtopic.php?id=498</id>
		<entry>
			<title type="html"><![CDATA[Re: MS MPI Package in SF]]></title>
			<link rel="alternate" href="http://forums.approximatrix.com/viewtopic.php?pid=2239#p2239" />
			<content type="html"><![CDATA[<p>I actually looked at the package project, and it appears to use the same Fortran code that your compiler is failing on compilation.&nbsp; I think the actual problem is that GNU Fortran 4.8 did not allow the use of <em>SIZEOF</em> in the parameter declarations, but GNU Fortran 4.9 and higher did.</p><p>For your case, I would change the offending line from:</p><div class="codebox"><pre><code>PARAMETER (MPI_ADDRESS_KIND=SIZEOF(IGNORED_POINTER__))</code></pre></div><p>to either the 32-bit equivalent:</p><div class="codebox"><pre><code>PARAMETER (MPI_ADDRESS_KIND=4)</code></pre></div><p>or 64-bit equivalent:</p><div class="codebox"><pre><code>PARAMETER (MPI_ADDRESS_KIND=8)</code></pre></div><p>depending on your project type.</p>]]></content>
			<author>
				<name><![CDATA[jeff]]></name>
				<uri>http://forums.approximatrix.com/profile.php?id=2</uri>
			</author>
			<updated>2015-10-15T13:00:10Z</updated>
			<id>http://forums.approximatrix.com/viewtopic.php?pid=2239#p2239</id>
		</entry>
		<entry>
			<title type="html"><![CDATA[Re: MS MPI Package in SF]]></title>
			<link rel="alternate" href="http://forums.approximatrix.com/viewtopic.php?pid=2238#p2238" />
			<content type="html"><![CDATA[<p>Thanks Jeff! Good to know that I&#039;m not misunderstanding things.</p>]]></content>
			<author>
				<name><![CDATA[wwang328]]></name>
				<uri>http://forums.approximatrix.com/profile.php?id=3567</uri>
			</author>
			<updated>2015-10-14T21:21:40Z</updated>
			<id>http://forums.approximatrix.com/viewtopic.php?pid=2238#p2238</id>
		</entry>
		<entry>
			<title type="html"><![CDATA[Re: MS MPI Package in SF]]></title>
			<link rel="alternate" href="http://forums.approximatrix.com/viewtopic.php?pid=2237#p2237" />
			<content type="html"><![CDATA[<p>I believe the Fortran files you mentioned were possibly modified when creating the included <em>.mod</em> module files, but the modified Fortran was not properly included with the package.&nbsp; Let me check our build system, and I&#039;ll let you know when the updated package is available.</p>]]></content>
			<author>
				<name><![CDATA[jeff]]></name>
				<uri>http://forums.approximatrix.com/profile.php?id=2</uri>
			</author>
			<updated>2015-10-14T21:18:26Z</updated>
			<id>http://forums.approximatrix.com/viewtopic.php?pid=2237#p2237</id>
		</entry>
		<entry>
			<title type="html"><![CDATA[MS MPI Package in SF]]></title>
			<link rel="alternate" href="http://forums.approximatrix.com/viewtopic.php?pid=2236#p2236" />
			<content type="html"><![CDATA[<p>Hi Jeff,</p><p>&nbsp; &nbsp; &nbsp;After a few trials in OpenMP with unsatisfactory improvement by the nature of my problem, I switched to MPI and downloaded the MS MPI package from SF. </p><p>&nbsp; &nbsp; &nbsp; <a href="http://forums.approximatrix.com/viewtopic.php?id=462"> This link </a> was helpful in guiding me through. But I found some problems, mostly due to my older version of SF (version 1.45). My version doesn&#039;t have the <em> Start </em> menu as mentioned in that link. </p><p>&nbsp; &nbsp; &nbsp; So here is what I did. I copied the prime_mpi.f90 under </p><p>&nbsp; &nbsp; &nbsp; </p><div class="codebox"><pre><code> C:\Users\Wenya\AppData\Local\sfpm\32\share\msmpi\examples </code></pre></div><p>&nbsp; &nbsp; <br />&nbsp; &nbsp; &nbsp; and mpi.f90 <br />&nbsp; &nbsp; <br />&nbsp; &nbsp; &nbsp; </p><div class="codebox"><pre><code> C:\Users\Wenya\AppData\Local\sfpm\64\include </code></pre></div><p>&nbsp; &nbsp; &nbsp; into my working directory (somewhere on D disk), make a new project with the two files, and compile and run. </p><p>&nbsp; &nbsp; &nbsp; It should work. But SF shows the following error <br />&nbsp; &nbsp;<br />&nbsp; &nbsp; &nbsp; </p><div class="codebox"><pre><code>     &quot;C:\Program Files (x86)\Simply Fortran\mingw-w64\bin\gfortran.exe&quot; -c -o &quot;build\mpi.o&quot; -g -m32 -IC:/Users/Wenya/AppData/Local/sfpm/32/include -fcheck=all -fno-range-check -Jmodules &quot;.\mpi.f90&quot;
.\mpi.f90:362.35:

       PARAMETER (MPI_ADDRESS_KIND=SIZEOF(IGNORED_POINTER__))
                                   1
Error: Intrinsic function &#039;sizeof&#039; at (1) is not permitted in an initialization expression
.\mpi.f90:486.11:

       USE MPI_CONSTANTS,ONLY:MPI_ADDRESS_KIND
           1
Fatal Error: File &#039;mpi_constants.mod&#039; opened at (1) is not a GNU Fortran module file
Error(E42): Last command making (build\mpi.o) returned a bad status
Error(E02): Make execution terminated</code></pre></div><p> </p><p>&nbsp; &nbsp; &nbsp; The problem seems lie in the </p><div class="codebox"><pre><code> SIZEOF </code></pre></div><p> that messes up the parameter </p><div class="codebox"><pre><code> MPI_ADDRESS_KIND </code></pre></div><p> in the MPI_CONSTANTS module. </p><p>&nbsp; &nbsp; &nbsp; &nbsp;What should I do to fix that? According to the <a href="http://forums.approximatrix.com/viewtopic.php?id=462"> previous post by brendan </a>, SF should have no problem in running examples, right? Did I miss something? Thank you very much!</p>]]></content>
			<author>
				<name><![CDATA[wwang328]]></name>
				<uri>http://forums.approximatrix.com/profile.php?id=3567</uri>
			</author>
			<updated>2015-10-13T23:17:38Z</updated>
			<id>http://forums.approximatrix.com/viewtopic.php?pid=2236#p2236</id>
		</entry>
</feed>
