<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom">
	<channel>
		<title>AMD working on new multi-GPU technology</title>
		<description>Discuss AMD working on new multi-GPU technology</description>
		<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology</link>
		<lastBuildDate>Fri, 24 May 2013 23:49:51 +0100</lastBuildDate>
		<generator>JComments</generator>
		<atom:link href="http://www.fudzilla.com/component/jcomments/feed/com_k2/21625/10" rel="self" type="application/rss+xml" />
		<item>
			<title>Thisguy says:</title>
			<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-15118</link>
			<description><![CDATA[nVidia's Optimus is switchable graphics, which means they can be activated when more gpu power is needed, and switched to igp when it is not, i think. This saves power in mobile... ect... Although, in the graph it is Llano + turks, and I'm sure turks has more gpu power than 5550. That would skew the results a little bit...]]></description>
			<dc:creator>Thisguy</dc:creator>
			<pubDate>Wed, 26 Jan 2011 00:29:13 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-15118</guid>
		</item>
		<item>
			<title>Naterm says:</title>
			<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-15018</link>
			<description><![CDATA[Didn't nVidia already do this and call it 'Optimus'? Except it would work with intel's graphics as well?]]></description>
			<dc:creator>Naterm</dc:creator>
			<pubDate>Tue, 25 Jan 2011 11:33:48 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-15018</guid>
		</item>
		<item>
			<title>Sinister213 says:</title>
			<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-15014</link>
			<description><![CDATA[amd will never be within pissing distance of intel.]]></description>
			<dc:creator>Sinister213</dc:creator>
			<pubDate>Tue, 25 Jan 2011 09:10:52 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-15014</guid>
		</item>
		<item>
			<title>asianbullet says:</title>
			<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14996</link>
			<description><![CDATA[THOMASG - I fully take your point & am aware of that. But I am hoping that they would take another set of graphical processes aside from the rendering of frames, to implement "cross talk" between the APU & the discrete GPU. It would be ideal if the APU graphical side could take care of say 'physics' whilst the discrete GPU can take care of the rendering. Although there might be a bottle-neck in the sense that the computational CPU side of the APU might not be able to keep up.]]></description>
			<dc:creator>asianbullet</dc:creator>
			<pubDate>Tue, 25 Jan 2011 00:42:32 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14996</guid>
		</item>
		<item>
			<title>Bl0bb3r says:</title>
			<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14979</link>
			<description><![CDATA[Something like this: http://en.wikipedia.org/wiki/CrossFire#Second-generation_.28Software_CrossFire.29 ? Probably and improved variant...]]></description>
			<dc:creator>Bl0bb3r</dc:creator>
			<pubDate>Mon, 24 Jan 2011 20:53:19 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14979</guid>
		</item>
		<item>
			<title>Pyronick says:</title>
			<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14978</link>
			<description><![CDATA[AMD needs to implement the PCI Express controller in the APU-SOC. Then design a new interface which is signal compatible with PCI Express, but also transfers specific CrossFire data.]]></description>
			<dc:creator>Pyronick</dc:creator>
			<pubDate>Mon, 24 Jan 2011 20:43:37 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14978</guid>
		</item>
		<item>
			<title>Bl0bb3r says:</title>
			<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14969</link>
			<description><![CDATA[@asianbullet, probably Llano's GPU will be on the level of Redwood? @Simic, "We are quite sure that AMD will come up with something similar to Nvidia Optimus and the discrete GPU will be able to power down or up depending on the load." I'm quite sure you shouldn't be sure, and that AMD already has its Optimus, for some time now and that's not the issue here. Probably they need something better to Crossfire, something that scales well through the bus itself in the absence of a Crossfire connector.]]></description>
			<dc:creator>Bl0bb3r</dc:creator>
			<pubDate>Mon, 24 Jan 2011 19:31:40 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14969</guid>
		</item>
		<item>
			<title>thomasg says:</title>
			<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14968</link>
			<description><![CDATA[The reason they only made certain GPU's compatible is that if you pair it with a discrete GPU that is much faster than the integrated one, it's going to hold back the discrete GPU. If you ran a 6900 series card with the 6310 in the APU, it would be slower than just using the discrete graphics because the 6950/70 would always be waiting on the 6310 to rende the other frame or the othe half of that frame. If you have, say, a 5450, it would probably be faster at AFR between it and the 6310 core than if it rendered eveything by itself. You can only be twice as fast as your slowest chip in a dual GPU setup.]]></description>
			<dc:creator>thomasg</dc:creator>
			<pubDate>Mon, 24 Jan 2011 19:30:08 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14968</guid>
		</item>
		<item>
			<title>asianbullet says:</title>
			<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14943</link>
			<description><![CDATA[AMD certainly moving in the right direction with introducing "co-op" between te APU & discrete GPU. However, the slide above I'm quite dissapointed in the comparison of parts AMD has chosen. The last config is inconsistant, as the previous configs uses the Radeon 5550 as the discrete GPU, whereass the last config uses the 'Turks' GPU which is the Radoen 6670. So there's obviously going to be an advantage there. I'm guessing if they're going with the previous incarnation on Hybrid Crossfire X. Then only certain discrete GPU's will be compatible with the new APU's. Now that would be very disapointing to any business looking to upgrade to APU but wanting to keeo their existing discrete GPU.]]></description>
			<dc:creator>asianbullet</dc:creator>
			<pubDate>Mon, 24 Jan 2011 15:41:43 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14943</guid>
		</item>
		<item>
			<title>Naterm says:</title>
			<link>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14928</link>
			<description><![CDATA[Sorry, but "PC-related corportations" as you put it, sounds like a much bigger market than the PC gaming populace, which is quite small. AMD needs to get a decent number of server design wins with Bulldozer. I expect they will, I wouldn't think there would be a repeat of the Opteron 6100/4100 fiasco. On the other hand, I don't seem them getting nearly as many designs as intel will get with it's combination of Westmere-EX (just slots into current Nehalem-EX designs) and Sandy Bridge-EP.]]></description>
			<dc:creator>Naterm</dc:creator>
			<pubDate>Mon, 24 Jan 2011 13:37:17 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/21625-amd-working-on-new-multi-gpu-technology#comment-14928</guid>
		</item>
	</channel>
</rss>
