I have to say, as big of a flame as that was (and I can't say I condone it) this guy (or girl) just said basically everything I felt while reading this article. Yale? Wow. Remind me never to transfer there.
The majority of the article is doubtlessly right, but his arrogant greater-than-thou attitude and unnecessary use of techno buzzwords reminds me more of William Gibson than any great mind of our past. Perhaps this more insight into the author's own intellectual insecurity than it does into the topic at hand.
I think a lot of the discussion is missing the point. Legality issues are clear. They can publish basically what they want. Unless it is heinously out of wack and offends the wrong people they can do what they want. Also, they are a private company and so would like to promote themselves. However, I'm sure they make the same "managerial decisions" with people other than linux buffs. The point is that it is demonstrating the drift of Linux towards mainstream commercialism, pulling away from the "moral" techie underground.
Without a recompile, old Apps should not gain any particular speed benefits because they never call on 64 bit instructions. Unless the core was designed in some spectacular way, taking out of order processing to an all new level by combining instructions and optimizing on the fly into 64 bit this would have no benefit on old apps.
The real question is not when they are going to name the next 386 on steroids. The question should be when are they going to switch to 64 bit? (forgive me if Williamette is) And what about ISA? It's had it's reign. Let it go happily. I think we've all felt the chokehold of backward compatibility for long enough.
No one is saying that everything Microsoft ever did was bad (well, not no one, but not many people). In fact, I think that, purely from the standpoint of GUI design, microsoft has been ahead with win 95/98 for quite a while. The problem is that the more techy stuff doesn't make much sense the way it's arranged under WIN and it doesn't work half the time either.
I don't think anyone is against guiness (especially if they still have their prompt), they are against the poor programming, and monopolistic nature of Microsoft.
Actually, Voodoo's 4x is equivalent to nVidia's 2x. nVidia intereprets it as 2x both axis, Voodoo things bigger numbers = better so 4x the area. Both just downsample for antialiasing.
Really that's what all the console companies do. Timing is essential in product release, especially in a such a highly competitive market with few leaders.
The thing is a PIII 733 with a GeForce 2 type card. The thing can render like a mother. Sure stuff has been faked in the past, and the present info may very well be distorted. But, just look at the hardware. It's impressive stuff.
Actually the x-box is very much designed with gaming in mind. Processor is running at at least twice the clock of the PS2 (hopefully making up for the smaller word size) with excellent 3d hardware. Unlike a normal computer the RAM is shared between 3d and the CPU (or so I have been told). This will greatly reduce memory flow bottlenecks in graphics systems and increase frame rate and system flexibility. If it had been released be a previously existing name brand I believe it would be quite a competitor. For that reason alone I see it falling behind. And, like everyone else has said so far, we haven't seen the games yet.
Also, HDTV excluded, I don't think TVs scan more than 640x480. However, the impressive thing in the demo was not fill rate, but transforms. I think the change from 320x200 to 1024x768 wouldn't even make that much difference in the framerate.
Comparing simply the CPU power, it will be almost twice as fast as the competition (in mhz only). I think PS2 is going to be almost 300mhz, and the dolphin is going to be 350 - 400mhz. However, both the PS2 and Dolpin are going to be running on 128+ bit machines (definitely 128 for PS2 I'm not sure about the Dolphin). That will contribute a lot to game performance.
While sega is ahead of the competition, despite their 2.5M consoles sold, there are a lot of people who have intentionally NOT bought a sega because they are waiting for a PS2 or Dolphin which will have far surperior hardware. As for your "more polys than pixels" argument, really that's not very true. Even if more polygons for detail is not necessary, more polygons for things like reflection, and better atmosphereics, shadowing will always be a good thing and add dramatically to the less blatant aspects of realism.
In the background of the ping-pong ball movie I heard something about "turns out 120hz". That can't be the framerate. Can anyone understand what he's saying there?
Considering almost all 3d software is now written for an API like OpenGL or DirectX(groan), whether or not they are aware of it, developers are writing for hardware T&L even if they can't count on it for the speed benefit.
I think the thing people have to remember is that, given a 10mhz 386 with enough disk space for a reasonable amount of virtual RAM, you could render a seen with twice as many pingpong balls. It might take a while, but it could happen. The ps1 for instance easily could have produced that output (though not at that resolution) The point is the framerate more than anything else. By posting the images and labeling them as screenshots, the hidden assumptions is that these are screenshots from something reasonably playable. I think the butterfly demo is cake to render. The pingpong balls look more impressive, but considering the kind of heuristics that are applicable to it, it is a lot simpler (and still we don't know the framerate). It's hard to be impressed until they post more information.
Really, even excepting heuristics, the butterfly images really aren't that impressive. Each one could quite possibly be made of only two polygons (four triangles) sharing six verticies mapped with a texture map which has an alpha channel. Make the texture high enough resolution and you can't tell that the edges are a little blurry. What's more impressive is the ping pong ball demo, really. Of course, there are some heuristics which could be applied to that to to eliminate some almost redundant calculation within each ping-pong ball (for example consider all the points on a ball to be the same distance eliminating a multiply or two and some trig). However, considering the number of balls it is quite impressive. I would be more interested to see the framerate on that demo. Does anyone know if it's possible to find a copy of the demo anywhere?
Slashdot Mirror
I have to say, as big of a flame as that was (and I can't say I condone it) this guy (or girl) just said basically everything I felt while reading this article. Yale? Wow. Remind me never to transfer there.
The majority of the article is doubtlessly right, but his arrogant greater-than-thou attitude and unnecessary use of techno buzzwords reminds me more of William Gibson than any great mind of our past. Perhaps this more insight into the author's own intellectual insecurity than it does into the topic at hand.
/.ed?
I think a lot of the discussion is missing the point. Legality issues are clear. They can publish basically what they want. Unless it is heinously out of wack and offends the wrong people they can do what they want. Also, they are a private company and so would like to promote themselves. However, I'm sure they make the same "managerial decisions" with people other than linux buffs. The point is that it is demonstrating the drift of Linux towards mainstream commercialism, pulling away from the "moral" techie underground.
It doesn't need more RAM or disk space.
Without a recompile, old Apps should not gain any particular speed benefits because they never call on 64 bit instructions. Unless the core was designed in some spectacular way, taking out of order processing to an all new level by combining instructions and optimizing on the fly into 64 bit this would have no benefit on old apps.
A recompile may make it nicely functional on a 64 bit system, but it wouldn't be particularly optimized which is the main purpose of the conversion.
Is that 64 bit aware just like previous Windows releases awareness of 32 bit?
Also, the Alpha was designed with massive multiprocessing in mind. I seriously doubt SGI will ever make a system with 512 Pentium 4s in it.
The real question is not when they are going to name the next 386 on steroids. The question should be when are they going to switch to 64 bit? (forgive me if Williamette is) And what about ISA? It's had it's reign. Let it go happily. I think we've all felt the chokehold of backward compatibility for long enough.
But geeks wouldn't buy a Pentium 4 if they changed the name either. I'd still go for the alpha (dribble dribble)
No one is saying that everything Microsoft ever did was bad (well, not no one, but not many people). In fact, I think that, purely from the standpoint of GUI design, microsoft has been ahead with win 95/98 for quite a while. The problem is that the more techy stuff doesn't make much sense the way it's arranged under WIN and it doesn't work half the time either.
I don't think anyone is against guiness (especially if they still have their prompt), they are against the poor programming, and monopolistic nature of Microsoft.
Actually, Voodoo's 4x is equivalent to nVidia's 2x. nVidia intereprets it as 2x both axis, Voodoo things bigger numbers = better so 4x the area. Both just downsample for antialiasing.
Really that's what all the console companies do. Timing is essential in product release, especially in a such a highly competitive market with few leaders.
The thing is a PIII 733 with a GeForce 2 type card. The thing can render like a mother. Sure stuff has been faked in the past, and the present info may very well be distorted. But, just look at the hardware. It's impressive stuff.
Actually the x-box is very much designed with gaming in mind. Processor is running at at least twice the clock of the PS2 (hopefully making up for the smaller word size) with excellent 3d hardware. Unlike a normal computer the RAM is shared between 3d and the CPU (or so I have been told). This will greatly reduce memory flow bottlenecks in graphics systems and increase frame rate and system flexibility. If it had been released be a previously existing name brand I believe it would be quite a competitor. For that reason alone I see it falling behind. And, like everyone else has said so far, we haven't seen the games yet.
Also, HDTV excluded, I don't think TVs scan more than 640x480.
However, the impressive thing in the demo was not fill rate, but transforms. I think the change from 320x200 to 1024x768 wouldn't even make that much difference in the framerate.
Comparing simply the CPU power, it will be almost twice as fast as the competition (in mhz only).
I think PS2 is going to be almost 300mhz, and the dolphin is going to be 350 - 400mhz. However, both the PS2 and Dolpin are going to be running on 128+ bit machines (definitely 128 for PS2 I'm not sure about the Dolphin). That will contribute a lot to game performance.
Will it do real FSAA or the fudged, render at twice the resolution and then sample down, kind that the voodoo does?
While sega is ahead of the competition, despite their 2.5M consoles sold, there are a lot of people who have intentionally NOT bought a sega because they are waiting for a PS2 or Dolphin which will have far surperior hardware. As for your "more polys than pixels" argument, really that's not very true. Even if more polygons for detail is not necessary, more polygons for things like reflection, and better atmosphereics, shadowing will always be a good thing and add dramatically to the less blatant aspects of realism.
In the background of the ping-pong ball movie I heard something about "turns out 120hz". That can't be the framerate. Can anyone understand what he's saying there?
Considering almost all 3d software is now written for an API like OpenGL or DirectX(groan), whether or not they are aware of it, developers are writing for hardware T&L even if they can't count on it for the speed benefit.
I think the thing people have to remember is that, given a 10mhz 386 with enough disk space for a reasonable amount of virtual RAM, you could render a seen with twice as many pingpong balls. It might take a while, but it could happen. The ps1 for instance easily could have produced that output (though not at that resolution) The point is the framerate more than anything else. By posting the images and labeling them as screenshots, the hidden assumptions is that these are screenshots from something reasonably playable. I think the butterfly demo is cake to render. The pingpong balls look more impressive, but considering the kind of heuristics that are applicable to it, it is a lot simpler (and still we don't know the framerate). It's hard to be impressed until they post more information.
Really, even excepting heuristics, the butterfly images really aren't that impressive. Each one could quite possibly be made of only two polygons (four triangles) sharing six verticies mapped with a texture map which has an alpha channel. Make the texture high enough resolution and you can't tell that the edges are a little blurry. What's more impressive is the ping pong ball demo, really. Of course, there are some heuristics which could be applied to that to to eliminate some almost redundant calculation within each ping-pong ball (for example consider all the points on a ball to be the same distance eliminating a multiply or two and some trig). However, considering the number of balls it is quite impressive. I would be more interested to see the framerate on that demo. Does anyone know if it's possible to find a copy of the demo anywhere?