Thoughts on realistic new features on Next Gen cards?

[mx125race]

Lower prices is what i want...Just look at the GTX, around $450...It's slowly going down, but c'mon, even though it one of the best out now, I think it coulda been better. I, myself, do not have one, but one of my friends does and it's not all that super impressive. Still a great card overall, just think it coulda been better. And 1,000 for a single processor :eek: Hopefully this winter PC prices drop more than expected, especially at retail stores. Shiznitizzle fo shizzle!!! Patriot 1 gig stick of DDR400 was around $250 at a local Best Buy. And OEM memory, or upgrades if you got a retail PC such as Dell, costs around that, too. If they dropped the prices, they'd get a lot more sales. There's a lot of people that enjoy computers but can't waste their money on them. I don't think RAM should cost almost as much as a 6800 GT IMO, do you?

[retsam]

i hope the nvidia gods are listening!!.... full trilinear and angle independent af!!!

[nutball]

Like others said above, what I *want* is better IQ -- better AF, AA, etc. No more jaggies and sparklies please NVIDIA, kthx.

Stuff like physics chips a) are not going to happen in the GPU (because it makes *no* sense) and b) are "meh" anyway IMO.

I do wonder whether the mythical PPP will make into the next generation -- gut feeling is that it won't, but it'll be in the generation after next.

[nutball]

I would have to say 48 unified pipes is the next step thats for sure NV bashed the idea but they are going along with it anyway as it's a must for Dx10

Unified pixel + vertex shader pipes are not required for DX10, only a unified shader API (ie. same functionality available in shader programs for both purposes). How the underlying hardware delivers that unification is a matter for the IHVs.

There may well be good reason for keeping vertex + pixel pipelines seperate, as NVIDIA have already said, because although they may be required to implement orthogonal functionality in the shader language, typical use cases may remain very different. If function X is commonly used in pixel shaders, but rarely used in vertex shaders, and function Y common in vertex shaders but not pixel shaders, it would make sense to spend transistors optimising function X in the pixel pipes and function Y in the vertex pipes, rather than the same total number of trannies providing sub-par implementations of functions X + Y to both pipelines, if you see what I'm getting at.

[MUYA]

Link(s)?

Please tell me what you are basing this comment off of. Thanks.
In an interview with 3D center and bit-tech NV reps questioned unified efficiency when compared with current day "defined" functioning pipelines etc. They said they would make the tranistion to unified shaders when it suited them IIRC. It's all about comprehension, some say "bash" some say "PR" ;)

[Raje]

In an interview with 3D center and bit-tech NV reps questioned unified efficiency when compared with current day "defined" functioning pipelines etc. They said they would make the tranistion to unified shaders when it suited them IIRC. It's all about comprehension, some say "bash" some say "PR" ;)

Yeah, I've read those interviews and quoted them in previous posts.

I wasn't questioning the part about NVIDIA's "bashing" or "downtalking", etc. of Unified Pipelines. In fact, I tend to agree with NVIDIA's outlook on the subject, which I feel can be sumed up as such: "Debating unified against separate shader architecture is not really the important question. The strategy is simply to make the vertex and pixel pipelines go fast. The tactic is how you build an architecture to execute that strategy. We're just trying to work out what is the most efficient way." While ATI's approach leaves me with a feeling of "Even if this isn't the most efficient way of doing it, it sure sounds great on paper and people are already spreading the word thanks to Xbox360 vs. PS3 and a missunderstanding of API vs. Hardware implementation." It's just like what Nutball says in his post a few earlier in this thread.

Doesn't mean NVIDIA will succeed in making the most efficient architecture, but it's nice to at least hear that coming from NVIDIA's chief scientist especially when ATI's rep says things like "I’d love to say that Nvidia are going to be stuck when it comes to Longhorn."

What I am wondering is why Dazz thinks that 48 unified pipes is "the next step", that 48 unified pipes is "a must for DX10", and, more than that, why Dazz says that NVIDIA "are going along with it anyway". I think it would be significant for NVIDIA to announce they now have a unified architecture on the hardware level.

He's said almost this exact same thing twice now (once in this thread and once in another). Both times I've asked for links or some sort of clarification. Not to bash him or make him look bad (there's enough of that **** on other boards), but because I am curious if NVIDIA has said somewhere that they will actually be going with 48 unified pipelines because DX10 requires it or if this is just what Dazz thinks.

[shabby]

Nvidia didnt say anything about 48 unified pipes, dazz is just spreading fud.

[MUYA]

Yeah, I've read those interviews and quoted them in previous posts.

I wasn't questioning the part about NVIDIA's "bashing" or "downtalking", etc. of Unified Pipelines. In fact, I tend to agree with NVIDIA's outlook on the subject, which I feel can be sumed up as such: "Debating unified against separate shader architecture is not really the important question. The strategy is simply to make the vertex and pixel pipelines go fast. The tactic is how you build an architecture to execute that strategy. We're just trying to work out what is the most efficient way." While ATI's approach leaves me with a feeling of "Even if this isn't the most efficient way of doing it, it sure sounds great on paper and people are already spreading the word thanks to Xbox360 vs. PS3 and a missunderstanding of API vs. Hardware implementation." It's just like what Nutball says in his post a few earlier in this thread.

Doesn't mean NVIDIA will succeed in making the most efficient architecture, but it's nice to at least hear that coming from NVIDIA's chief scientist especially when ATI's rep says things like "I’d love to say that Nvidia are going to be stuck when it comes to Longhorn."

What I am wondering is why Dazz thinks that 48 unified pipes is "the next step", that 48 unified pipes is "a must for DX10", and, more than that, why Dazz says that NVIDIA "are going along with it anyway". I think it would be significant for NVIDIA to announce they now have a unified architecture on the hardware level.

He's said almost this exact same thing twice now (once in this thread and once in another). Both times I've asked for links or some sort of clarification. Not to bash him or make him look bad (there's enough of that **** on other boards), but because I am curious if NVIDIA has said somewhere that they will actually be going with 48 unified pipelines because DX10 requires it or if this is just what Dazz thinks.
whatca talkin about Willis? Really?