The problem I see is that while it may be possible to break the light barrier without breaking causality and using up infinite amounts of energy (and getting infinite mass), we ourselves may be keeping us from discovering it. He's right; this needs non-mainstream thinking. Creativity is severely dampened by this-is-impossible cries. Some might see a challenge in it to disprove this, but even then, the fact that it is considered impossible is cemented in the mind, thereby having an impact on creativity. Also, the fact that sometimes, the scientific community behaves like the church condemning heretics (just read the part with the difficulties getting a hearing about this exotic propulsion concepts), and that consequently, there are MANY crackpots in these "forbidden zones" which create an enormous noise, do not make things really easier. This might be too complicated for an innovation made by some weird genius in his basement, but the powers that could handle it might be too narrow-minded.
I doubt Vista's WDM architecture is mature and stable enough to be released. Thus, it would not make ANY sense to send nVidia & co. the specs right now, since they are still subject to change. Furthermore, the IHVs WILL demand knowledge about the internal structure in order to optimize their drivers. Also, don't forget that Autodesk, Alias, nVidia itself, XSI etc. demand OpenGL support; don't underestimate this pressure.
Time will tell. Once Vista is released, we'll see.
AFAIK it is a driver issue, actually. If MS wanted to kill off OpenGL, they would simply dump it. However, since writing an OpenGL ICD is far from being an easy task, the d3d-gl-bridge serves as a default 1.4 GL implementation. This is useful for cheap graphics chips which have drivers lacking good GL support (my old laptops Lynx3DM chip for example). IHVs still have the option to provide their own ICD simply by replacing the OpenGL32.dll. Since D3D and OGL are just interfaces to the driver, it is possible for them to mix the two, for example by providing an overall graphics server (which acts like a GL server too) and D3D/OGL apps being clients.
I still don't understand why the compiler was moved into the GL server. I mean, it makes writing ICDs even harder than it is already, and doesn't have any obvious benefits. Why not simply standardize some bytecodes and have people write free GLSL->bytecode compilers?
You are right, it is annoying. I'd have done it differently -.oga for Ogg Audio files (right now, Ogg Vorbis), and.ogv for Ogg Video. Mind you, Audio/Video are not tied to Vorbis/Theora, but these extensions would be much easier to understand. Also, decent DirectShow Theora decoders are still missing:(
"A full OpenGl desktop will be problematic when you want to run a 'windowed' version of Quake in for example, as the applicaiton will be expecting to have full control of the OpenGL/GPU and not expecting the first priority to be going to the Desktop Environment."
This is wrong. OpenGL is actually better suited for this than Direct3D, since OpenGL has a client/server architecture. OSX proves this - it is easily possible to play in windowed mode with no slowdowns. So, no 3D apps get broken. D3D needs to be redesigned for this, in short it needs a similar client/server-architecture. I wouldnt be surprised to see Direct3D 10 heading in this direction.
I would love to see KDE 4 with a tiled window manager like Ion and scaled windows like you described. (The only overlapping windows being dialogs.) This could be switchable, between "tiled" and "normal" mode.
Re:GNOME's audio backend GStreamer to use DRM
on
A Look at GNOME 2.14
·
· Score: 1
The thing is, usually Joe does not know how to install Xine, win32codecs, ffmpeg, libdvdread, libdvdcss... he expects it to work out of the box. If Linux doesnt, and Windows does, the point goes to Windows.
The average user should NOT use Xine, since you can be sued for using it to watch DRMed content like CSS-encrypted DVDs. So far GStreamer is the only system able to playback this stuff with no legal uncertainities.
Also, forget about people throwing away their DVDs. They'd rather throw the Linux distro away because "this junk does not play DVDs".
Re:GNOME's audio backend GStreamer to use DRM
on
A Look at GNOME 2.14
·
· Score: 2, Insightful
Xine is more stable, but GStreamer has the better design (it already surpassed Windows' DirectShow). Also, with Xine, VLC, and mplayer, it is impossible to play DVDs legally. Thus, GStreamer's closed-source modules are a good thing. No DVDs means no new Joe Averages for the Linux desktop, no increased market share and therefore no greater influence useful for fighting MS lock-ins by giving standards weight. Right now, Linux distros work out-of-the-box with standards like Ogg and OpenDocument, but MS can still push wmv/wma simply because the Linux bastion has no power to push their stuff.
Simple. Our old friends RIAA and MPAA gotta love this.
The media industry does NOT need PCs. On the other hand, PC and laptop vendors do need media as its one reason more to buy a computer. If the studios want, there will be no video playback for PC, they can rely on dedicated DVD players. This wouldnt hurt them, but it would hurt IBM/Lenovo, Dell, Acer, and MS since they deliver the most used OS for the computers. So the *AAs can force MS and the IHVs to play along. As a side effect, ANYTHING can be DRM'ed. This is both good and bad, but the bad side really overshadows the good one, since I DO NOT trust the content producers. However, there is the chance of all this breaking down due to faulty implementations and pissed customers (like "now this piece of shit tells me that I am not allowed to watch my damn video again just because I got my Windows reinstalled!") just like what happened with copy-protected CDs.
IBM, Dell etc. are not the evil ones here, neither is MS. Its the *AAs who don't care about destroying the universal computer.
Correct. Ironically, FEAR impressed me most with sequences like these, and NOT with its graphics (being a 3d-graphics-coding hobbyist apparently saturates). I found it quite impressive to be blown out of through the shattered window when a really big explosion incinerated the building I was walking in. Or the nuclear reactor meltdown, which causes a HUGE atomic explosion, everything gets really bright, and in the distant, the shockwave can be seen, nearing and smashing everything in its way..... really awesome, and does NOT need an expensive card. (Well, FEAR does, but such animations don't.)
Definitely not graphics
on
What is Next-Gen?
·
· Score: 3, Interesting
Graphics reached a "good enough" point. The next major breakthrough is likely to be real-time raytracing and/or real-time global illumination. Also, animation still sucks. Completely believable animation with real-time reactions to environmental changes and to player actions is still far off (it involves physics and AI too).
An example: A guard patrols an area. You are hidden behind a wall, waiting for the right moment to sneak past the guard to the room's other side. Then, you accidentally hit a bucket. The guard hears the sound, and runs to investigate it. No problem so far, this can be done with premodeled animation sequences (walking, standing, running...)
But then there is a rock on the ground. The guard hits it with his left foot. What happens? In real life, the guard would fall down. Now this is quite hard since the animation has to change in real-time. It involves physics (rock shape, amount of force, collision location...) and AI (since the animation has to change in a convincing manner, and this is achieved by letting an AI decide what to do next). This further leads to letting the guard stand up, checking himself if there are serious injuries etc. None of this is even remotely possible today.
So, you want next-gen with "next-gen" being purely technical? Look for advanced animation.
However, it IS true that today, games tend to be very similar, and there are much fewer games with new ideas. Also, there are less genres, for example there are no point-and-click adventures anymore, which is very sad since they were prime examples of gameplay over technology.
Speak for yourself. Actually, just operating at the TOP of the Maslow hierarchy is optimal. I am a passionate coder. An Amish way of life would mean for me to give up the thing I really like to do. Its like taking away a writer's pen or typewriter and forever denying him any chance of writing ever again. In short, "back to nature" means to give up LOTS of thinking and science. This may be your dream, but it is certainly not mine.
Also, you forget about the medical problems the Amish have.
Also stuff like "By 2025, we're going to be back in the Stone Age" and "when the Ghawar field waters out, you can kiss your lifestyle goodbye". It implies that one should forget about dreams, forget about technology, forget about thinking since everybody will be forced to be a farmer etc. Creative creators like painters, poets, writers, and engineers & software developers are basically told that they will never be able to do what they can do best ever again. It is an extremely fatalist point of view, which makes me wonder why people claiming this don't simply commit suicide. Instead of draining all hope and motivation people should encourage investitions into research & alternative energy infrastructures. While I agree that it is foolish to waste energy like it is done now (SUVs, throwing away stuff thats edible or still working, expensive stuff bought just because its "in"...) I won't sit around lamenting that everything is doomed. For the record, I do cut down consumption in my life. My computer is pretty much the only expensive investition, I don't have SUVs or big TVs.
Re:Huh? All those shots remind me of.....
on
KDE 4 Screenshots
·
· Score: 1
Hey hey, slow down. I meant the layout, the theme. Its much more gnome-like. This has NOTHING to do with usability.
Other than that, I agree with the gnome-dumbing-down-point.
Re:Cultural differences?
on
KDE 4 Screenshots
·
· Score: 0, Flamebait
Then again, after drinking US coffee and US beer, it becomes quite clear that the USA is not the place to be looking for examples of good taste either.:)
Huh? All those shots remind me of.....
on
KDE 4 Screenshots
·
· Score: 1
.... gnome!
Am I the only one seeing odd similarities between KDE4 and Gnome screenshots? Well, isn't all that important to me, as I use KDE not because of its optics, but because its tech is so much better than Gnome's. I mean, kioslaves vs. gnomevfs, konqueror vs. nautilus etc. (Also, I don't know if a gnome counterpart to KParts exists). The underlying systems are much better integrated and designed. The UI itself is a matter of taste.
/dev/dsp is always taken hostage unless the sound driver supports multiple sound streams, either by internal software mixing or by mixing in hardware. This is the very reason why sound daemons like arts or esd exist.
Of course, this changes once dmix is stable enough. And I hope that its latency and accuracy is not as bad as esd's.
Actually, Qt signals can not be achieved by template-based signals, since Qt signals allow limited introspection, which is impossible with template-based ones. This comes in handy when serializing signal-slot-connections, for example.
Scrollbars have been around for a long time. Taskbar: the same. These are basics of GUIs, and were present in the late 80s already (the scrollbar is even older). Simple but groundbreaking advances are rare nowadays simply because there has been put a lot of work into UIs already. The easy innovations are already here.
It is telling that in one of the videos (the zoom one?) they talk about the animations making the desktop feel more "physical" (as if it's a good thing). D'oh! Why should we want to be artificially limited to something similar to common physics? Fine if it's for a game, but for a UI?
I wouldn't say "physical". I'd say "solid". Some desktops feel like patchworks: they work but have many quirks here and there. More important, the latency varies greatly (when moving large windows for example). A solid feeling requires these two to be solved. Flickering is also bad. All of this is not *necessary*, but it makes working with the desktop less annoying.
Indeed, one wonderful example comes from the GUI of the game Aquanox Revelation. There, I can move windows with animated rendered content, and there is NO lagging increase at all, no matter how many windows are open. With a geforce4 ti4400 and 1280x1024x32bit 4xAA.
You Sir are completely wrong. Compiz is a combined window manager/compositing manager. KDE uses its own compositing manager (kcompmgr) and its likely that KDE 4 will have a combined WM/CM too.
Wrong. RIGHT NOW the GPUs are wasted. They sit there, with a lot functionality, doing 2D tasks that are a dead end. No one expects 200 million parallax-mapped triangles in an UI, which leads me to the next line:
Also, not all desktops have GPUs fast enough to support all that. I suppose this would make the hardware manufacturers pretty happy eh? All the integrated graphics chips would need to be higher spec'ed.
Which is wrong too. The effects demonstrated in the videos require _very little_ GPU power. A typical shooter demands 100 times more computations. Today, ALL integrated chipsets with graphics cores support basic 3D. Intels Extreme Graphics 9 even supports Direct3D 9, so what's the deal?
For very little UI/usability gain, this sort of stuff would make machines that were perfectly decent, slower and less responsive.
First, these machines must be older than ~3 years. Second, you can always choose another X server. Right now the real problem is not the hardware - its the missing 3D functionality in the drivers.
I am not a luddite, far from it. I'd rather see _real_ improvements.
Like what? There have been many attempts at doing something else than a desktop. The thing is, nothing beats it. Its pretty sure that as long as we have 2D screens with a fixed frame, the desktop is here to stay. This may change with flexible OLED sheets like the ones in "Red Planet", with some sort of 3D displays, or with direct neural interfaces, all of them being still quite far off. Also, don't forget that today, often the hard drive is the bottleneck. If FRAMs/MRAMs/whatever-persistent-RAM gets ready, it will most likely cause a huge advance, since booting no longer loads stuff in mem (because its unnecessary by then), programs can access files directly without loading into mem speeding up things more, sleep mode becomes a lot easier (simply dont erase the temp files in the ram)..
In fact, Xgl is one GOOD progress. Its absolutely foolish to force an additional 2D system when the 3D system can do everything the 2D one is capable of, just MUCH MUCH faster and with a lot more flexibility (blending for example). Absence of stalls while drawing reduces (if not eliminates) subjective "lagging", which is very important.
Somewhere else you mentioned the Amiga UI. Point is, window redrawing is NOT fluent there. Try moving a window, it takes its time to redraw. Also, you forget about the simple fact that today's resolutions contain a hell of a lot more data. 1280x1024x32bit = 5 MByte for the framebuffer. Compare this to the typical Amiga Workbench resolution 640x256x4colors (2bit): 40960 byte. Turn on 16 colors, and performance and latency got a lot worse.
BTW I wonder if the 2D performance of video cards has actually improved over the years, I recall getting a 3D vid card a few years ago and noticing the 2D performance was poorer than my old 2D video card - lower fps in software rendering mode. Don't see many 2D benchmarks being done nowadays...
Today, GPUs do not have a dedicated 2D core. The space it takes in the die simply doesnt pay off. Instead, the driver acts as the emulation layer. (At least newer nVidia and ATI chips do this.)
Well, its a new tool for UI design. As with all tools, one has to learn how to use it properly first. For example, OSX shows that alphablending can be useful.
The main advantage is the sheer responsiveness. Everything feels more fluent, and lags are no longer noticeable.
Slashdot Mirror
The problem I see is that while it may be possible to break the light barrier without breaking causality and using up infinite amounts of energy (and getting infinite mass), we ourselves may be keeping us from discovering it. He's right; this needs non-mainstream thinking. Creativity is severely dampened by this-is-impossible cries. Some might see a challenge in it to disprove this, but even then, the fact that it is considered impossible is cemented in the mind, thereby having an impact on creativity. Also, the fact that sometimes, the scientific community behaves like the church condemning heretics (just read the part with the difficulties getting a hearing about this exotic propulsion concepts), and that consequently, there are MANY crackpots in these "forbidden zones" which create an enormous noise, do not make things really easier. This might be too complicated for an innovation made by some weird genius in his basement, but the powers that could handle it might be too narrow-minded.
I doubt Vista's WDM architecture is mature and stable enough to be released. Thus, it would not make ANY sense to send nVidia & co. the specs right now, since they are still subject to change. Furthermore, the IHVs WILL demand knowledge about the internal structure in order to optimize their drivers. Also, don't forget that Autodesk, Alias, nVidia itself, XSI etc. demand OpenGL support; don't underestimate this pressure.
Time will tell. Once Vista is released, we'll see.
AFAIK it is a driver issue, actually. If MS wanted to kill off OpenGL, they would simply dump it. However, since writing an OpenGL ICD is far from being an easy task, the d3d-gl-bridge serves as a default 1.4 GL implementation. This is useful for cheap graphics chips which have drivers lacking good GL support (my old laptops Lynx3DM chip for example). IHVs still have the option to provide their own ICD simply by replacing the OpenGL32.dll. Since D3D and OGL are just interfaces to the driver, it is possible for them to mix the two, for example by providing an overall graphics server (which acts like a GL server too) and D3D/OGL apps being clients.
I still don't understand why the compiler was moved into the GL server. I mean, it makes writing ICDs even harder than it is already, and doesn't have any obvious benefits. Why not simply standardize some bytecodes and have people write free GLSL->bytecode compilers?
You are right, it is annoying. I'd have done it differently - .oga for Ogg Audio files (right now, Ogg Vorbis), and .ogv for Ogg Video. Mind you, Audio/Video are not tied to Vorbis/Theora, but these extensions would be much easier to understand. Also, decent DirectShow Theora decoders are still missing :(
"A full OpenGl desktop will be problematic when you want to run a 'windowed' version of Quake in for example, as the applicaiton will be expecting to have full control of the OpenGL/GPU and not expecting the first priority to be going to the Desktop Environment."
This is wrong. OpenGL is actually better suited for this than Direct3D, since OpenGL has a client/server architecture. OSX proves this - it is easily possible to play in windowed mode with no slowdowns. So, no 3D apps get broken. D3D needs to be redesigned for this, in short it needs a similar client/server-architecture. I wouldnt be surprised to see Direct3D 10 heading in this direction.
I would love to see KDE 4 with a tiled window manager like Ion and scaled windows like you described. (The only overlapping windows being dialogs.) This could be switchable, between "tiled" and "normal" mode.
The thing is, usually Joe does not know how to install Xine, win32codecs, ffmpeg, libdvdread, libdvdcss... he expects it to work out of the box. If Linux doesnt, and Windows does, the point goes to Windows.
The average user should NOT use Xine, since you can be sued for using it to watch DRMed content like CSS-encrypted DVDs. So far GStreamer is the only system able to playback this stuff with no legal uncertainities.
Also, forget about people throwing away their DVDs. They'd rather throw the Linux distro away because "this junk does not play DVDs".
Xine is more stable, but GStreamer has the better design (it already surpassed Windows' DirectShow). Also, with Xine, VLC, and mplayer, it is impossible to play DVDs legally. Thus, GStreamer's closed-source modules are a good thing. No DVDs means no new Joe Averages for the Linux desktop, no increased market share and therefore no greater influence useful for fighting MS lock-ins by giving standards weight. Right now, Linux distros work out-of-the-box with standards like Ogg and OpenDocument, but MS can still push wmv/wma simply because the Linux bastion has no power to push their stuff.
Simple. Our old friends RIAA and MPAA gotta love this.
The media industry does NOT need PCs. On the other hand, PC and laptop vendors do need media as its one reason more to buy a computer. If the studios want, there will be no video playback for PC, they can rely on dedicated DVD players. This wouldnt hurt them, but it would hurt IBM/Lenovo, Dell, Acer, and MS since they deliver the most used OS for the computers. So the *AAs can force MS and the IHVs to play along. As a side effect, ANYTHING can be DRM'ed. This is both good and bad, but the bad side really overshadows the good one, since I DO NOT trust the content producers. However, there is the chance of all this breaking down due to faulty implementations and pissed customers (like "now this piece of shit tells me that I am not allowed to watch my damn video again just because I got my Windows reinstalled!") just like what happened with copy-protected CDs.
IBM, Dell etc. are not the evil ones here, neither is MS. Its the *AAs who don't care about destroying the universal computer.
Correct. Ironically, FEAR impressed me most with sequences like these, and NOT with its graphics (being a 3d-graphics-coding hobbyist apparently saturates). I found it quite impressive to be blown out of through the shattered window when a really big explosion incinerated the building I was walking in. Or the nuclear reactor meltdown, which causes a HUGE atomic explosion, everything gets really bright, and in the distant, the shockwave can be seen, nearing and smashing everything in its way..... really awesome, and does NOT need an expensive card. (Well, FEAR does, but such animations don't.)
Graphics reached a "good enough" point. The next major breakthrough is likely to be real-time raytracing and/or real-time global illumination. Also, animation still sucks. Completely believable animation with real-time reactions to environmental changes and to player actions is still far off (it involves physics and AI too).
An example: A guard patrols an area. You are hidden behind a wall, waiting for the right moment to sneak past the guard to the room's other side. Then, you accidentally hit a bucket. The guard hears the sound, and runs to investigate it. No problem so far, this can be done with premodeled animation sequences (walking, standing, running...)
But then there is a rock on the ground. The guard hits it with his left foot. What happens? In real life, the guard would fall down. Now this is quite hard since the animation has to change in real-time. It involves physics (rock shape, amount of force, collision location...) and AI (since the animation has to change in a convincing manner, and this is achieved by letting an AI decide what to do next). This further leads to letting the guard stand up, checking himself if there are serious injuries etc. None of this is even remotely possible today.
So, you want next-gen with "next-gen" being purely technical? Look for advanced animation.
However, it IS true that today, games tend to be very similar, and there are much fewer games with new ideas. Also, there are less genres, for example there are no point-and-click adventures anymore, which is very sad since they were prime examples of gameplay over technology.
Speak for yourself. Actually, just operating at the TOP of the Maslow hierarchy is optimal. I am a passionate coder. An Amish way of life would mean for me to give up the thing I really like to do. Its like taking away a writer's pen or typewriter and forever denying him any chance of writing ever again. In short, "back to nature" means to give up LOTS of thinking and science. This may be your dream, but it is certainly not mine.
Also, you forget about the medical problems the Amish have.
Also stuff like "By 2025, we're going to be back in the Stone Age" and "when the Ghawar field waters out, you can kiss your lifestyle goodbye". It implies that one should forget about dreams, forget about technology, forget about thinking since everybody will be forced to be a farmer etc.
Creative creators like painters, poets, writers, and engineers & software developers are basically told that they will never be able to do what they can do best ever again. It is an extremely fatalist point of view, which makes me wonder why people claiming this don't simply commit suicide. Instead of draining all hope and motivation people should encourage investitions into research & alternative energy infrastructures. While I agree that it is foolish to waste energy like it is done now (SUVs, throwing away stuff thats edible or still working, expensive stuff bought just because its "in"...) I won't sit around lamenting that everything is doomed. For the record, I do cut down consumption in my life. My computer is pretty much the only expensive investition, I don't have SUVs or big TVs.
Hey hey, slow down. I meant the layout, the theme. Its much more gnome-like. This has NOTHING to do with usability.
Other than that, I agree with the gnome-dumbing-down-point.
Then again, after drinking US coffee and US beer, it becomes quite clear that the USA is not the place to be looking for examples of good taste either. :)
.... gnome!
Am I the only one seeing odd similarities between KDE4 and Gnome screenshots? Well, isn't all that important to me, as I use KDE not because of its optics, but because its tech is so much better than Gnome's. I mean, kioslaves vs. gnomevfs, konqueror vs. nautilus etc. (Also, I don't know if a gnome counterpart to KParts exists). The underlying systems are much better integrated and designed. The UI itself is a matter of taste.
/dev/dsp is always taken hostage unless the sound driver supports multiple sound streams, either by internal software mixing or by mixing in hardware. This is the very reason why sound daemons like arts or esd exist.
Of course, this changes once dmix is stable enough. And I hope that its latency and accuracy is not as bad as esd's.
Actually, Qt signals can not be achieved by template-based signals, since Qt signals allow limited introspection, which is impossible with template-based ones. This comes in handy when serializing signal-slot-connections, for example.
Scrollbars have been around for a long time. Taskbar: the same. These are basics of GUIs, and were present in the late 80s already (the scrollbar is even older). Simple but groundbreaking advances are rare nowadays simply because there has been put a lot of work into UIs already. The easy innovations are already here.
It is telling that in one of the videos (the zoom one?) they talk about the animations making the desktop feel more "physical" (as if it's a good thing). D'oh! Why should we want to be artificially limited to something similar to common physics? Fine if it's for a game, but for a UI?
I wouldn't say "physical". I'd say "solid". Some desktops feel like patchworks: they work but have many quirks here and there. More important, the latency varies greatly (when moving large windows for example). A solid feeling requires these two to be solved. Flickering is also bad. All of this is not *necessary*, but it makes working with the desktop less annoying.
Indeed, one wonderful example comes from the GUI of the game Aquanox Revelation. There, I can move windows with animated rendered content, and there is NO lagging increase at all, no matter how many windows are open. With a geforce4 ti4400 and 1280x1024x32bit 4xAA.
You Sir are completely wrong. Compiz is a combined window manager/compositing manager. KDE uses its own compositing manager (kcompmgr) and its likely that KDE 4 will have a combined WM/CM too.
OK it's a waste of GPU. Happy now?
Wrong. RIGHT NOW the GPUs are wasted. They sit there, with a lot functionality, doing 2D tasks that are a dead end. No one expects 200 million parallax-mapped triangles in an UI, which leads me to the next line:
Also, not all desktops have GPUs fast enough to support all that. I suppose this would make the hardware manufacturers pretty happy eh? All the integrated graphics chips would need to be higher spec'ed.
Which is wrong too. The effects demonstrated in the videos require _very little_ GPU power. A typical shooter demands 100 times more computations. Today, ALL integrated chipsets with graphics cores support basic 3D. Intels Extreme Graphics 9 even supports Direct3D 9, so what's the deal?
For very little UI/usability gain, this sort of stuff would make machines that were perfectly decent, slower and less responsive.
First, these machines must be older than ~3 years. Second, you can always choose another X server. Right now the real problem is not the hardware - its the missing 3D functionality in the drivers.
I am not a luddite, far from it. I'd rather see _real_ improvements.
Like what? There have been many attempts at doing something else than a desktop. The thing is, nothing beats it. Its pretty sure that as long as we have 2D screens with a fixed frame, the desktop is here to stay. This may change with flexible OLED sheets like the ones in "Red Planet", with some sort of 3D displays, or with direct neural interfaces, all of them being still quite far off. Also, don't forget that today, often the hard drive is the bottleneck. If FRAMs/MRAMs/whatever-persistent-RAM gets ready, it
will most likely cause a huge advance, since booting no longer loads stuff in mem (because its unnecessary by then), programs can access files directly without loading into mem speeding up things more, sleep mode becomes a lot easier (simply dont erase the temp files in the ram)..
In fact, Xgl is one GOOD progress. Its absolutely foolish to force an additional 2D system when the 3D system can do everything the 2D one is capable of, just MUCH MUCH faster and with a lot more flexibility (blending for example). Absence of stalls while drawing reduces (if not eliminates) subjective "lagging", which is very important.
Somewhere else you mentioned the Amiga UI. Point is, window redrawing is NOT fluent there. Try moving a window, it takes its time to redraw. Also, you forget about the simple fact that today's resolutions contain a hell of a lot more data. 1280x1024x32bit = 5 MByte for the framebuffer. Compare this to the typical Amiga Workbench resolution 640x256x4colors (2bit): 40960 byte. Turn on 16 colors, and performance and latency got a lot worse.
BTW I wonder if the 2D performance of video cards has actually improved over the years, I recall getting a 3D vid card a few years ago and noticing the 2D performance was poorer than my old 2D video card - lower fps in software rendering mode. Don't see many 2D benchmarks being done nowadays...
Today, GPUs do not have a dedicated 2D core. The space it takes in the die simply doesnt pay off. Instead, the driver acts as the emulation layer. (At least newer nVidia and ATI chips do this.)
Well, its a new tool for UI design. As with all tools, one has to learn how to use it properly first. For example, OSX shows that alphablending can be useful.
The main advantage is the sheer responsiveness. Everything feels more fluent, and lags are no longer noticeable.