<?xml version="1.0" encoding="UTF-8"?>
<!-- Generated on Fri, 25 May 2012 14:32:21 -0700 -->
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <atom:link href="http://software.intel.com/en-us/articles/nuts-and-bolts-of-multithreaded-programming-1/feed/" rel="self" type="application/rss+xml" />
    <title>Intel Software Network Comments Feed</title>
    <link>http://software.intel.com/en-us/articles/nuts-and-bolts-of-multithreaded-programming-1</link>
    <description></description>
    <language>en-us</language>
    <item>
      <title>By Mikael</title>
      <description><![CDATA[ 
OpenMP is slow and buggy. You should suggest MPI instead.
The beowulf recent discussion list on OpenMP vs MPI you will see that the opinion is that MPI wins hands down in terms of performance.
 ]]></description>
      <link>http://software.intel.com/en-us/articles/nuts-and-bolts-of-multithreaded-programming-1/#comment-71</link>
      <pubDate>Sat, 08 Mar 2008 04:58:39 -0800</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/nuts-and-bolts-of-multithreaded-programming-1/#comment-71</guid>
    </item>
    <item>
      <title>By test</title>
      <description><![CDATA[ 
test
 ]]></description>
      <link>http://software.intel.com/en-us/articles/nuts-and-bolts-of-multithreaded-programming-1/#comment-72</link>
      <pubDate>Tue, 11 Mar 2008 11:50:46 -0700</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/nuts-and-bolts-of-multithreaded-programming-1/#comment-72</guid>
    </item>
    <item>
      <title>By Tim Mattson</title>
      <description><![CDATA[ 
I've used both OpenMP and MPI.  I guess I can claim to be an expert with both. And I like them both.  

It's not an issue of picking one vs. the other.   Sometimes OpenMP is the right way to go.  Sometimes MPI is the right way to go.  They both have their place.

You mention a discussion on the Beowulf list.  Well, I'd be hard pressed to name a more biased audience.  Nothing against Beowulf or the people on the list ... but that's a cluster list.  If you are interested in cluster computing or think you might need portability between an SMP and a cluster, of course you'd pick MPI.  It's a no brainer.

But if you know you will be working on a NUMA or SMP platform, OpenMP has some real advantages.  (Support for incremental parallellism for starters).
 ]]></description>
      <link>http://software.intel.com/en-us/articles/nuts-and-bolts-of-multithreaded-programming-1/#comment-73</link>
      <pubDate>Tue, 11 Mar 2008 11:52:27 -0700</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/nuts-and-bolts-of-multithreaded-programming-1/#comment-73</guid>
    </item>
    <item>
      <title>By Tim Mattson</title>
      <description><![CDATA[ 
And I take exception to your comment that OpenMP is buggy.  Is there an OpenMP compiler out there that has been giving you problems?  Take it up with that vendor, don't slander the whole standard because of your experiences with a particular standard.  OpenMP is an API and to call it &quot;buggy&quot; just doesn't make sense.
 ]]></description>
      <link>http://software.intel.com/en-us/articles/nuts-and-bolts-of-multithreaded-programming-1/#comment-74</link>
      <pubDate>Tue, 11 Mar 2008 11:52:46 -0700</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/nuts-and-bolts-of-multithreaded-programming-1/#comment-74</guid>
    </item>
  </channel></rss>
