I don't agree. Just because they are offering some extras doesn't mean a single GPU from them has to be able to do everything. When you buy a GPU, you buy a GPU and shouldn't expect to have anything extra. They DO offer extras and that's nothing but good. From there, it's up to you to decide if the performance drop derived from S3D or PhysX is worth for you or if it's worth to buy a second card. With Ati is the same, they offer Eyefinity and that's cool, you still have to buy 3 monitors, they won't give them to you and you still need more than one GPU if you trully want to play the latest DX11 games on 3 monitors at decent resolution and fps. (I only mention Ati, because that's the only other GPU company to compare with)
What I mean is that it's all your choice. You want the same experience that you always had or you want more? If you want more it's clear that you will have to pay more or be happy with what a single GPU can do. i.e. I could live with 40-50 fps with S3D and PhysX enabled and I'm sure that a single Fermi card will be able to deliver that in 99% of the games. I've been playing Stereo3D since Nvidia introduced their drivers* and I can tell you that you don't need more fps than you would need without, you do need more Hz in the monitor but that's all.
*way back in 2002? 2004? I actually don't remember when it was, but a long time ago for sure. I stopped playing Stereo3D when games started using massive post-processing effects that broke the 3D and introduced a lot of ghosting. Also post effects where always rendered in the front instead at the depth they should be, etc. Stereo3D has always been the reason for my preference for Nvidia cards in the past and one of the reasons I still use a CRT. I don't like the 3D Vision too much, on the other hand, because it has to be with their glasses and LCD monitors AFAIK, although I'm not sure.
I agree with ya 110%. I'm extra sensitive to FPS, and for me, good gaming is 120FPS. Less, and I tend to get headaches. And truly, I don't think expecting that in a single card is realistic either...that's kinda my point...Fermi isn't just a single card, it's a complete platform, amd with SLi added in as a feature, it's GOTTA take 2 cards or more. That's perfectly fine to me. We know what Fermi is already...the details are out. Performance isn't a question...merely drivers are left, as single-card performance isn't the full story, to me. And there's no doubting they can pull it off.
And while Eyefinity takes 3 monitors, the end cost is the same as nV's S3D(thanks for that abbreviation!), so that means little to me. Heck, I've already got the monitors. But like I said...it's not ready, so to me, even as an ATI fanboy, ATI has failed to successfully launch Eyefinity and 5800-series. Old news, and what ATi does really has no effect on Fermi, or vice-versa.
Take a look at my sig...first scores are the best of ATi's DX9 series, second is DX10. Eother company meets, or exceeds, the expectations put out by those platforms already. But here we are entering a new platform, DX11...they've got time to kill to release, renew, and knowing nV and re-labelling, even a bit of recycling.
I guess part of the pressure on Nv is due to them stopping production of GT200 cards, but I think they've properly managed this, so that stores aren't left holding old stock.
That's the usual thing, but I believe this time they are going to try something different. My opinion is based in something they said in an Anandtech article tbh. They asked Nvidia about pricing and they said "You will be pleasantly surprised by the price when it launches". Maybe I'm giving too much credit to that quote, but I think that releasing a card that is tempting for everyone, not just gamers is the only way to go for Nvidia considering what they want to do and what their future strategy seems to be. GPGPU is not going to be anything if they don't manage to get some "mainstream" * people aboard. And it's their interest to make GPGPU successful, because they are much much faster there.
* With mainstream there I mean the kind of people that has mainstream graphics but powerful ($600-1000) CPUs, because they do have high computing necesities, although not visual ones. If buying a $400 card along with a $400 CPU is going to be better for them than buying a $800 CPU they will buy one, but if it costs $600 they will not. Looking at the future it's better for Nvidia to sell the cards at lower margins and capitalize in HPC sales.
See, I get something different form that...I hear nV say, when mentioning cost..."Hey, you don't need a super-powerful cpu...we got you covered. Buy that cheap cpu, and spend that money on graphics. We can crunch the same numbers they can, and we can do it faster." I see them saying you only need a $200 cpu(Hello Phenom2 C3), and here, take our $600 gpu for your power-computing needs.
That's why AMD is a consideration....thier cost has always fit well with AMD cpus. But if you got an AMD cpu, why not buy an AMD videocard too? What do they have that makes them better?