Slashdot Mirror
Nvidia Fakes Fermi Boards At GPU Tech Conference
fragMasterFlash writes with this excerpt from SemiAccurate:
'In a really pathetic display, Nvidia actually faked the introduction of its latest video card, because it simply doesn't have boards to show. Why? Because it didn't get enough parts to properly bring them up, much less make demo boards. ... Notice that the three screws that hold the end plate on are, well, generic wood screws. Large flat -head Phillips screws. Home Depot-grade screws that don't even sit flush. If a card is real, you hold it on with the bolts on either side of the DVI connector. Go look at any GPU you have; do you see wood screws that don't mount flush or DVI flanking bolts? ... If you look at the back of the fake Fermi, [from this PC Watch picture], you can see that the expected DVI connector wires are not there, just solder-filled holes. No stubs, no tool marks from where they would be cut out. Basically, the DVI port isn't connected to anything with solder, so they had to use screws on the plate."
Little update found on this article: http://www.fudzilla.com/content/view/15798/34/
Having built many a prototype board in my day I can tell you I have utilized all manner of odds and ends including not only wood screws but wood as well - I don't think it means the card is a fake, it may be an engineering prototype or a software development board or whatever. I personally don't see anything in the photos that screams to me "FAKE" !
Those do not look like wood screws to me. not even close. They appear way too small and they dont appear to be counter sunk. Go to lowes and see if you can find any wood screws that match. They do remind me of the ones used to mount motherboards or for mounting 5 1/4 and 3 1/2 drives. And my geforce 7800 gtx has those stand offs with both dvi connectors. I didnt realize that was novel.
Notice the source. The site semi accurate is run by a guy, Charlie Demerjian, who was fired from The Inquirer for a number of reasons, including making shit up. In particular, this guy has it in for nVidia. I don't remember the details of why he has it in for them, I think they cut him out of the information loop because he leaked some info he wasn't supposed to. Regardless, he hates nVidia and does everything he can to make them look bad. In his case, that includes just straight out making shit up.
So that's why he's making such a big deal of this being a fake. He wants it to be fake because, well I dunno, I guess that is somehow a "win" in his mind.
Personally I find it funny since companies do mockups for demonstrations all the time. Wouldn't at all surprise me if the card he was holding was such a mockup.
At any rate as with most things in life, you want to check sources, and on the Internet that is doubly true. Some people have an agenda to push and will... modify, to put it mildly, the truth to suit their needs. I though we'd all be well aware of that after all the political BS of recent years :P.
OK. He got the screws wrong. Big deal. Try reading the article.
Some of the things NVidia did on their "working board" include: covering the SLI connector, not having the DVI connector wires go through vias, place the PCI-E power connectors wrong from where the board shows they should be, cut off the end of the board with a saw right though where there was more stuff, have half the vents on the back of the card completely blocked...
This isn't just "they used the wrong screws", this is "total fake that couldn't possibly work". Saying it was a working board was a total lie.
Comment forecast: Bits of genius surrounded by a sea of mediocrity.
I believe it forces asking for the root account for run level 1.
Dear GP: If that's the case, try sticking a "2" on the end of your boot params (ie. select the line, hit e, edit the line with the mention of /boot on it, and add a " 2" to the end, then hit b to boot).
I'm sorry, but this screws in the end plate are not wood screws. I work with wood on a regular basis and spent over a decade as a manufacturing engineer in electronics manufacturing. These screws are common assembly screws in electronics, not furniture. It is also common to leave off components on proto, demo or even production PCAs. Many circuits are designed to be partially populated using a single board with various levels of features. As far as "First Silicon" is concerned, if a chip is working to spec, there is no reason not to use it. While this may not be a production board (I have no way of knowing), it could be a working prototype. I'm beginning to think the writer is a bit of a drama queen.
Since when do "generic wood screws" come chromed and without a tapered head? Has the author actually seen a generic wood screw before?
Summary is crap, article is slashdotted. Next, please.
Eagles may soar, but weasels don't get sucked into jet engines.
The author has a grudge against nVidia. Read some of his past work paying close attention to how many times he has been wrong before.
Generally, the very first chips that come back are known as A0. If only metal layer changes need to be done to fix issues, further revisions would be known as A1, A2, A3.
If a full chip spin is done (or anything more than a metal fix), you start over with B0, and further metal spins are B1, B2, B3...
So, A1 means they probably got A0 back and it had enough issues to later call for doing a metal spin. So one entire premise of this article is quite like false. Ask anyone who has ever done work with real hardware and they'll tell you the same thing about how silicon revisions are named. I've never seen a company that has started out with their first silicon called A1.
Just like the first Amiga prototype. Basically just wirewrapped boards cabled together. Actually used at a trade-show, but hidden under a table.
/ The Arrow
"How lovely you are. So lovely in my straightjacket..." - Nny
Grudge or not, at least this time it appears that he got the mockup thing right. Meanwhile, if you feel the need to attack the messenger to discredit the message (are you even able to refute the accusations?) then it says more about you than about the messenger.
Slashdot, fix your code or at least hire someone who is competent at it to do it for you.
Just on a point of information, the "PR drone" was actually Jen-Hsun Huang, company president and CEO. If the card he was waving around was a mockup, he surely knew about it.
Not that I see that it matters. Huang openly admitted they're at least "a few months" away from production, and it was strongly implied at the press conference that GeForce models would come before Quadro and Tesla (lots of airy talk about high-end customers running to different cycles). It was a cute spot that this was, most likely, not a real card, but it's not as if it blows open a huge lie.
Agreed. Speaking as an Engineer who has worked on hardware between first manufacture and first sale, this is no big deal. Prototypes are expensive, and usually not pretty. And you just don't let the folks in marketing (or the executives) touch your prototypes - you usually don't have enough to use yourself, much less to loan out for a few days, and risk getting broken at the hands of photographers and the like who don't take proper precautions in handling the boards. Not to mention, they look pretty ugly - faceplates and covers may not be ready yet (or even designed), they might have a few mod wires (though that happens occasionally on shipping products too), and the ones that the engineers work with probably have lots of extra wires hanging off in every direction. If I work on a new board, I may have to pick a random heatsink out of the box of spare parts, or to have it built without all the mechanical pieces (like standoffs and covers) because they get in the way of me doing my job.
So when the marketing types want something to photograph or show off to the world? I pick out something that's dead as a doornail. That's one of the few good uses for all the boards I kill! (Or I might give it to the mechanical engineer for shock and vibration testing, or packaging drop tests, or things like that.)
So what's the big deal? If they start talking about their latest greatest product too long before it's ready, that'll just keep their customers from buying NVIDIA's current products. Many a company has been bitten (or even put out of business) by doing that; see the Osborne Effect. Obviously, they think it's far enough along that they can start talking about it without risking their revenue stream.
Those original prototype Amigas (with the 5.25" drives and the expansion chimney) were in metal cases, not wooden. These generally had the older chips, Portia instead of Paula, Daphne instead of Denise, as I recall.
This is a natural part of product development... final, consumer-ready products don't spring to life fully borne, and in the case of something like the Amiga, the developer's units (which only went out to a handful of developers) were designed to get hardware into hands as quickly as possible, rather than waiting for final system details, final silicon, shipping OS, etc. Developers who get these things understand this well.
As for the nVidia thing... those were hardly wood screws, those are chromed machine screws most likely. I've been a carpenter and cabinetmaker for longer than a computer designer... and that, pretty long. Does this guy even know what a wood screw looks like?
And as well... demonstrating a "Fully Functional Fermi" is hardly the same as claiming to be demonstrating a production ready board. They're not necessarily claiming a final PCB design, final hardware around that PCB, completed drivers, or even finial silicon. Anyone in the actual hardware business would understand this. People who don't understand the development process should maybe stick to videogame reviews.
-Dave Haynie