<?xml version="1.0" encoding="iso-8859-1"?><!-- generator="b2evolution/2.4.5" -->
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:admin="http://webns.net/mvcb/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:content="http://purl.org/rss/1.0/modules/content/">
	<channel>
		<title>Brent Baude - Latest comments on Semi-optimized PS3 stages</title>
		<link>http://blogs.gentoo.org/ranger?disp=comments</link>
		<description></description>
		<language>en-US</language>
		<docs>http://backend.userland.com/rss</docs>
		<admin:generatorAgent rdf:resource="http://b2evolution.net/?v=2.4.5"/>
		<ttl>60</ttl>
				<item>
			<title>In response to: Semi-optimized PS3 stages</title>
			<pubDate>Sat, 22 Sep 2007 15:26:21 +0000</pubDate>
			<dc:creator>Fallen- [Visitor]</dc:creator>
			<guid isPermaLink="false">c19191@http://blogs.gentoo.org/</guid>
			<description>Hello, I have my PS3 with Gentoo now :D and I want to optimize CFLAGS to install other things like Gnome.Should I replace mcpu=cell or do I leave mcpu=970?  on:&lt;br /&gt;
http://gentoo-wiki.com/Safe_Cflags#Cell_Broadband_Engine_.28Playstation_3.29&lt;br /&gt;
the recommendation is cell. &lt;br /&gt;
&lt;br /&gt;
Thank you for your help! Best regards.</description>
			<content:encoded><![CDATA[Hello, I have my PS3 with Gentoo now :D and I want to optimize CFLAGS to install other things like Gnome.Should I replace mcpu=cell or do I leave mcpu=970?  on:<br />
http://gentoo-wiki.com/Safe_Cflags#Cell_Broadband_Engine_.28Playstation_3.29<br />
the recommendation is cell. <br />
<br />
Thank you for your help! Best regards.]]></content:encoded>
			<link>http://blogs.gentoo.org/ranger/2007/03/08/semi_optimized_ps3_stages#c19191</link>
		</item>
				<item>
			<title>In response to: Semi-optimized PS3 stages</title>
			<pubDate>Sat, 18 Aug 2007 05:16:54 +0000</pubDate>
			<dc:creator>popper [Visitor]</dc:creator>
			<guid isPermaLink="false">c19022@http://blogs.gentoo.org/</guid>
			<description>hi brent.&lt;br /&gt;
&lt;br /&gt;
just a passing thought as you bring up &lt;br /&gt;
the flags.&lt;br /&gt;
&lt;br /&gt;
i was rereading http://forum.beyond3d.com/showthread.php?s=7c9350bc2751cdda0bb5aee91b329bf7&amp;amp;p=893620#post893620 again and remembered about &lt;br /&gt;
using -funroll-loops to increase speed.&lt;br /&gt;
&lt;br /&gt;
also, while i assume noones yet &lt;br /&gt;
replaced any core PPC linux code with &lt;br /&gt;
Altivec routines?,  perhaps its time &lt;br /&gt;
to see if some optimisations can be found.&lt;br /&gt;
&lt;br /&gt;
it seems such a waste to have the &lt;br /&gt;
Altivec chip sat there (almost) idle &lt;br /&gt;
taking up power and not being used in &lt;br /&gt;
any of the current core OS code.&lt;br /&gt;
&lt;br /&gt;
markos's freevec still seems unused &lt;br /&gt;
anywere obvious in any part of the PPC OS codebase, and yet freevec is full of &lt;br /&gt;
tested code that works for most app &lt;br /&gt;
speed options....&lt;br /&gt;
&lt;br /&gt;
he implyed it could be used to replace &lt;br /&gt;
or supliment lots of core code and get &lt;br /&gt;
a massive speed boost, such a shame &lt;br /&gt;
noone seems to have tryed to run tests&lt;br /&gt;
to see if it can boost the PS3 effort. &lt;br /&gt;
 &lt;br /&gt;
&lt;br /&gt;
for readers, that thread was good for test code&lt;br /&gt;
code and otpimisations to try.&lt;br /&gt;
-----------------&lt;br /&gt;
&lt;br /&gt;
-O0: some basic and minimum optimization &lt;br /&gt;
-O2: strong, low-level optimization that benefits most programs &lt;br /&gt;
-O3: intense, low-level optimization analysis with basic loop optimization &lt;br /&gt;
 &lt;br /&gt;
--------------&lt;br /&gt;
&lt;br /&gt;
i think gcc4 still has broken -03 and still produces slower code than the old ones, but i may be wrong or was it just 4.0.3&lt;br /&gt;
</description>
			<content:encoded><![CDATA[hi brent.<br />
<br />
just a passing thought as you bring up <br />
the flags.<br />
<br />
i was rereading http://forum.beyond3d.com/showthread.php?s=7c9350bc2751cdda0bb5aee91b329bf7&amp;p=893620#post893620 again and remembered about <br />
using -funroll-loops to increase speed.<br />
<br />
also, while i assume noones yet <br />
replaced any core PPC linux code with <br />
Altivec routines?,  perhaps its time <br />
to see if some optimisations can be found.<br />
<br />
it seems such a waste to have the <br />
Altivec chip sat there (almost) idle <br />
taking up power and not being used in <br />
any of the current core OS code.<br />
<br />
markos's freevec still seems unused <br />
anywere obvious in any part of the PPC OS codebase, and yet freevec is full of <br />
tested code that works for most app <br />
speed options....<br />
<br />
he implyed it could be used to replace <br />
or supliment lots of core code and get <br />
a massive speed boost, such a shame <br />
noone seems to have tryed to run tests<br />
to see if it can boost the PS3 effort. <br />
 <br />
<br />
for readers, that thread was good for test code<br />
code and otpimisations to try.<br />
-----------------<br />
<br />
-O0: some basic and minimum optimization <br />
-O2: strong, low-level optimization that benefits most programs <br />
-O3: intense, low-level optimization analysis with basic loop optimization <br />
 <br />
--------------<br />
<br />
i think gcc4 still has broken -03 and still produces slower code than the old ones, but i may be wrong or was it just 4.0.3<br />
]]></content:encoded>
			<link>http://blogs.gentoo.org/ranger/2007/03/08/semi_optimized_ps3_stages#c19022</link>
		</item>
				<item>
			<title>In response to: Semi-optimized PS3 stages</title>
			<pubDate>Sun, 29 Jul 2007 22:01:19 +0000</pubDate>
			<dc:creator>Brent BAUDE [Member]</dc:creator>
			<guid isPermaLink="false">c18993@http://blogs.gentoo.org/</guid>
			<description>In my humble opinion, optimized can be a result of several different things.  I would not consider -0-anything the only way to optimize.  The passing of the 970 flags was what my original announcement was referring to.  Prior to then, as I mentioned in the post, the CFLAGS were not getting set properly in the various stages as they were built, specifically in stage3.  &lt;br /&gt;
&lt;br /&gt;
Just for reference, we're working to become more gcc-4.3 savy so that we can actually begin to pass the &quot;cell&quot; mcpu for increased optimization.  My hope is that will make things a little more interesting.</description>
			<content:encoded><![CDATA[In my humble opinion, optimized can be a result of several different things.  I would not consider -0-anything the only way to optimize.  The passing of the 970 flags was what my original announcement was referring to.  Prior to then, as I mentioned in the post, the CFLAGS were not getting set properly in the various stages as they were built, specifically in stage3.  <br />
<br />
Just for reference, we're working to become more gcc-4.3 savy so that we can actually begin to pass the "cell" mcpu for increased optimization.  My hope is that will make things a little more interesting.]]></content:encoded>
			<link>http://blogs.gentoo.org/ranger/2007/03/08/semi_optimized_ps3_stages#c18993</link>
		</item>
				<item>
			<title>In response to: Semi-optimized PS3 stages</title>
			<pubDate>Fri, 13 Jul 2007 22:08:37 +0000</pubDate>
			<dc:creator>Stroller [Visitor]</dc:creator>
			<guid isPermaLink="false">c18978@http://blogs.gentoo.org/</guid>
			<description>Hi there,&lt;br /&gt;
&lt;br /&gt;
Is -O2 really optimal for the PS3? I have little experience of optimising CFLAGS but considering the PS3 doesn't have so much RAM, might -Os be better? Do you know if anyone has done any tests?&lt;br /&gt;
&lt;br /&gt;
Stroller. </description>
			<content:encoded><![CDATA[Hi there,<br />
<br />
Is -O2 really optimal for the PS3? I have little experience of optimising CFLAGS but considering the PS3 doesn't have so much RAM, might -Os be better? Do you know if anyone has done any tests?<br />
<br />
Stroller. ]]></content:encoded>
			<link>http://blogs.gentoo.org/ranger/2007/03/08/semi_optimized_ps3_stages#c18978</link>
		</item>
			</channel>
</rss>
