<?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 on: Memory system rewrite</title>
	<atom:link href="http://www.lxdream.org/news/archives/113/feed" rel="self" type="application/rss+xml" />
	<link>http://www.lxdream.org/news/archives/113</link>
	<description>lxdream news and updates</description>
	<lastBuildDate>Tue, 22 Jun 2010 15:26:49 -0700</lastBuildDate>
	<generator>http://wordpress.org/?v=2.8.5</generator>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
		<item>
		<title>By: Gondos</title>
		<link>http://www.lxdream.org/news/archives/113/comment-page-1#comment-211</link>
		<dc:creator>Gondos</dc:creator>
		<pubDate>Mon, 19 Jan 2009 21:27:38 +0000</pubDate>
		<guid isPermaLink="false">http://www.lxdream.org/news/?p=113#comment-211</guid>
		<description>Hi Nathan,
Good to see some steady progress on your project, I know it&#039;s hard to keep focused; I sadly gave up working on the DC emulator I was toying with.
If you want some tips for your dynrec, I came up with a 1-pass DRC that gave pretty decent code. Given the instruction set of the SH4, constant propagation is a great addition (and not too difficult to add) to the register allocator and can really reduce the size of the generated code (even further if you consider &quot;MOV.x @(disp, PC), Rn&quot; instructions to be generating constants). Also lazy flushing the T bit can virtually remove the need for setx operations in a basic block and allows for some sweet &quot;cmp reg1,reg2/jx&quot; combos (especially if you allow your basic blocks to continue through branches not taken).
Well, that was just my 2 cents; I&#039;m sure you already have nice plans for your dynamic recompiler :)
Keep up the good work,
Gondos</description>
		<content:encoded><![CDATA[<p>Hi Nathan,<br />
Good to see some steady progress on your project, I know it&#8217;s hard to keep focused; I sadly gave up working on the DC emulator I was toying with.<br />
If you want some tips for your dynrec, I came up with a 1-pass DRC that gave pretty decent code. Given the instruction set of the SH4, constant propagation is a great addition (and not too difficult to add) to the register allocator and can really reduce the size of the generated code (even further if you consider &#8220;MOV.x @(disp, PC), Rn&#8221; instructions to be generating constants). Also lazy flushing the T bit can virtually remove the need for setx operations in a basic block and allows for some sweet &#8220;cmp reg1,reg2/jx&#8221; combos (especially if you allow your basic blocks to continue through branches not taken).<br />
Well, that was just my 2 cents; I&#8217;m sure you already have nice plans for your dynamic recompiler <img src='http://www.lxdream.org/news/wp-includes/images/smilies/icon_smile.gif' alt=':)' class='wp-smiley' /><br />
Keep up the good work,<br />
Gondos</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Lahr</title>
		<link>http://www.lxdream.org/news/archives/113/comment-page-1#comment-209</link>
		<dc:creator>Lahr</dc:creator>
		<pubDate>Fri, 16 Jan 2009 14:32:03 +0000</pubDate>
		<guid isPermaLink="false">http://www.lxdream.org/news/?p=113#comment-209</guid>
		<description>Awesome. You can NEVER go wrong with a performance update.</description>
		<content:encoded><![CDATA[<p>Awesome. You can NEVER go wrong with a performance update.</p>
]]></content:encoded>
	</item>
</channel>
</rss>
