CE-Oh no he didn't! Part LVIII: NVIDIA CEO doesn't know what Larrabee is, doesn't care
By Richard Lawler 
posted August 27th 2008 3:22PM
posted August 27th 2008 3:22PM
NVIDIA CEO Jen-Hsun Huang just can't resist throwing more jabs at Intel, distracting the inaugural NVISION crowd from Battlestar Galactica star Tricia Helfer with the claim that "Larrabee hasn't shipped so you don't know what it is and I don't know what it is." The fact that we do know what it is -- a next-gen hybrid CPU / GPU -- shouldn't be a concern according to Huang, because "By the time it does ship, Nvidia's technology will be so far advanced it won't matter." Besides stuffing Usain Bolt-type speed into a GPU the company will keep busy working on its WinMo smartphone hardware, and software for the not-exactly-Atom-killing VIA Nano, but forget about that rumored x86-compatible hardware 'cuz, as Jen-Hsun reminds us, "the Internet doesn't run on x86." For a company that lacks innovation, is "a joke," and at least four years behind, Intel must be doing something right, because the competition can't keep its name out of their mouths.
Hey, Nvidia? When Intel starts creating chipsets for YOUR exceedingly popular CPUs, let me know. Until then, it's the other way around still, ain't it?
Yeah so if they think that poorly about Intel I wonder how they feel about AMD.
As for my personal opinion, I don't really care one way or another and neither should anyone else, neither company has a product to market yet and the winner will be the one company first to market with a processor that can do 100% real-time ray tracing (0% rasterization) in a game at decent resolutions - anyone who disagrees with me is someone who would buy a Phenom processor now or a GTX280 over a 4870.
The unfortunate thing is that 1. 98% of the people reading that comment wont know what it means and 2. it wont happen. Whatever is the cheapest will win. If history is anything to go by, the best product is rarely the winner.
anyone who disagrees with me is someone who would buy a Phenom processor now or a GTX280 over a 4870.
You mean a fanboy?
methinks the Huang doth protest too much.
8hr+ battery life for the masses. Wake me then. Will donate some shekels to the winner.
Interesting that I haven't heard one positive comment about the CEO. Every company screws up sometimes. The big Microsoft's and RIM's didn't just become monsters overnight. Everyone seems to have forgotten what this company has done for the gaming community. They may be going through a hickup, but Huang is widely viewed as one of the best in the business, as is the company itself. All this talk of GPU's going away is nonsense. If this does ever happen, it won't be before aliens take over the earth. There is still incredible growth potential in visual computing. In ten years we will all look back at the PS3 and laugh at ourselves for thinking that its graphics were "advanced" the same way that we look back at Atari now.
I don't see the Larrabee doing that well. Yet Intel are throwing a lot of money at it. It is an interesting concept, but they are trying to reinvent the wheel. While it may be an excellent GPGPU, I don't hold much hope of it competing against dedicated GPUs with specific enhancements for DirectX and OpenGL.
Intel's approach seems to be as open for general purpose use as possible (which should be good competition for nVidia's Tesla and ATI's FireStream), but it takes care of DirextX and OpenGL specific functions in software. Doing this in software is going to come at a large performance hit, and it is arguable that keeping the x86 instruction set for graphics tasks is a waste of die space. Intel's solution at this stage just seems to be to throw more cores at the problem, but that will only do so much. As many as 25 cores are required to handle most current games, yet it seems an initial 32 core version is planned, and initial figures also indicate that will be a 300W TDP beast. ATI and nVidia's offerings are power hungry too, but Larrabee promises to be moreso.
As far as uptake goes, Intel aren't going to have an easy time convincing gamers that their offerings are far superior to ATI and nVidia's. Given Intel's current intergrated graphics offerings and their performance (or lack of), I think Intel are going to have a PR battle here. Convincing game developers to take advantage of the Larrabee's widely parallel (and thus difficult to program) architechure is going to be a problem, even if it is an x86 instruction set.
I can see Intel hoping that the graphics world will turn to it's architecture en masse. They want to upset ATI and nVidia, and in order to do that, Intel will easily throw massive ammounts of money at this, and I can see Intel selling Larrabee cards at a huge loss to do so. Of course that will send ATI and nVidia to the FTC and DOJ, crying "Antitrust!" ;)
Huang, where is my automobile?
He looks like he just caught a throwing star hurled at his face. That's his "aww hell no, you just fucked up now!" look.
First off, I too thought of "Dear Leader" when I saw that picture.
Second I highly doubt that the first generation of Larabee's will be available as a separate board, they will be "just" that much more than the previous generation of onboard designs so that all the major re-sellers can use those boards and they will spec out so that the final system will be able to play a few games at decent resolutions (labeled on the box/flyer etc.)
Done THAT way Intel is able to say that their knew design is ALREADY installed on most new computers sold today.
After probably 6mth to a year THEN they will probably go with the stand alone variety, 1st the current onboard design with some more memory and then an "upgrade" to that.
I doubt that a mid-high GPU will be out and competitive for at least a year (if ever-maybe because at that point they would like to at least match AMD on the video side as well)
Arrogant idiots! LOL! You guys have no idea how powerful your graphic cards are.
The fastest PC processors - Intel quad-core QX9775 performs over 51 GIGAFLOPS. GPUs in PCs are considerably more powerful in pure FLOPS. For example, in the GeForce 8 Series the nVidia 8800 Ultra performs around 576 GIGAFLOPS on 128 Processing elements. This equates to around 4.5 GIGAFLOPS per element, compared with 2.75 per core for the Intel quad-core QX9775 . It should be noted that the 8800 series performs only single precision calculations, and that while GPUs are highly efficient at calculations they are not as flexible as a general purpose CPU. On August 12, 2008 AMD released the ATI Radeon HD 4870X2 graphics card with two Radeon R770 GPUs totalling 2.4 teraFLOPs(1 teraflops = 1000gigaflops)
In computing, FLOPS (or flops or flop/s) is an acronym meaning FLoating point Operations Per Second. The FLOPS is a measure of a computer's performance, especially in fields of scientific calculations that make heavy use of floating point calculations, similar to instructions per second.
http://www.dailytech.com/NVIDIAs+NVISION+2008+JenHsun+Huang+Keynote/article12774.htm
Ir really don't see why we schould drop what we have achieved (decent looking Games without too much waste of resources) just to let Intel change the rules of the game and get a competitive GPU again (as well as an argument to sell 400 core cpu's, which would be pretty much worthless, otherwise.
Seriously, guys. Pretty much every ray-tracing image i saw had a strong plastic look and didn't really outperform state-of-the-art rasterization images, except for refletions. Is this worth the waste of erssources?