<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	>
<channel>
	<title>Comments on: STM Compiler</title>
	<atom:link href="http://softwareblogs.intel.com/2007/09/16/stm-compiler/feed/" rel="self" type="application/rss+xml" />
	<link>http://softwareblogs.intel.com/2007/09/16/stm-compiler/</link>
	<description></description>
	<pubDate>Sat, 30 Aug 2008 02:12:35 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.6.1</generator>
		<item>
		<title>By: Joe Crumm</title>
		<link>http://softwareblogs.intel.com/2007/09/16/stm-compiler/#comment-6904</link>
		<dc:creator>Joe Crumm</dc:creator>
		<pubDate>Wed, 19 Sep 2007 21:42:21 +0000</pubDate>
		<guid isPermaLink="false">http://softwareblogs.intel.com/2007/09/16/stm-compiler/#comment-6904</guid>
		<description>Xinmin: Yes indeed, I somehow missed the third sample.

jc</description>
		<content:encoded><![CDATA[<p>Xinmin: Yes indeed, I somehow missed the third sample.</p>
<p>jc</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Xinmin Tian</title>
		<link>http://softwareblogs.intel.com/2007/09/16/stm-compiler/#comment-6899</link>
		<dc:creator>Xinmin Tian</dc:creator>
		<pubDate>Wed, 19 Sep 2007 19:57:33 +0000</pubDate>
		<guid isPermaLink="false">http://softwareblogs.intel.com/2007/09/16/stm-compiler/#comment-6899</guid>
		<description>The sample code #3 we posted uses the POSIX threads and __tm_atomic on Linux platform. 

-Xinmin</description>
		<content:encoded><![CDATA[<p>The sample code #3 we posted uses the POSIX threads and __tm_atomic on Linux platform. </p>
<p>-Xinmin</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Jim Cownie</title>
		<link>http://softwareblogs.intel.com/2007/09/16/stm-compiler/#comment-6865</link>
		<dc:creator>Jim Cownie</dc:creator>
		<pubDate>Tue, 18 Sep 2007 08:00:14 +0000</pubDate>
		<guid isPermaLink="false">http://softwareblogs.intel.com/2007/09/16/stm-compiler/#comment-6865</guid>
		<description>Joe, you can use the TM atomic blocks with any threading subsystem. We're using OpenMP just because it makes it easier to write small, portable pieces of code.

Good luck, please provide more feedback.

-- Jim</description>
		<content:encoded><![CDATA[<p>Joe, you can use the TM atomic blocks with any threading subsystem. We're using OpenMP just because it makes it easier to write small, portable pieces of code.</p>
<p>Good luck, please provide more feedback.</p>
<p>-- Jim</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Joe Crumm</title>
		<link>http://softwareblogs.intel.com/2007/09/16/stm-compiler/#comment-6861</link>
		<dc:creator>Joe Crumm</dc:creator>
		<pubDate>Tue, 18 Sep 2007 03:28:10 +0000</pubDate>
		<guid isPermaLink="false">http://softwareblogs.intel.com/2007/09/16/stm-compiler/#comment-6861</guid>
		<description>Very interesting technology. Looking at your samples, I was wondering if the atomic blocks work with Windows/Linux threads or does one have to use OpenMP constructs for parallelism? (or do I have to learn OpenMP first to use it?)

jc</description>
		<content:encoded><![CDATA[<p>Very interesting technology. Looking at your samples, I was wondering if the atomic blocks work with Windows/Linux threads or does one have to use OpenMP constructs for parallelism? (or do I have to learn OpenMP first to use it?)</p>
<p>jc</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Joseph Seigh</title>
		<link>http://softwareblogs.intel.com/2007/09/16/stm-compiler/#comment-6842</link>
		<dc:creator>Joseph Seigh</dc:creator>
		<pubDate>Mon, 17 Sep 2007 10:20:15 +0000</pubDate>
		<guid isPermaLink="false">http://softwareblogs.intel.com/2007/09/16/stm-compiler/#comment-6842</guid>
		<description>I once did an STM implmentation in Java as an exercise to see what the api and implementation issues would be. Language issues were mainly Java closures and abstract pointer support being really really awkward. C++ has the same issues.

You can solve the issues by putting support in the compiler but this creates portability issues. Since STM is about scalability, you're talking about very large applications. At this point it would require a huge leap of faith and significant resources to port the application to STM even if portability wasn't an issue. Non portability makes it worse.</description>
		<content:encoded><![CDATA[<p>I once did an STM implmentation in Java as an exercise to see what the api and implementation issues would be. Language issues were mainly Java closures and abstract pointer support being really really awkward. C++ has the same issues.</p>
<p>You can solve the issues by putting support in the compiler but this creates portability issues. Since STM is about scalability, you're talking about very large applications. At this point it would require a huge leap of faith and significant resources to port the application to STM even if portability wasn't an issue. Non portability makes it worse.</p>
]]></content:encoded>
	</item>
</channel>
</rss>
