<?xml version="1.0" encoding="UTF-8"?>
<!-- Generated on Wed, 25 Nov 2009 09:23:52 -0800 -->
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <atom:link href="http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/feed/" rel="self" type="application/rss+xml" />
    <title>Intel Software Network Comments feed</title>
    <link>http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/feed/</link>
    <description></description>
    <language>en-us</language>
    <item>
      <title>By hendlerd</title>
      <description><![CDATA[ 
Nice article. Guidelines for choosing between OpenMP and TBB will also be useful. ]]></description>
      <link>http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-9259</link>
      <pubDate>Wed, 26 Nov 2008 12:18:27 -0800</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-9259</guid>
    </item>
    <item>
      <title>By atacoral</title>
      <description><![CDATA[ Thank you for the article, one question though.
Is it not possible to call the function that sets thread priority of the underlying threading library in a parallel section?  Or is it possible but not good practice? ]]></description>
      <link>http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-21650</link>
      <pubDate>Wed, 25 Mar 2009 01:53:28 -0700</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-21650</guid>
    </item>
    <item>
      <title>By Andrew Binstock</title>
      <description><![CDATA[ @Atacoral: I'm sorry, I don't understand your question. Does it refer to something in this article, or is it a general threading question? If it's from the article, could you give me more context? 

Also, I am not sure what you mean by a parallel section. Finally, thread priority is generally not set at the library level. Rather, the typical implementation is that threads have a default priority and threads that need higher or lower priority have their priority set on a per-thread basis.  ]]></description>
      <link>http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-21722</link>
      <pubDate>Wed, 25 Mar 2009 22:05:32 -0700</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-21722</guid>
    </item>
    <item>
      <title>By atacoral</title>
      <description><![CDATA[ Sorry for the ambiguous question, let me try to clarify:

Quoting from the part Limitations of OpenMP (the paragraph before the last) "As a result, if you need to tweak one of these activities (such as raising thread priority), you cannot use OpenMP"

a- What I meant was why can't I set thread priority if I am using OpenMP ?
b- Does OpenMP use posix threads on a Linux OS?
c- So, if I am working on a Linux OS, can't I use pthread_setschedparam(), or I can but I shouldn't ? ]]></description>
      <link>http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-22965</link>
      <pubDate>Tue, 21 Apr 2009 06:59:50 -0700</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-22965</guid>
    </item>
    <item>
      <title>By Andrew Binstock</title>
      <description><![CDATA[ OpenMP, as I understand it, uses the native threads of the platform it runs on. Which would mean POSIX threads on Linux. 

You can certainly try to figure out which thread in the OpenMP thread pool is doing what work and try to change its priority, but you cannot do this through OpenMP. As a result, you have to do it without OpenMP knowing about it. Hence, you run the risk of changing expectations that OpenMP might have about thread state. This could lead to sudden unexpected failures that you will have great difficulty finding in a debugger.

In summary, you might be able to do it (though I think it would be difficult), but you should not. ]]></description>
      <link>http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-23007</link>
      <pubDate>Tue, 21 Apr 2009 22:59:57 -0700</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-23007</guid>
    </item>
    <item>
      <title>By Dny</title>
      <description><![CDATA[ Hello,

A very good article for describing usage of OpenMP Vs Explicit Threading.
But My question is 
If we want to scale the application performance with increase in core then using explicit threading then what are different aspects we need to consider like data decomposition between all threads.

In other words how could we enable free lunch using explicit threading?

Thanking You,

Regards,
Dny ]]></description>
      <link>http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-23011</link>
      <pubDate>Wed, 22 Apr 2009 03:14:28 -0700</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-23011</guid>
    </item>
    <item>
      <title>By atacoral</title>
      <description><![CDATA[ Thank you for the explanation, point taken. ]]></description>
      <link>http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-23014</link>
      <pubDate>Wed, 22 Apr 2009 03:47:35 -0700</pubDate>
      <guid isPermaLink="true">http://software.intel.com/en-us/articles/choosing-between-openmp-and-explicit-threading-methods/#comment-23014</guid>
    </item>
  </channel></rss>