<?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 28nm graphics coming in 2H 2011</title>
		<description>Discuss AMD 28nm graphics coming in 2H 2011</description>
		<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011</link>
		<lastBuildDate>Sun, 26 May 2013 10:29:39 +0100</lastBuildDate>
		<generator>JComments</generator>
		<atom:link href="http://www.fudzilla.com/component/jcomments/feed/com_k2/20891/10" rel="self" type="application/rss+xml" />
		<item>
			<title>Naterm says:</title>
			<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9749</link>
			<description><![CDATA[For the record, AMD's share price is sitting at a pathetic $7.50 a share with a market cap of just 5.09 billion dollars. nVidia has a share price of $13.75 and a market cap of just under $8 billion dollars. nVidia's p/e ratio is also over eight times higher than AMD's. That could be a bad thing. It could be speculation that nVidia will be able to really push into new markets. They could grow massively, it might be a bubble. We're starting to find out and it's looking like nVidia will be very successful pushing into HPC. HPC is definitely a growth market as well, it's starting to make it's way outside of academia and into business. You don't even want to compare AMD and intel.]]></description>
			<dc:creator>Naterm</dc:creator>
			<pubDate>Sat, 20 Nov 2010 12:00:05 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9749</guid>
		</item>
		<item>
			<title>Naterm says:</title>
			<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9748</link>
			<description><![CDATA[Yeah buddy, I've only owned AMD stock for most of the past decade. My interest in their performance goes a lot further than if their GPUs are good for playing games or not. As a shareholder, I wish AMD wouldn't waste it's time going after the high-end desktop GPU market. It matters less and less each quarter. nVidia's strategy of shifting $500 gaming GPUs into $2000 HPC accelerators is a good business plan. Not only are the margins per ASIC sold a lot higher, but they can eventually sell a lot more high-end chips. That's why I wish AMD would get their GPGPU shit together. Maybe their share price would actually significantly rise for a change. Pushing Bobcat into the mega scale cloud server market would also be a really sound financial move.]]></description>
			<dc:creator>Naterm</dc:creator>
			<pubDate>Sat, 20 Nov 2010 11:51:05 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9748</guid>
		</item>
		<item>
			<title>jedm33 says:</title>
			<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9744</link>
			<description><![CDATA[likewise...but you are the worst... who needs double precision and ecc support in playing 3d games??? its just a waste of power and die space, and who cares about HPC...the gamers??? thanks but no thanks... BTW did nvidia removed ecc support in thier gtx580??? YES they did... lolx!!!]]></description>
			<dc:creator>jedm33</dc:creator>
			<pubDate>Sat, 20 Nov 2010 08:03:32 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9744</guid>
		</item>
		<item>
			<title>Naterm says:</title>
			<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9739</link>
			<description><![CDATA[Further more, just as with ECC, double precision performance is extremely important in HPC applications. Look at the jump in adoption between the GT200-based Tesla cards and the GF100-based Tesla cards. It's likely ten fold. It's all down to the much higher DP throughput and ECC support.]]></description>
			<dc:creator>Naterm</dc:creator>
			<pubDate>Sat, 20 Nov 2010 04:18:26 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9739</guid>
		</item>
		<item>
			<title>Naterm says:</title>
			<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9738</link>
			<description><![CDATA[ECC isn't needed? The stupid shit fanboys will say. ECC is definitely needed. Just look at some of the recent findings on error rates in DRAM. It's getting to the point where regular desktop machines probably need ECC for their main memory. I don't know when we'll see this, but it's likely going to happen.]]></description>
			<dc:creator>Naterm</dc:creator>
			<pubDate>Sat, 20 Nov 2010 04:15:07 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9738</guid>
		</item>
		<item>
			<title>hoohoo says:</title>
			<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9710</link>
			<description><![CDATA[One last comment. Promise ;-) If you build a GPGU accelerated super, or are making commercial 3rd party software, or like Apple are using OpenCL in the OS package, you will probably go to NV or AMD and pay for engineering expertise to help in programming the GPUs. So, NV & AMD make money (at least should) out of GPGPU that ppl like me will probably never pay. FWIW I work in academe and have spent some of my time getting scientists to take this GPGPU stuff seriously. A lot of the $ for AMD/NV from GPGPU may not be obvious. A lot of the $ may be masked by NDAs.]]></description>
			<dc:creator>hoohoo</dc:creator>
			<pubDate>Sat, 20 Nov 2010 01:52:49 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9710</guid>
		</item>
		<item>
			<title>hoohoo says:</title>
			<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9708</link>
			<description><![CDATA[Tough call about AMD vs NV in GPGPU going forward. CUDA got the jump on CAL in terms of acceptance, and CUDA only runs on NV. NV & AMD GPUs are both very good at parallel FP computations with simple dependencies between threads of execution. It is such a young market, 1 or 2 product generations for AMD to match NV's CUDA effort is not a very long time. I have an irrational bias against OpenCL (and OpenMP). I come out of the POSIX threads world, what can I say, and CUDA is detailed and meticulous like POSIX threads. I'm comfortable with that. Nevertheless OpenCL may still totally level the playing field.]]></description>
			<dc:creator>hoohoo</dc:creator>
			<pubDate>Sat, 20 Nov 2010 01:27:45 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9708</guid>
		</item>
		<item>
			<title>hoohoo says:</title>
			<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9705</link>
			<description><![CDATA[C2050 & 480 are built using the same GF100 chip: the C2050 manages about 4x the DP FP throughput than the 480. Maybe NV burns some gates to make the video cards? C2050 uses ECC RAM, not sure what that costs. C2050 comes with 4GB RAM, the 480 with max 1.5GB - so the size of on-card data is different, that is a product differentiator too. NV is segmenting the market, a venerable business practice. Also, my mistake, C1060 is I think actually the chip from the GTX280, not G80.]]></description>
			<dc:creator>hoohoo</dc:creator>
			<pubDate>Sat, 20 Nov 2010 00:57:43 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9705</guid>
		</item>
		<item>
			<title>loadwick says:</title>
			<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9700</link>
			<description><![CDATA[Are the 'normal' non-HPC chips really that much slower/crippled than the HPC counter parts? just seems such a massive premium to pay, ie you could have 5x GTX580's working away for the price of one C2050??]]></description>
			<dc:creator>loadwick</dc:creator>
			<pubDate>Sat, 20 Nov 2010 00:28:46 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9700</guid>
		</item>
		<item>
			<title>hoohoo says:</title>
			<link>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9678</link>
			<description><![CDATA[A comment about the "impending demise of nVidia". Video cards based on the 100 series chips may sell at a loss, but I very much doubt that nVidia's HPC products do: C2050 card, GF100 series chip = approx $2500 C1060 card, G80 variant IIRC = approx $1200 Them's not small markups!]]></description>
			<dc:creator>hoohoo</dc:creator>
			<pubDate>Fri, 19 Nov 2010 21:52:56 +0100</pubDate>
			<guid>http://www.fudzilla.com/home/item/20891-amd-28nm-graphics-coming-in-2h-2011#comment-9678</guid>
		</item>
	</channel>
</rss>
