Two things.
One, I noticed that sweet post, Gee. It's a really neat guide for what to expect of RV630, and it lines up so well with G84 that we almost don't need actual benchmarks.
But it's a really nice way to bound RV630 speculation.
I'm not knowledgeable enough to further extrapolate how RV630's lower intended resolution will skew the results (as alluded to in the new Eric Demers interview, the higher the res, the less the texturing units are needed, which suits RV630's lower res and proportionally greater texturing power), but it's an interesting extra data point. Then again, perhaps not, if one is supposed to think of res not in absolute terms, but relative to the GPU (where 8x6 would be very roughly the same for RV630 as 16x12 would be for R600). So, like leone said, res may also affect our guesstimates.
Two, upon rereading my previous post, I hope it's clear (especially given this post) that "hamster wheel" applies to me as much as anyone else, and so is intended as a convivial deprecation/metaphor/hamstropomorphication, not a patronizing comment.
Julidz, it's really down to the artists. One would think that, given the same artist, the game that's more shader-intensive (be it more short shaders or some long shaders) will look more polished and so better because the artist had more to work with, but it may not be that straightforward. Real time 3D is a compromise, and I wonder how apparent the differences would be b/w the compromises we're discussing. And, as Morgoth said, there may be a limit to how much you can substitute shadering for texturing.
But shader intensive doesn't have to mean longer shaders, just more shader calculation relative to texturing. For instance, I think it's fair to say that Humus' Doom 3 patch (substituted shader calculations for texture lookups) made D3 more shader-intensive. I'm guessing it's still not pushing as many shader calcs as FEAR, though.
One, I noticed that sweet post, Gee. It's a really neat guide for what to expect of RV630, and it lines up so well with G84 that we almost don't need actual benchmarks.
But it's a really nice way to bound RV630 speculation. I'm not knowledgeable enough to further extrapolate how RV630's lower intended resolution will skew the results (as alluded to in the new Eric Demers interview, the higher the res, the less the texturing units are needed, which suits RV630's lower res and proportionally greater texturing power), but it's an interesting extra data point. Then again, perhaps not, if one is supposed to think of res not in absolute terms, but relative to the GPU (where 8x6 would be very roughly the same for RV630 as 16x12 would be for R600). So, like leone said, res may also affect our guesstimates.
Two, upon rereading my previous post, I hope it's clear (especially given this post) that "hamster wheel" applies to me as much as anyone else, and so is intended as a convivial deprecation/metaphor/hamstropomorphication, not a patronizing comment.
Julidz, it's really down to the artists. One would think that, given the same artist, the game that's more shader-intensive (be it more short shaders or some long shaders) will look more polished and so better because the artist had more to work with, but it may not be that straightforward. Real time 3D is a compromise, and I wonder how apparent the differences would be b/w the compromises we're discussing. And, as Morgoth said, there may be a limit to how much you can substitute shadering for texturing.
But shader intensive doesn't have to mean longer shaders, just more shader calculation relative to texturing. For instance, I think it's fair to say that Humus' Doom 3 patch (substituted shader calculations for texture lookups) made D3 more shader-intensive. I'm guessing it's still not pushing as many shader calcs as FEAR, though.
