<?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"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	>
<channel>
	<title>Comments for Real Time Software Rendering</title>
	<atom:link href="http://code.dawnofthegeeks.com/comments/feed/" rel="self" type="application/rss+xml" />
	<link>http://code.dawnofthegeeks.com</link>
	<description>Doing Things the Hard Way</description>
	<pubDate>Sun, 14 Mar 2010 01:26:28 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.7.1</generator>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
		<item>
		<title>Comment on FYI: New Vs Malloc by Seth Morris</title>
		<link>http://code.dawnofthegeeks.com/2009/05/04/fyi-new-vs-malloc/comment-page-1/#comment-53</link>
		<dc:creator>Seth Morris</dc:creator>
		<pubDate>Sun, 05 Jul 2009 23:13:59 +0000</pubDate>
		<guid isPermaLink="false">http://code.dawnofthegeeks.com/?p=52#comment-53</guid>
		<description>I the performance difference is vital to your code, then you probably need to look at the design anyway. New and delete can be overridden on a more granular level than malloc and free and mixing the two (new/delete for for objects, malloc/free for intrinsic types) makes for hard-to-maintain code.

In most code you'd rather wrap intrinsic allocations in "smart"(ish) objects anyway so they can be deleted if an exception is thrown (malloc'd memory won't) and you don't have to remember to add a free to every return statement if you use multiple returns.

Most Win32 code, even server-side, spends most of its time waiting on users, IO completion, etc., and every code I've seen that had a serious problem with allocation performance was better solved by managing heaps, getting allocations out of loops, moving to stack-based intrinsics, etc.

Having said that, I *do* like alloca(), which cleans itself up because it just opens some stack space.</description>
		<content:encoded><![CDATA[<p>I the performance difference is vital to your code, then you probably need to look at the design anyway. New and delete can be overridden on a more granular level than malloc and free and mixing the two (new/delete for for objects, malloc/free for intrinsic types) makes for hard-to-maintain code.</p>
<p>In most code you&#8217;d rather wrap intrinsic allocations in &#8220;smart&#8221;(ish) objects anyway so they can be deleted if an exception is thrown (malloc&#8217;d memory won&#8217;t) and you don&#8217;t have to remember to add a free to every return statement if you use multiple returns.</p>
<p>Most Win32 code, even server-side, spends most of its time waiting on users, IO completion, etc., and every code I&#8217;ve seen that had a serious problem with allocation performance was better solved by managing heaps, getting allocations out of loops, moving to stack-based intrinsics, etc.</p>
<p>Having said that, I *do* like alloca(), which cleans itself up because it just opens some stack space.</p>
]]></content:encoded>
	</item>
	<item>
		<title>Comment on About by Java Lesson 1: Initializing the screen and creating some static &#171; Real Time Software Rendering</title>
		<link>http://code.dawnofthegeeks.com/about/comment-page-1/#comment-2</link>
		<dc:creator>Java Lesson 1: Initializing the screen and creating some static &#171; Real Time Software Rendering</dc:creator>
		<pubDate>Wed, 06 May 2009 21:45:15 +0000</pubDate>
		<guid isPermaLink="false">http://code.dawnofthegeeks.com/?page_id=2#comment-2</guid>
		<description>[...] About [...]</description>
		<content:encoded><![CDATA[<p>[...] About [...]</p>
]]></content:encoded>
	</item>
</channel>
</rss>
