• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Fermi In Trouble ?

its pretty serious in that nVidia has badmouthed pretty much every one like some foul mouthed brat on the street corner and so are pretty much out of alternative markets except for super computing and graphics card. heavily rumoured they arent even in running fo next gen consoles, chipset business pretty much gone and now looks like they cant keep incoem from discrete graphics. the laptop business is also not looking great in future.
begining to think they badly need new management
 
Drunkenmaster, in that case would it not be better to skip this gen and continue with the 2xx until they sort out production? what way has the archeticture changed that they need a new part for DX11 instead of implimenting it on the 2-series?

its pretty serious in that nVidia has badmouthed pretty much every one like some foul mouthed brat on the street corner and so are pretty much out of alternative markets except for super computing and graphics card. heavily rumoured they arent even in running fo next gen consoles, chipset business pretty much gone and now looks like they cant keep incoem from discrete graphics. the laptop business is also not looking great in future.
begining to think they badly need new management

Wonder what their game plan is when AMD and intel start to move towards all-in-one GPU-CPU.
 
Last edited:
Wonder what their game plan is when AMD and intel start to move towards all-in-one GPU-CPU.

The have two options, firstly they will push their 'GPGPU' into markets that have been traditionally dominated by CPU's and leave the consumer market for Intel and AMD to slug it out between them. There is already evidence of this as Nvidia have already started that it's GT300 line will be targeting businesses first.

Their second option is to follow Intel's and AMD's line and design their own x86 chip with a GPU integrated into it. Although Nvidia doesn't hold it's own x86 licence it does have a controlling stake in VIA who do own a x86 licence so don't be surprised to see new VIA chips coming out with integrated GPU's.
 
There's been a rumour about them getting into the CPU business for quite some time... http://arstechnica.com/hardware/new...umors-resurfaceis-it-a-plausible-scenario.ars

They can't produce an x86 chip because they refuse to pay Intel the licence for Intel's IPR for the x86. If they attempt this then Intel will sue and win.

AMD have a licence from Intel for x86 and SSE, and Intel have a licence from AMD for x86-64 extensions.

Unless nV persuade microsoft to create a new code base for windows and all their products on their own bespoke instruction set (cat in hell's chance).
 
The key to all this is the compiler, or more accurately the new forms of compiler (JIT based on the available processors). GCC for example is so old that's positively prehistorical in it's approach to compilation - I look at the compilation with respect to adding GPGPU to it and decided that the whole way that GCC attempts it is wrong and is specifically tailored for SIMD operations rather than SPMD.
 
They can't produce an x86 chip because they refuse to pay Intel the licence for Intel's IPR for the x86. If they attempt this then Intel will sue and win.

AMD have a licence from Intel for x86 and SSE, and Intel have a licence from AMD for x86-64 extensions.

Unless nV persuade microsoft to create a new code base for windows and all their products on their own bespoke instruction set (cat in hell's chance).

I wouldn't say its entirely out of the question - theres an NT kernel compiled for RISC.
 
I wonder if there reading all forum what talk about the new cards there bringing out.

Possibly laughing and saying "LOL there falling into our trap, we just like keeping them waiting"
 
Drunkenmaster, in that case would it not be better to skip this gen and continue with the 2xx until they sort out production? what way has the archeticture changed that they need a new part for DX11 instead of implimenting it on the 2-series?



Wonder what their game plan is when AMD and intel start to move towards all-in-one GPU-CPU.

Nope, literally because they've halted production, the gt200 series was also underclocked, its around 10% under the targeted clocks, largely because its switch to 55nm and 65nm wasn't as well as wanted, because TSMC had issues there. They stopped production because to compete with 5770 parts, they literally make a loss at the same costs for similar performance but less features.

AS for Larabee, its not a joke, the only silicon being seen is test silicon thats not whats to be released, they are essentially privately making say a 2900xt/3870, with the intention of releasing a 4870/5870 powered card at a later date. They are making, and learning and improving and doing it behind closed doors.

The problem with people lending Nvidia money in the short term is, in the long term their volume sales will dissappear, intergrated and chipset is down the pan due to on cpu's with intergrated gpu's in them. Nvidia can't ever compete with them, that leaves them far smaller shipping volume in the future, meaning their R&D costs are recouped on smaller and smaller numbers of cards, which increases prices. Intel will produce their own, AMD will move to produce their own cards, Nvidia will be stuck on TSMC, and likely streets behind AMD/intel, or literally give AMD money for every Nvidia card sold by having them built essentially by AMD. Not exactly the best way to stay competitive in a shrinking market they can't compete in.

Larabee doesn't have to be fantastic to start off with. Nvidia offering nothing but midrange expensive GPU's to say Dell to sell in their rigs. While both Dell and AMD are selling literally millions of cpu's, mobo's gpu's in the cpu's and discrete gpu's from low to high end. Dell in the future will have no reason to buy Nvidia cards for a select few midrange rigs, especially as Nvidia continue to be a generation behind and more expensive for less performance in the mid and low end where most of their profits used to come from.

Dell and people like that are the majority of gpu sales, end user sales are next to nothing in comparison. Dell and those guys like platforms for big discounts that hit nice price points, they like keywords and fancy features they can sell their computers with. They like new and faster things to persuade Joe Average to shell out £400-800 every few years on a new computer, Nvidia won't offer anything useful to these guys anymore.

The GPGPU market Nvidia has put way to much effort into, while AMD has fallen onto an equally powerful gpgpu without even trying, is simply to small to keep paying for its own R&D, Nvidia probably spent a few hundred million on R&D this year alone, the entire revenue(not profit, but revenue) for the GPGPU sector was $80million or so last year.

They can't make CPU's, while they might own part of Via, they are distinctly different companies, and as such the licensing agreement will without question not cover Nvidia making CPU's under the Via name. Even if they did they'd be hampered by making it somewhere like TSMC, way behind the competition in manufacturing and cost, ignoring their lack of experience and market power to produce and use the CPU. They can't afford to force a new cpu/mobo/gpu/gpgpu platform out, they don't have the capital, the products, the competitive prices or the economy for Dell or anyone else to sell such computers, again ignoring the billions in developement it would cost that they don't have.

Their only win that I can think of recently is getting the DS mobile chips, but its 10-20million miniscule profit chips, and while they got that they lost the much larger profit PS4 contract, so overall its a significant loss in income.

The real thing Nvidia, realistically, have to do is switch to a new small efficient ATi style architecture, and SOON, very very soon. Even if they have to eat crow and get laughed at for copying AMD, its the only viable way to compete on cost. Even if they do that, they still won't have a platform to sell to the OEM's.

Again this is why the AMD move to buy ATi was so freaking monumentally brilliant. It made AMD the platform that made the trillionaire arab oil guys see the potential and buy them.

Long term future Nvidia can't compete on cost, manufacturing, maybe performance due to the two afforementioned problems, which leaves them, I honestly don't know where. No where, I realistically see them dead within 2 years in the GPU market. I can't see GPGPu market taking off big enough to cover their R&D costs in that time either. Will they become a niche company that designs some fairly simple IC's that it sells the design for other people to make, maybe.

Keep in mind bumpgate, the half a billion they've had to pay Sony/Dell/HP and plenty of others to cover their failed products, the fact they are barely shipping anything over $100 at the moment to any of those guys, the superb treatment all those guys are getting from ATi, the massive switch of Apple to ATi cards in the past couple months, same for Dell. When Nvidia get back on track, considering the problems, the treatment, the refusal to accept responsibility, the bad PR those companies got for their products failing, the better prices from both Intel and AMD, can you honestly see many of them willingly choosing to go with more expensive late products from a company thats screwed them?


IF ATi had not switch architecture so dramatically, if AMD hadn't bought them they'd both be in a similar situation to Nvidia, which would also make Nvidia far stronger than they are now with ATi having the same delay problems, but they aren't.
 
drunkenmaster, i can see them making a nice product for the handheld market.

They don't have a x86 license, fair enough, but x86 is not the only processor technology out there.

ARM is a big force as well, if they can find some way of creating a MCM design based around ARM and their GPU IP, couple that with Android and you have yourself a very decent netbook, they could seriously hit intel in the ********.
 
TI might be bringing their own fab back online due to issues with TSMC and other foundary partners... if nVidia got in bed with them it would make it even more viable.

Its a big if tho - but if it ever did happen it would jump nVidia ahead of the 40nm process - its about the only chance I can see they have if TSMC continues to screw up.
 
Last edited:
drunkenmaster, i can see them making a nice product for the handheld market.

They don't have a x86 license, fair enough, but x86 is not the only processor technology out there.

ARM is a big force as well, if they can find some way of creating a MCM design based around ARM and their GPU IP, couple that with Android and you have yourself a very decent netbook, they could seriously hit intel in the ********.

ARM and nVidia would make for good bed fellows... and I honestly don't see it outside of possibility of seeing a RISC port of windows if there was enough interest as a lot of the work has already been done.
 
I wouldn't say jump into bed as per se. ARM is a technology for license, nVidia would have to build around that IP, using their own area of expertise.
 
Tho I really wouldn't want to see ARM dragged down by nVidia - I have a bit of a soft spot for ARM from back in teh day.
 
I've said before, ARM's weakness is their GPU. nV's strength is their GPU.

The issue is the two companies work have fundamentally different cultures, however they have 'worked' together on netbooks against Qualcom etc.
 
Business is Business at the end of the day thou, we need a competitive environment, Intel's LRB will not provide that to us with Gaming. We need nVidia to stick around.
 
Back
Top Bottom