Just wanted to get back to Li Mu Bai's B3D post* about Nintendo using some proprietary fixed function units in addition to the standard shaders. Mind you, I might be getting the terminology wrong, as I'm not a GPU guru or anything (I only deal with software rendering). Anyway, I do at least like to pretend I can understand the logic of going this route.
When Rogue Squadron came out, lots of us were mystified as to how Factor 5 could get so much graphical fidelity out of the purple lunchbox. The title had effects like self-shadowing and bump mapping applied to just about everything, yet the Gamecube managed not to ignite. The explanation we got back then was that Factor 5 was performing some sort of assembly code magic, creating something out of nothing. Yeah, well... not really. The Gamecube had those effects built-in, it's just that Factor 5 were competent enough to read the effing manual, as it were.
If Gamecube had Xbox-style programmable shader units in place of these built-in effects, it would have been able to achieve the same results. While that's true, they would have run much much slower. The big trade-off for the flexibility of a do-anything programmable shader is efficiency. Seeing as how comparatively slow the Flipper ran, this was a big deal. The important thing was finding the most useful, common effects to build into the silicon in lieu of the more flexible but slower alternative.
Li Mu Bai's post talks about lighting as being the most likely focus of these units in the Wii U. If that's true, then it would be kind of a major thing to overlook when porting your game to the Wii U. As lighting is a huge part of what a GPU is calculating, you would be making it jump through hoops it doesn't actually need to, slowing down everything else in the process.
That brings us to the past few weeks, and the 'leaks' from disgruntled anonymous developers about the Wii U. They're complaining about how it doesn't have enough shaders, so its graphics aren't powerful [sic]. You can see why some of us are more than a little skeptical about comments like that. If it doesn't have enough shaders, then it has something else in their place. Try using it.
* - the link he posted after the four quoted paragraphs should point here instead.
When Rogue Squadron came out, lots of us were mystified as to how Factor 5 could get so much graphical fidelity out of the purple lunchbox. The title had effects like self-shadowing and bump mapping applied to just about everything, yet the Gamecube managed not to ignite. The explanation we got back then was that Factor 5 was performing some sort of assembly code magic, creating something out of nothing. Yeah, well... not really. The Gamecube had those effects built-in, it's just that Factor 5 were competent enough to read the effing manual, as it were.
If Gamecube had Xbox-style programmable shader units in place of these built-in effects, it would have been able to achieve the same results. While that's true, they would have run much much slower. The big trade-off for the flexibility of a do-anything programmable shader is efficiency. Seeing as how comparatively slow the Flipper ran, this was a big deal. The important thing was finding the most useful, common effects to build into the silicon in lieu of the more flexible but slower alternative.
Li Mu Bai's post talks about lighting as being the most likely focus of these units in the Wii U. If that's true, then it would be kind of a major thing to overlook when porting your game to the Wii U. As lighting is a huge part of what a GPU is calculating, you would be making it jump through hoops it doesn't actually need to, slowing down everything else in the process.
That brings us to the past few weeks, and the 'leaks' from disgruntled anonymous developers about the Wii U. They're complaining about how it doesn't have enough shaders, so its graphics aren't powerful [sic]. You can see why some of us are more than a little skeptical about comments like that. If it doesn't have enough shaders, then it has something else in their place. Try using it.
* - the link he posted after the four quoted paragraphs should point here instead.
