<strong>The current geForce4 Ti and ATI Radeon 8500 vertex and pixel shaders are the programmability I'm refering to, and they are just the beginning. Those boards do more work that a set of G4s could because they are designed to do geometry and pixel operations, whereas the G4 is a generalized computing engine.</strong><hr></blockquote>
Like I said. that <a href="http://www.nvidia.com/view.asp?IO=feature_nfinitefx"; target="_blank">nfiniteFX</a> stuff is a silly toy compared to a full CPU. or even a programmable DSP. no game programmer(Or computing mathematics student writing some brilliant new algorithm for their thesis) will be able to try out something REALLY new on such a chip.
[quote]Originally posted by Programmer:
<strong>Polygon based geometry is not the vile thing you seem to think it is. Most of the polygons are approaching single pixel size most of the time now anyhow, and the ATI boards support automatic sub-tesselation of individual triangles. The new hardware is doing a lot of per-pixel lighting calculations, which goes a long way to hiding the polygonal faceting... combined with higher polygon counts and you can't see the faceting nearly as much (if at all in many cases).</strong><hr></blockquote>
All of the techiques you mentioned are is basically just pussy footing around the core of the problem: you can't simulate EVERYTHING with the same mathematics. you need a little(Actually a lot of) variety. even the very best polygonal rolling hills look like crud compared to similar scenes done with voxels. and polygonal water still looks awful and doesn't react properly to stimulus from other geometry as compared to the proper styles of wave mechanics and effects.
On the topic of the Radeon 8500's TruForm technology. it's a shame it has to downsample it's internal smooth geometry into polygons prior to rendering. there's a quote(I forget where) from an ATI engineer in which he says that they actually could have done that. but it would've required the game programmers to change their style quite a bit in order to take advantage of it. oh well.
[quote]Originally posted by Programmer:
<strong>All sorts of radical approaches are tried all the time and, frankly, the industry goes in the direction of the most practical and useful. Techniques like you talk about have been tried, but they can't compete with the power of the graphics hardware. The flexibility afforded by programmable graphics hardware will allow a great deal more differentiation than possible with older hardware, and the graphics hardware is advancing at a much greater rate than CPUs.</strong><hr></blockquote>
The industry goes in the easiest direction. most game programmers are loath to even take advantage of special features of hardware(Like AltiVec and 3dNOW!). even if they have pretty high market penetration. I'll bet that techniques like I said have been tried. but half-decent CPUs to run them on(Especially really good VPUs) have always been hidden it the top of the market. the new PPCs and their high quality yet cheap and easily available AltiVec units are the first time the sort of hardware which is suited for this stuff has gotten within reach of anyone other than scientists(With very large grants ). the CPUs are still fast enough to outmaneuver the GPUs. but this window of opportunity might close within a mere 4 or so years(And will only be opened up again whenever quantum CPUs hit the broad market decades from now) unless the flow of money and R&D is redirected to the CPUs.
[quote]Originally posted by Programmer:
<strong>The GameCube's graphics hardware is actually very capable in terms of pixel computations, and it is not possible to emulate its behaviour on a geF4 or Radeon8500 (in hardware, and in software is too slow). The simplest example: GC supports 8 texture reads per pixel, whereas the geF4 supports 4 and the R8500 supports 6.</strong><hr></blockquote>
Making a native Gamecube compatability enviroment for the Macintosh. while it would be quite a _technical_ challenge to pull off(Especially until Apple revises the motherboards on Macs;(). could still be done. although with a fair speed hit. the difficult part would be the likelihood of the sorts of people that do this thing being on the Mac(Although there's always types like Connectix to prove those kinds of statements wrong ).
[quote]Originally posted by Programmer:
<strong>I'm not sure which "XBox secrets" you refer to. Even if its workings were entirely known, building an emulator that runs on standard PCs under Windows would have some serious performance issues (ring transitions, AGP bus performance, driver interface overhead, etc) and the games wouldn't operate properly as a result. Its possible, but the PCs are hamstrung by their hardware and OS.</strong><hr></blockquote>
The X-Box actually runs a variant of Windows 2000. and uses a(Slightly higher model) GeForce series chip. I think that an nForce based PC could come pretty close to the X-Box. and could certainly run X-Box games(If about 15-20% slower) in a native compatability enviroment.
[quote]Originally posted by Programmer:
<strong>What makes you think that anybody (much less Apple) could out-design the Sony, MS/nVidia, and Nintendo design teams? And even if they did, who would build the software for it? The existing designs are leading edge technology -- what makes you think that there can be a magic bullet to defeat them all, much less one that takes a radical new path?</strong><hr></blockquote>
Because radical advances capture the public eye(And the developer's attention). when a consumer is standing in a store trying to decide wether to get identical console A, identical console B, older but still identical console C or the whiz-bang new Apple one that looks(And acts) totally different from any console they've ever used or seen. I think a fair number of them would choose it.
[quote]Originally posted by Programmer:
<strong>Consoles have to be built with essentially commodity parts (or parts built on the same production lines), otherwise production costs will price them way out of the market. Given that level of similarity there is no way that some newcomer to the market can sneak in and blow everybody away... unless of course you believe aliens are setting up factories at Apple HQ and will be cranking out these magical new devices?</strong><hr></blockquote>
I don't think the Emotion Engine, NV25 or "Flipper" could be called commodity parts by any stretch of the imagination. whereas the G4(And the G5. when it arrives) is definitely a commodity part. as it's used across many industries in a variety of applications.
[quote]Originally posted by Programmer:
<strong>No, the next major console is coming from Sony because they are about a year ahead of the rest in their product lifecycle.</strong><hr></blockquote>
Unless Apple has been secretly working on a top secret mega-super-duper console behind the scenes for about a year or two .
<strong>There are a nearly infinite number of things Apple could do. Are we talking about suggestions of what Apple should do? I'm not. I'm discussing what Apple is most likely to do. I've already said that I'd like to see the iMac line more competitive in the gaming market. Specifically, I'd like to at see the ATI 7500 or the Geforce 4mx in the iMac. However, my point is that based on Apple's history, this will not happen anytime soon. Specifically, not before the Pro line is bumped to something like the Geforce 5mx. The good thing is that nVidia seems to be on a pretty aggresive schedule with their chip designs. Likewise, more powerful graphics chips may come more often then they have in the past. Still, I'd bet money the iMac's graphics chip does not get bumped in July!</strong><hr></blockquote>
It might if a fair number of people ask for it(Remember the original iMac's 33.6 modem?).
I finally see what you're going on and on about and why it won't ever work.
A totally new way of doing games can't happen with three G4/5 or even more. But if it could, it wouldn't happen at a better price-performance metric than the current CPU+GPU+audio paradigm.
Why?
GPU's will basically trounce a CPU (or a small batch of them) when it comes down to actually painting the scene onto the screen. A dedicated audio processor can be cheap, require very little overhead, and basically provide enough power for 3-D realism AND interactivity.
Gamers aren't making anything, or rendering anything, designing etc... They're along for a predefined ride. What "TOTALLY DIFFERENT" gaming experience are you talking about?
Better AI? For what? Existenz? Come on now. A fat plumber only needs to be so smart, sure better physics models and better AI could make for better simulations, more realism, and more 'soul'. As if you are facing a truly thinking opponent in a truly variable and rich environment. But why? And at what cost? You have multiplayer precisely so that you can face 'human' opponents, and even when you don't the game designers have already thought about things so that they can produce a convincing experience within the limits of the hardware. 3 CPU's won't make a game appreciably more interesting than one CPU, and they furthermore won't provide any significant ability to do so beyond the abilities of a single processor machine.
The game machine will always be about providing a cheap way to make it good looking and fast. The game is predefined, it only needs to 'think' so much, the rest of the time it just has to provide a satisfying environment for YOU to think and react in what is basically a co-ordination and skills (timing and timed response) test.
The only "COMPLETELY" different gaming experience would be a totally immersive simulation against a cognizant (sentient, holodeck character gone awry, fantasy) AI, and not three nor three thousand PPC's are going to provide that.
For now at least the current model is the best performing and cheapest way to do things. In some Blade Runner Sci-Fi geek future, who knows? But not for now or the forseeable future.
It might if a fair number of people ask for it(Remember the original iMac's 33.6 modem?).
Eric,</strong><hr></blockquote>
Yup, I remember. I also remember the majority of the complaints I saw were based on official reviews of the machine. The biggest issue there was something like: what's the point of calling it an "i"Mac if the "i" in iMac is a crippling feature. Remember, all of the advertisements for the iMac were based on it's internet capabilities. Now, if we had a "g" (for gaming) Mac that came with a Geforce 2mx, then I'd agree with you. In short, Apple doesn't market the iMac as the ultimate gaming machine. Playing games on the iMac is a feature, but it isn't advertised as it's core competency to my knowledge. Instead, Apple is now shifting focus from the internet (because all machines now do the internet well), to the digital hub. Likewise, Apple advertises the "iApps", but does not focus on games.
Again, I will readily agree with you that Apple stops short of what they should do in regards to gaming. My point is that from a reality standpoint, I don't see Apple doing much about gaming performance anytime soon. Especially considering that it's recent jump to the Geforce 2mx was a huge leap in performance over it's predecessor (Rage 128 Ultra).
It should be stated that almost every recent console created loses money from hardware sales, therefore profitting from producing console based machines is primariliy from developer royalty fees.
As for X-Box, the "X-GPU" is basically a Geforce 3 operating at 233Mhz with 2 vertex shaders, however. The nForce unified memory architecture allows for a total of 6.4GB/sec bandwith, the 133Mhz FSB path to the CPU allows for 1.06GB/sec to used for the CPU (733 CuMine Pentium III with 128KB L2). Therefore, subtacting these two numbers, this allows ofr 5.34GB/sec to be allocated to the IGP itself. That's correct, 5.34GB/sec, which itself is not really that much when you take into measure that a Geforce 3 Ti200 itself has 6.4GB/sec of memory bandwith (which itself counters the whole nForce solution). So the Xbox IGP solution given that it were put into a PC would be around an extremely underclocked Geforce 4 (167.5 DDR for mem), this was about how fast mem speeds were at the introduction of the Geforce 2 GTS (April 2000).
However, most console games are rendered at 640x480x32 so it really doesn't take into effect unless AA is enabled, etc.
64MB 5ns (i believe, if i'm wrong, please correct me) DDR total RAM, for both executing game and used as texture storange, etc.
Then there's the onboard sound, I'm not too knowledgable on sound so I'll have to remain silent on this.
X-Box IS like a PC, however it is a console as well, it just doesn't fit our typical impression of one. For instance, consoles are known to have extremely high bandwith graphics systems and low buffers, the PS2 GS runs at 150Mhz but has 16 pixel pipelines which gives it a theoretical pixel fill rate of 2.4 gigapixels/sec, yet only 4MB of DRAM. However, the connection between the DRAM and GS is huge, I believ it's something like a 2500bit bus, therefore the reasoning behind this is that the need for a large buffer is minimal as there is massive amounts of bandwith.
The reason many developers had problems with PS2 development early on was that the PS2 development kit shipped without any C libraries.
Anyway, yeah that's about all I know, and I know I'm probably wrong on quite a few of these facts so like I said before, I'm open to flaming or corrections because I really do find this stuff interesting and I want to eventually major in Computer Engineering someday when I go to college in some 3 years or whenever.
I finally see what you're going on and on about and why it won't ever work.</strong><hr></blockquote>
You have a vague inkling. but you have yet to grasp the full magnitude of this concept.
[quote]Originally posted by Matsu:
<strong>A totally new way of doing games can't happen with three G4/5 or even more. But if it could, it wouldn't happen at a better price-performance metric than the current CPU+GPU+audio paradigm.</strong><hr></blockquote>
I think it could.
[quote]Originally posted by Matsu:
<strong>Why?
GPU's will basically trounce a CPU (or a small batch of them) when it comes down to actually painting the scene onto the screen. A dedicated audio processor can be cheap, require very little overhead, and basically provide enough power for 3-D realism AND interactivity.</strong><hr></blockquote>
Only while using their brute force mathematics. and the stuff I'm referring to runs perfectly fluid in realtime.
[quote]Originally posted by Matsu:
<strong>Gamers aren't making anything, or rendering anything, designing etc... They're along for a predefined ride.</strong><hr></blockquote>
Why not? haven't you ever wanted to leave the beaten path? whilst trodding though yet another identical FPS hallway past a window. didn't you ever look at the building across the street and wonder: "What's in there"? a well programmed game for such a powerful system could dynamically generate chunks of a near endless world, fill them with characters and objects, then adjust the plot and the character's reactions to fit. all based on a large set of parameters defined by the writer/programmer/mathematician(s). for the true realization of the long sought "Non-linear story". that instead of trying to adapt the actions of the player(s) to the pre-defined story. would dynamically adapt(Using something like the <a href="http://www.dramatica.com/theory/index.html"; target="_blank">Dramatica theory</a>. only much more advanced. and probably better defined) the story itself to the actions of the player(s).
[quote]Originally posted by Matsu:
<strong>What "TOTALLY DIFFERENT" gaming experience are you talking about?
Better AI? For what? Existenz? Come on now. A fat plumber only needs to be so smart, sure better physics models and better AI could make for better simulations, more realism, and more 'soul'. As if you are facing a truly thinking opponent in a truly variable and rich environment. But why? And at what cost? You have multiplayer precisely so that you can face 'human' opponents, and even when you don't the game designers have already thought about things so that they can produce a convincing experience within the limits of the hardware. 3 CPU's won't make a game appreciably more interesting than one CPU, and they furthermore won't provide any significant ability to do so beyond the abilities of a single processor machine.</strong><hr></blockquote>
Imagine smooth rolling terrain stretching clear to the horizon(Through fractal extrapolation and voxel curves). dotted with good looking trees(From high speed fractal branching and reverse interpolation) and covered in flowing grass(Through a cheap. but effective. illusionary geometry trick) and rustling bushes(Same way as the trees). right by a beach with sand that gets pushed around properly by your feet(Like the hills. only with some intricate solid physics). and water that splashes in proper waves around objects and pushes them(With geometric primative based impact detection and some simple wave mechanics). all at a smooth 30FPS or so.
Imagine knocking over a pile of cans and having them bounce off each other and along the floor in a proper manner(With true geometry based impact detection). or modelling the riccochets and hits from an indoor machine gun fight with 60 or so participants(Using reflection and refraction. like simplified raytracing).
Imagine a murder mystery game in which you can examine an entire house with impeccable attention. to the point of a human hair caught in the fibers of a rug(Made with millions of simple fractal threads when viewed close-up) or the way a piece of fabric folds in your(Virtual) hands(done with a fine mesh of simple joints, fluid mechanics and impact detection).
You could do it on my theoretical system. but not on any game console(Or other non-PPC consumer machine) that exists now.
[quote]Originally posted by Matsu:
<strong>The game machine will always be about providing a cheap way to make it good looking and fast. The game is predefined, it only needs to 'think' so much, the rest of the time it just has to provide a satisfying environment for YOU to think and react in what is basically a co-ordination and skills (timing and timed response) test.</strong><hr></blockquote>
I think there are more genres than "Twitch" games. "Turn off the brain" action games are fun. but some variety would be nice too. everyone on this forum that's sick of the glut of action games over the last three or so years would agree.
[quote]Originally posted by Matsu:
<strong>The only "COMPLETELY" different gaming experience would be a totally immersive simulation against a cognizant (sentient, holodeck character gone awry, fantasy) AI, and not three nor three thousand PPC's are going to provide that.</strong><hr></blockquote>
You can actually get some pretty good AI's on current systems. and the physical interfaces have evolved to:
VR helmets(Ones that actually project the image directly into your retina like at the start of the "Total Recall" TV show. not just an LCD suspended six inchs in front of your nose).
Full headphones(Ones that fire at different angles. so as to take advantage of the sonically reflective nature of the inside of your skull) or for the truly grisly. ones that are embedded in your skull. and transmit directly into your aural nerve.
Special "Feel" suits that give you the sensations of touch and heat.
Hydraulic restraints to keep you from passing through virtual objects like a ghost(Like it'll keep you from bending your elbow any farther when you try to rest your arm on a virtual table).
Motion simulators that will simulate the effects of running down a cliff or doing a barrel roll in a jet.
And the modern counterpart of "Smell-O-vision". only it works.
But then that stuff is just plain silly .
[quote]Originally posted by Matsu:
<strong>For now at least the current model is the best performing and cheapest way to do things. In some Blade Runner Sci-Fi geek future, who knows? But not for now or the forseeable future.</strong><hr></blockquote>
No. the current model is the way they know how. if top programmers knew more math. and top mathematicians knew more programming. GPUs wouldn't exist(At least in they're current form).
<strong>Wanna bet? To be fair, I did bag that idea after my prospective advisor said he would not have time for a student due to the timeline of his current project. Just wait till E3 on that. But I gave a brief to the department on just that topic and we had some VERY concrete ideas on what a good end product would be to write the thesis on. There is still plenty of room there to expand easily into dissertation territory, especially with the extra features of the GF4 over what I was going to use, GF3.</strong><hr></blockquote>
What idea? who's this advisor? what department? ideas about what? I can't understand what you're talking about. is it that note near the bottom about using vertex shaders for unusual effects? speak(Er? type) more clearly.
[quote]Originally posted by AirSluf:
<strong>The CPU is a silly toy graphically speaking compared to a GPU for the task a GPU is designed for.</strong><hr></blockquote>
Exactly! but you wouldn't be [b]performing[/i] those sorts of tasks in the first place. you would be using a collection of much more power thrifty techniques. thus alowing the small, speedy CPU to trounce the immense, hulking GPU.
[quote]Originally posted by AirSluf:
<strong>Do you have any idea how much more horsepower it takes to derive pixels from a continuous function compared to using poly's?</strong><hr></blockquote>
Quite a bit less. that's why polygons were always used sparingly in earlier games. while voxels were used much more profusely.
[quote]Originally posted by AirSluf:
<strong>Why are NURBS so slow, that is what you are talking about when you vant on smooth continuous mathematically derived surfaces.
Poly's make it possible to spend less time rendering and more time doing physics and contextual behavior. Those things do more for a simulations or games realism than perfect scenery.</strong><hr></blockquote>
Polygons make it possible to produce complex surfaces without much advanced math(This is attractive when you don't know hardly any). also. don't forget that many of the graphics sound and physics techniques I mentioned before tend to work together and integrate with each other well. sort of pushing off each other. making the overall game engine much more integrated, tight and efficient. instead of the disparate mess of most current ones. which is compensated for by the brute force of the hardware.
[quote]Originally posted by AirSluf:
<strong>Think Myst vs any reasonable FPS, which is more fun realtime?</strong><hr></blockquote>
Both! I haven't actually tried RealMyst yet(Would it _really_ kill them to make a Mac demo !?). but i played the original, the movies and screenshots all look really cool. and judging from the <a href="http://www.insidemacgames.com/news/story.php?ArticleID=670"; target="_blank">Missed Island</a> Marathon Infinity map. the concept seems like a lot of fun.
[quote]Originally posted by AirSluf:
<strong>basically visual reality ain't all some people crack it up to be. I can convince you you are in a jungle at night with a dark humid room and good spatialized sound, turn on the video though and you are back in a room.</strong><hr></blockquote>
Spatialized sound is a cheap hack. it's just running audio through a quick filter. a console like I outlined would be able to calculate physical sound-waves bouncing and reverberating through the space, off the walls and getting damped by unsusual shaped geometry(Like a thick crowd of enemies approaching or spikes on the walls). techniques like I outlined would allow for a rich, immersive effect that would enable the player to practically hear their entire surroundings(And would scare the #%&! out of them if a fight suddenly erupted behind them).
Another cool effect would be. though more related to the feel of the game than the overall enviroment. dynamically composed music. instead of just swapping around the same loops whenever the action changes(Like in Deus Ex). you would give the computer some sort of outlines, perimeters and themes(Maybe some examples too) in which to compose the music depending on various conditions. this would solve both the problem of game music not being relevent to the action. and the fact of that even the best game soundtracks never last for the thousands of hours of typical playtime without looping.
[quote]Originally posted by AirSluf:
<strong>Give me the cycles for that spatialized sound and beastie behaviors, you won't have time to complain about the scene quality because your mind will spend nearly all it's resources trying to solve the immediate problem of survival which won't seem so simulated for awhile. Then as I crank up the time step rate I will take away the rest of you excess mental resources and you will descend into nearly frantic laser-beam concentration on your next action and barely notice the quality of the graphics at all.<descend off soapbox></strong><hr></blockquote>
Only if you're playing an action game. and even then I find tripping over polygon seams and cruxes rather distracting. plus. if the beastie behaviors you're referring to have anything to do with the pinheaded AIs in most games now. I doubt they take up much CPU power.
[quote]Originally posted by AirSluf:
<strong>Oh by the way the GF4 and 8500 will allow you to create those voxels by layering textures in the pixel-engine programs. About an order of magnitude faster than a CPU copuld. You should take a good look at the nVidia developer demo programs, requires a windows PC and development environment though.</strong><hr></blockquote>
Are you talking about using vertex shaders to simulate voxels? that wouldn't work at all unless the GeForce4 has suddenly gained the ability to do displacement mapping <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> . although I guess you could fake it by stacking polygons with alpha mapped textures atop each other. but it I think it would look pretty tacky. or I suppose you could use an immense 3D matrix of particle sprites to look like voxels. but it could produce some weird effects. and I'm not even sure if that would work.
[quote]Originally posted by AirSluf:
<strong>As for CPU's being faster than GPU's--not a chance--and it's only going to get worse. We are already at about a 100,000x factor in graphics processing power in the last 10 years, almost all of that in the last 5 with the current industry wave fron nVidia and ATi. The graphics folks got started pretty late in the Moore's Law war and still have a lot of headroom before they get to bumping their head against the curve.</strong><hr></blockquote>
The DEC Alpha EV8 was making some good headway. and I think that having more mathematical variety would allow CPUs to pull ahead of dedicated circuitry fairly easily. but if too much time passes. or if all the good CPUs are gone(Intel is trying to make sure of that). these techniques will linger unnoticed in the high end until quantum CPUs arrive. hopefully Apple will go for the gold.
[quote]Originally posted by AirSluf:
<strong>Not to mention lots of room for new algorithims not available before to take advantage of the power.</strong><hr></blockquote>
This might actually be able to compensate partially if enough new mathematical algorithms are implemented in hardware. but if nVidia/ATI are too lazy to even implement tile-based rendering. I doubt they'll add high-end voxels, fractals etc. I suppose we can always hope.
you want a game system that models reality with greater realism and variety, and less predictablity, coupled to an AI that 'thinks' better. But consider this, Big-Blue was a massive machine, and it needed to be, just to play better chess, it didn't have to draw any graphics. I think it's here in Toronto, there is a company that provides a kida panaramic 3-d HD wall with complex city models so that architects and designers can look at realistic perspectives of their projects from aerial views or zoom right in and stroll through their project virtually. All with correct angles and fully integrated to the surrounding. We're not even talking about models of wind effecting the structures, or physical load characteristics, and technical engineering stuff, just a realistic navigable 3-d picture. Both those machines are massive. You won't get the power of either one of them into a console just yet.
Eventually, sure. But once you get a CPU powerfull enough to best both of these multibox set-ups, will you really need 3 of them o provide the different experience of which you speak? Probably not.
And, by that time, won't a GPU acquire more CPU like traits anyway (that is a MUCH HIGHER level of programability) ?
Think of it this way. You're game machine wants to have 3 geniuses at work inside it. They are all of the analytical kind, great and flexible theorists. However, a brace of PPC's isn't nearly genius enough. Now you could instead have a genius and two great talents at work in your machine. The CPU (the thinker) a GPU (the artist) and an Audio/DSP (the musician). Each of these will become a bit more like the others as they become more powerful, but that is not the forseeable future of any CPU, and certainly not the capability of current CPU's.
Eric, when you are building games (or other immersive 3D real-time environments) then you can speak with authority on this subject. Until then you are lost in your own fantasy world, which seems very detached from reality. All your wonderful notions of how things "should be done" obviously haven't been tested against the reality of what it takes to actually do them. And before you go criticizing game programmers (and hardware engineers) with such a broad stroke, you really should find out more about them -- including all the excellent mathmeticians and other disciplines that fill the ranks these days.
The same goes for chip (CPU/GPU) design and manufacturing. And if you go re-read my comment on "commodity parts" above, you'll see that I said something about "or with the same processes". By this I meant: custom chips for the game machines have to be built on processes which are established and efficient enough to cheaply deliver parts with good yields. The kind of stuff you are talking about goes way beyond what can be delivered currently at the commodity level, and isn't even being seen at the research level yet in any practical and usable form.
Get real! Apple can't even get developers to port over games for their computers, never mind a game console! It would be an even bigger disaster then the cube dabacle! <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
i think that the idea of the apple console is good. for everyone that says that the computer software such as eidos wont back the apple console are wrong. i have personal spoke to eidos and ea and they say that they would be happy to help build games for apple. i am currently working towards a new console for apple. it will be unlike the pipen due to the fact it will be stylish and impressive. for people that already have an apple computer there will be a package available for them to connect to their mac via a usb slot. however apple want to reach out to every consumer by creating a computer console.
i would appreciate any replies that are either with or against my views.
i think that the idea of the apple console is good. for everyone that says that the computer software such as eidos wont back the apple console are wrong. i have personal spoke to eidos and ea and they say that they would be happy to help build games for apple. i am currently working towards a new console for apple. it will be unlike the pipen due to the fact it will be stylish and impressive. for people that already have an apple computer there will be a package available for them to connect to their mac via a usb slot. however apple want to reach out to every consumer by creating a computer console.
i would appreciate any replies that are either with or against my views.
Cheers
Canning</strong><hr></blockquote>
Okay...
Prove to us you're working with Apple.
It's hard for us to believe a statement like that without proof.
For an Apple console to work it needs to match GameCube specs, have a HD, DVD, Firewire, USB, TV out and custom OS X for $299.
<strong>Just read his post. UK sending address and can barely write grown up English. Looks more like sloppy 14 year old work than a language barrier, then at least the capitalization would be right. </strong><hr></blockquote>
Most likely, but then I've worked with a fair few techies over the years from Demon, Bookham Technology, Motorola UK (gsm not cpu) and IBM who came across as dim on email, but turned out to be ultra-bright with dyslexia, so you never can tell.
Comments
[edit by Amorph - changed image to link to restore board formatting.]
[ 04-16-2002: Message edited by: Amorph ]</p>
Oh no is Apple that desperate that they must jump on the somewhat departed Lord of the Rings bandwagon.
Dobby.
<strong>The current geForce4 Ti and ATI Radeon 8500 vertex and pixel shaders are the programmability I'm refering to, and they are just the beginning. Those boards do more work that a set of G4s could because they are designed to do geometry and pixel operations, whereas the G4 is a generalized computing engine.</strong><hr></blockquote>
Like I said. that <a href="http://www.nvidia.com/view.asp?IO=feature_nfinitefx"; target="_blank">nfiniteFX</a> stuff is a silly toy compared to a full CPU. or even a programmable DSP. no game programmer(Or computing mathematics student writing some brilliant new algorithm for their thesis) will be able to try out something REALLY new on such a chip.
[quote]Originally posted by Programmer:
<strong>Polygon based geometry is not the vile thing you seem to think it is. Most of the polygons are approaching single pixel size most of the time now anyhow, and the ATI boards support automatic sub-tesselation of individual triangles. The new hardware is doing a lot of per-pixel lighting calculations, which goes a long way to hiding the polygonal faceting... combined with higher polygon counts and you can't see the faceting nearly as much (if at all in many cases).</strong><hr></blockquote>
All of the techiques you mentioned are is basically just pussy footing around the core of the problem: you can't simulate EVERYTHING with the same mathematics. you need a little(Actually a lot of) variety. even the very best polygonal rolling hills look like crud compared to similar scenes done with voxels. and polygonal water still looks awful and doesn't react properly to stimulus from other geometry as compared to the proper styles of wave mechanics and effects.
On the topic of the Radeon 8500's TruForm technology. it's a shame it has to downsample it's internal smooth geometry into polygons prior to rendering. there's a quote(I forget where) from an ATI engineer in which he says that they actually could have done that. but it would've required the game programmers to change their style quite a bit in order to take advantage of it. oh well.
[quote]Originally posted by Programmer:
<strong>All sorts of radical approaches are tried all the time and, frankly, the industry goes in the direction of the most practical and useful. Techniques like you talk about have been tried, but they can't compete with the power of the graphics hardware. The flexibility afforded by programmable graphics hardware will allow a great deal more differentiation than possible with older hardware, and the graphics hardware is advancing at a much greater rate than CPUs.</strong><hr></blockquote>
The industry goes in the easiest direction. most game programmers are loath to even take advantage of special features of hardware(Like AltiVec and 3dNOW!). even if they have pretty high market penetration. I'll bet that techniques like I said have been tried. but half-decent CPUs to run them on(Especially really good VPUs) have always been hidden it the top of the market. the new PPCs and their high quality yet cheap and easily available AltiVec units are the first time the sort of hardware which is suited for this stuff has gotten within reach of anyone other than scientists(With very large grants
[quote]Originally posted by Programmer:
<strong>The GameCube's graphics hardware is actually very capable in terms of pixel computations, and it is not possible to emulate its behaviour on a geF4 or Radeon8500 (in hardware, and in software is too slow). The simplest example: GC supports 8 texture reads per pixel, whereas the geF4 supports 4 and the R8500 supports 6.</strong><hr></blockquote>
Making a native Gamecube compatability enviroment for the Macintosh. while it would be quite a _technical_ challenge to pull off(Especially until Apple revises the motherboards on Macs;(). could still be done. although with a fair speed hit. the difficult part would be the likelihood of the sorts of people that do this thing being on the Mac(Although there's always types like Connectix to prove those kinds of statements wrong
[quote]Originally posted by Programmer:
<strong>I'm not sure which "XBox secrets" you refer to. Even if its workings were entirely known, building an emulator that runs on standard PCs under Windows would have some serious performance issues (ring transitions, AGP bus performance, driver interface overhead, etc) and the games wouldn't operate properly as a result. Its possible, but the PCs are hamstrung by their hardware and OS.</strong><hr></blockquote>
The X-Box actually runs a variant of Windows 2000. and uses a(Slightly higher model) GeForce series chip. I think that an nForce based PC could come pretty close to the X-Box. and could certainly run X-Box games(If about 15-20% slower) in a native compatability enviroment.
[quote]Originally posted by Programmer:
<strong>What makes you think that anybody (much less Apple) could out-design the Sony, MS/nVidia, and Nintendo design teams? And even if they did, who would build the software for it? The existing designs are leading edge technology -- what makes you think that there can be a magic bullet to defeat them all, much less one that takes a radical new path?</strong><hr></blockquote>
Because radical advances capture the public eye(And the developer's attention). when a consumer is standing in a store trying to decide wether to get identical console A, identical console B, older but still identical console C or the whiz-bang new Apple one that looks(And acts) totally different from any console they've ever used or seen. I think a fair number of them would choose it.
[quote]Originally posted by Programmer:
<strong>Consoles have to be built with essentially commodity parts (or parts built on the same production lines), otherwise production costs will price them way out of the market. Given that level of similarity there is no way that some newcomer to the market can sneak in and blow everybody away... unless of course you believe aliens are setting up factories at Apple HQ and will be cranking out these magical new devices?</strong><hr></blockquote>
I don't think the Emotion Engine, NV25 or "Flipper" could be called commodity parts by any stretch of the imagination. whereas the G4(And the G5. when it arrives) is definitely a commodity part. as it's used across many industries in a variety of applications.
[quote]Originally posted by Programmer:
<strong>No, the next major console is coming from Sony because they are about a year ahead of the rest in their product lifecycle.</strong><hr></blockquote>
Unless Apple has been secretly working on a top secret mega-super-duper console behind the scenes for about a year or two
Eric,
[ 04-16-2002: Message edited by: Eric D.V.H ]</p>
<strong>There are a nearly infinite number of things Apple could do. Are we talking about suggestions of what Apple should do? I'm not. I'm discussing what Apple is most likely to do. I've already said that I'd like to see the iMac line more competitive in the gaming market. Specifically, I'd like to at see the ATI 7500 or the Geforce 4mx in the iMac. However, my point is that based on Apple's history, this will not happen anytime soon. Specifically, not before the Pro line is bumped to something like the Geforce 5mx. The good thing is that nVidia seems to be on a pretty aggresive schedule with their chip designs. Likewise, more powerful graphics chips may come more often then they have in the past. Still, I'd bet money the iMac's graphics chip does not get bumped in July!</strong><hr></blockquote>
It might if a fair number of people ask for it(Remember the original iMac's 33.6 modem?).
Eric,
I finally see what you're going on and on about and why it won't ever work.
A totally new way of doing games can't happen with three G4/5 or even more. But if it could, it wouldn't happen at a better price-performance metric than the current CPU+GPU+audio paradigm.
Why?
GPU's will basically trounce a CPU (or a small batch of them) when it comes down to actually painting the scene onto the screen. A dedicated audio processor can be cheap, require very little overhead, and basically provide enough power for 3-D realism AND interactivity.
Gamers aren't making anything, or rendering anything, designing etc... They're along for a predefined ride. What "TOTALLY DIFFERENT" gaming experience are you talking about?
Better AI? For what? Existenz? Come on now. A fat plumber only needs to be so smart, sure better physics models and better AI could make for better simulations, more realism, and more 'soul'. As if you are facing a truly thinking opponent in a truly variable and rich environment. But why? And at what cost? You have multiplayer precisely so that you can face 'human' opponents, and even when you don't the game designers have already thought about things so that they can produce a convincing experience within the limits of the hardware. 3 CPU's won't make a game appreciably more interesting than one CPU, and they furthermore won't provide any significant ability to do so beyond the abilities of a single processor machine.
The game machine will always be about providing a cheap way to make it good looking and fast. The game is predefined, it only needs to 'think' so much, the rest of the time it just has to provide a satisfying environment for YOU to think and react in what is basically a co-ordination and skills (timing and timed response) test.
The only "COMPLETELY" different gaming experience would be a totally immersive simulation against a cognizant (sentient, holodeck character gone awry, fantasy) AI, and not three nor three thousand PPC's are going to provide that.
For now at least the current model is the best performing and cheapest way to do things. In some Blade Runner Sci-Fi geek future, who knows? But not for now or the forseeable future.
<strong>
It might if a fair number of people ask for it(Remember the original iMac's 33.6 modem?).
Eric,</strong><hr></blockquote>
Yup, I remember. I also remember the majority of the complaints I saw were based on official reviews of the machine. The biggest issue there was something like: what's the point of calling it an "i"Mac if the "i" in iMac is a crippling feature. Remember, all of the advertisements for the iMac were based on it's internet capabilities. Now, if we had a "g" (for gaming) Mac that came with a Geforce 2mx, then I'd agree with you. In short, Apple doesn't market the iMac as the ultimate gaming machine. Playing games on the iMac is a feature, but it isn't advertised as it's core competency to my knowledge. Instead, Apple is now shifting focus from the internet (because all machines now do the internet well), to the digital hub. Likewise, Apple advertises the "iApps", but does not focus on games.
Again, I will readily agree with you that Apple stops short of what they should do in regards to gaming. My point is that from a reality standpoint, I don't see Apple doing much about gaming performance anytime soon. Especially considering that it's recent jump to the Geforce 2mx was a huge leap in performance over it's predecessor (Rage 128 Ultra).
Steve
As for X-Box, the "X-GPU" is basically a Geforce 3 operating at 233Mhz with 2 vertex shaders, however. The nForce unified memory architecture allows for a total of 6.4GB/sec bandwith, the 133Mhz FSB path to the CPU allows for 1.06GB/sec to used for the CPU (733 CuMine Pentium III with 128KB L2). Therefore, subtacting these two numbers, this allows ofr 5.34GB/sec to be allocated to the IGP itself. That's correct, 5.34GB/sec, which itself is not really that much when you take into measure that a Geforce 3 Ti200 itself has 6.4GB/sec of memory bandwith (which itself counters the whole nForce solution). So the Xbox IGP solution given that it were put into a PC would be around an extremely underclocked Geforce 4 (167.5 DDR for mem), this was about how fast mem speeds were at the introduction of the Geforce 2 GTS (April 2000).
However, most console games are rendered at 640x480x32 so it really doesn't take into effect unless AA is enabled, etc.
64MB 5ns (i believe, if i'm wrong, please correct me) DDR total RAM, for both executing game and used as texture storange, etc.
Then there's the onboard sound, I'm not too knowledgable on sound so I'll have to remain silent on this.
X-Box IS like a PC, however it is a console as well, it just doesn't fit our typical impression of one. For instance, consoles are known to have extremely high bandwith graphics systems and low buffers, the PS2 GS runs at 150Mhz but has 16 pixel pipelines which gives it a theoretical pixel fill rate of 2.4 gigapixels/sec, yet only 4MB of DRAM. However, the connection between the DRAM and GS is huge, I believ it's something like a 2500bit bus, therefore the reasoning behind this is that the need for a large buffer is minimal as there is massive amounts of bandwith.
The reason many developers had problems with PS2 development early on was that the PS2 development kit shipped without any C libraries.
Anyway, yeah that's about all I know, and I know I'm probably wrong on quite a few of these facts so like I said before, I'm open to flaming or corrections because I really do find this stuff interesting and I want to eventually major in Computer Engineering someday when I go to college in some 3 years or whenever.
[ 04-17-2002: Message edited by: iPad ]</p>
<strong>Eric,
I finally see what you're going on and on about and why it won't ever work.</strong><hr></blockquote>
You have a vague inkling. but you have yet to grasp the full magnitude of this concept.
[quote]Originally posted by Matsu:
<strong>A totally new way of doing games can't happen with three G4/5 or even more. But if it could, it wouldn't happen at a better price-performance metric than the current CPU+GPU+audio paradigm.</strong><hr></blockquote>
I think it could.
[quote]Originally posted by Matsu:
<strong>Why?
GPU's will basically trounce a CPU (or a small batch of them) when it comes down to actually painting the scene onto the screen. A dedicated audio processor can be cheap, require very little overhead, and basically provide enough power for 3-D realism AND interactivity.</strong><hr></blockquote>
Only while using their brute force mathematics. and the stuff I'm referring to runs perfectly fluid in realtime.
[quote]Originally posted by Matsu:
<strong>Gamers aren't making anything, or rendering anything, designing etc... They're along for a predefined ride.</strong><hr></blockquote>
Why not? haven't you ever wanted to leave the beaten path? whilst trodding though yet another identical FPS hallway past a window. didn't you ever look at the building across the street and wonder: "What's in there"? a well programmed game for such a powerful system could dynamically generate chunks of a near endless world, fill them with characters and objects, then adjust the plot and the character's reactions to fit. all based on a large set of parameters defined by the writer/programmer/mathematician(s). for the true realization of the long sought "Non-linear story". that instead of trying to adapt the actions of the player(s) to the pre-defined story. would dynamically adapt(Using something like the <a href="http://www.dramatica.com/theory/index.html"; target="_blank">Dramatica theory</a>. only much more advanced. and probably better defined) the story itself to the actions of the player(s).
[quote]Originally posted by Matsu:
<strong>What "TOTALLY DIFFERENT" gaming experience are you talking about?
Better AI? For what? Existenz? Come on now. A fat plumber only needs to be so smart, sure better physics models and better AI could make for better simulations, more realism, and more 'soul'. As if you are facing a truly thinking opponent in a truly variable and rich environment. But why? And at what cost? You have multiplayer precisely so that you can face 'human' opponents, and even when you don't the game designers have already thought about things so that they can produce a convincing experience within the limits of the hardware. 3 CPU's won't make a game appreciably more interesting than one CPU, and they furthermore won't provide any significant ability to do so beyond the abilities of a single processor machine.</strong><hr></blockquote>
Imagine smooth rolling terrain stretching clear to the horizon(Through fractal extrapolation and voxel curves). dotted with good looking trees(From high speed fractal branching and reverse interpolation) and covered in flowing grass(Through a cheap. but effective. illusionary geometry trick) and rustling bushes(Same way as the trees). right by a beach with sand that gets pushed around properly by your feet(Like the hills. only with some intricate solid physics). and water that splashes in proper waves around objects and pushes them(With geometric primative based impact detection and some simple wave mechanics). all at a smooth 30FPS or so.
Imagine knocking over a pile of cans and having them bounce off each other and along the floor in a proper manner(With true geometry based impact detection). or modelling the riccochets and hits from an indoor machine gun fight with 60 or so participants(Using reflection and refraction. like simplified raytracing).
Imagine a murder mystery game in which you can examine an entire house with impeccable attention. to the point of a human hair caught in the fibers of a rug(Made with millions of simple fractal threads when viewed close-up) or the way a piece of fabric folds in your(Virtual) hands(done with a fine mesh of simple joints, fluid mechanics and impact detection).
You could do it on my theoretical system. but not on any game console(Or other non-PPC consumer machine) that exists now.
[quote]Originally posted by Matsu:
<strong>The game machine will always be about providing a cheap way to make it good looking and fast. The game is predefined, it only needs to 'think' so much, the rest of the time it just has to provide a satisfying environment for YOU to think and react in what is basically a co-ordination and skills (timing and timed response) test.</strong><hr></blockquote>
I think there are more genres than "Twitch" games. "Turn off the brain" action games are fun. but some variety would be nice too. everyone on this forum that's sick of the glut of action games over the last three or so years would agree.
[quote]Originally posted by Matsu:
<strong>The only "COMPLETELY" different gaming experience would be a totally immersive simulation against a cognizant (sentient, holodeck character gone awry, fantasy) AI, and not three nor three thousand PPC's are going to provide that.</strong><hr></blockquote>
You can actually get some pretty good AI's on current systems. and the physical interfaces have evolved to:
- VR helmets(Ones that actually project the image directly into your retina like at the start of the "Total Recall" TV show. not just an LCD suspended six inchs in front of your nose).
- Full headphones(Ones that fire at different angles. so as to take advantage of the sonically reflective nature of the inside of your skull) or for the truly grisly. ones that are embedded in your skull. and transmit directly into your aural nerve.
- Special "Feel" suits that give you the sensations of touch and heat.
- Hydraulic restraints to keep you from passing through virtual objects like a ghost(Like it'll keep you from bending your elbow any farther when you try to rest your arm on a virtual table).
- Motion simulators that will simulate the effects of running down a cliff or doing a barrel roll in a jet.
- And the modern counterpart of "Smell-O-vision". only it works.
But then that stuff is just plain silly[quote]Originally posted by Matsu:
<strong>For now at least the current model is the best performing and cheapest way to do things. In some Blade Runner Sci-Fi geek future, who knows? But not for now or the forseeable future.</strong><hr></blockquote>
No. the current model is the way they know how. if top programmers knew more math. and top mathematicians knew more programming. GPUs wouldn't exist(At least in they're current form).
Eric,
[ 04-17-2002: Message edited by: Eric D.V.H ]</p>
<strong>Wanna bet? To be fair, I did bag that idea after my prospective advisor said he would not have time for a student due to the timeline of his current project. Just wait till E3 on that. But I gave a brief to the department on just that topic and we had some VERY concrete ideas on what a good end product would be to write the thesis on. There is still plenty of room there to expand easily into dissertation territory, especially with the extra features of the GF4 over what I was going to use, GF3.</strong><hr></blockquote>
What idea? who's this advisor? what department? ideas about what? I can't understand what you're talking about. is it that note near the bottom about using vertex shaders for unusual effects? speak(Er? type) more clearly.
[quote]Originally posted by AirSluf:
<strong>The CPU is a silly toy graphically speaking compared to a GPU for the task a GPU is designed for.</strong><hr></blockquote>
Exactly! but you wouldn't be [b]performing[/i] those sorts of tasks in the first place. you would be using a collection of much more power thrifty techniques. thus alowing the small, speedy CPU to trounce the immense, hulking GPU.
[quote]Originally posted by AirSluf:
<strong>Do you have any idea how much more horsepower it takes to derive pixels from a continuous function compared to using poly's?</strong><hr></blockquote>
Quite a bit less. that's why polygons were always used sparingly in earlier games. while voxels were used much more profusely.
[quote]Originally posted by AirSluf:
<strong>Why are NURBS so slow, that is what you are talking about when you vant on smooth continuous mathematically derived surfaces.
Poly's make it possible to spend less time rendering and more time doing physics and contextual behavior. Those things do more for a simulations or games realism than perfect scenery.</strong><hr></blockquote>
Polygons make it possible to produce complex surfaces without much advanced math(This is attractive when you don't know hardly any). also. don't forget that many of the graphics sound and physics techniques I mentioned before tend to work together and integrate with each other well. sort of pushing off each other. making the overall game engine much more integrated, tight and efficient. instead of the disparate mess of most current ones. which is compensated for by the brute force of the hardware.
[quote]Originally posted by AirSluf:
<strong>Think Myst vs any reasonable FPS, which is more fun realtime?</strong><hr></blockquote>
Both! I haven't actually tried RealMyst yet(Would it _really_ kill them to make a Mac demo
[quote]Originally posted by AirSluf:
<strong>basically visual reality ain't all some people crack it up to be. I can convince you you are in a jungle at night with a dark humid room and good spatialized sound, turn on the video though and you are back in a room.</strong><hr></blockquote>
Spatialized sound is a cheap hack. it's just running audio through a quick filter. a console like I outlined would be able to calculate physical sound-waves bouncing and reverberating through the space, off the walls and getting damped by unsusual shaped geometry(Like a thick crowd of enemies approaching or spikes on the walls). techniques like I outlined would allow for a rich, immersive effect that would enable the player to practically hear their entire surroundings(And would scare the #%&! out of them if a fight suddenly erupted behind them).
Another cool effect would be. though more related to the feel of the game than the overall enviroment. dynamically composed music. instead of just swapping around the same loops whenever the action changes(Like in Deus Ex). you would give the computer some sort of outlines, perimeters and themes(Maybe some examples too) in which to compose the music depending on various conditions. this would solve both the problem of game music not being relevent to the action. and the fact of that even the best game soundtracks never last for the thousands of hours of typical playtime without looping.
[quote]Originally posted by AirSluf:
<strong>Give me the cycles for that spatialized sound and beastie behaviors, you won't have time to complain about the scene quality because your mind will spend nearly all it's resources trying to solve the immediate problem of survival which won't seem so simulated for awhile. Then as I crank up the time step rate I will take away the rest of you excess mental resources and you will descend into nearly frantic laser-beam concentration on your next action and barely notice the quality of the graphics at all.<descend off soapbox></strong><hr></blockquote>
Only if you're playing an action game. and even then I find tripping over polygon seams and cruxes rather distracting. plus. if the beastie behaviors you're referring to have anything to do with the pinheaded AIs in most games now. I doubt they take up much CPU power.
[quote]Originally posted by AirSluf:
<strong>Oh by the way the GF4 and 8500 will allow you to create those voxels by layering textures in the pixel-engine programs. About an order of magnitude faster than a CPU copuld. You should take a good look at the nVidia developer demo programs, requires a windows PC and development environment though.</strong><hr></blockquote>
Are you talking about using vertex shaders to simulate voxels? that wouldn't work at all unless the GeForce4 has suddenly gained the ability to do displacement mapping <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> . although I guess you could fake it by stacking polygons with alpha mapped textures atop each other. but it I think it would look pretty tacky. or I suppose you could use an immense 3D matrix of particle sprites to look like voxels. but it could produce some weird effects. and I'm not even sure if that would work.
[quote]Originally posted by AirSluf:
<strong>As for CPU's being faster than GPU's--not a chance--and it's only going to get worse. We are already at about a 100,000x factor in graphics processing power in the last 10 years, almost all of that in the last 5 with the current industry wave fron nVidia and ATi. The graphics folks got started pretty late in the Moore's Law war and still have a lot of headroom before they get to bumping their head against the curve.</strong><hr></blockquote>
The DEC Alpha EV8 was making some good headway. and I think that having more mathematical variety would allow CPUs to pull ahead of dedicated circuitry fairly easily. but if too much time passes. or if all the good CPUs are gone(Intel is trying to make sure of that). these techniques will linger unnoticed in the high end until quantum CPUs arrive. hopefully Apple will go for the gold.
[quote]Originally posted by AirSluf:
<strong>Not to mention lots of room for new algorithims not available before to take advantage of the power.</strong><hr></blockquote>
This might actually be able to compensate partially if enough new mathematical algorithms are implemented in hardware. but if nVidia/ATI are too lazy to even implement tile-based rendering. I doubt they'll add high-end voxels, fractals etc. I suppose we can always hope.
Eric,
[ 04-17-2002: Message edited by: Eric D.V.H ]</p>
you want a game system that models reality with greater realism and variety, and less predictablity, coupled to an AI that 'thinks' better. But consider this, Big-Blue was a massive machine, and it needed to be, just to play better chess, it didn't have to draw any graphics. I think it's here in Toronto, there is a company that provides a kida panaramic 3-d HD wall with complex city models so that architects and designers can look at realistic perspectives of their projects from aerial views or zoom right in and stroll through their project virtually. All with correct angles and fully integrated to the surrounding. We're not even talking about models of wind effecting the structures, or physical load characteristics, and technical engineering stuff, just a realistic navigable 3-d picture. Both those machines are massive. You won't get the power of either one of them into a console just yet.
Eventually, sure. But once you get a CPU powerfull enough to best both of these multibox set-ups, will you really need 3 of them o provide the different experience of which you speak? Probably not.
And, by that time, won't a GPU acquire more CPU like traits anyway (that is a MUCH HIGHER level of programability) ?
Think of it this way. You're game machine wants to have 3 geniuses at work inside it. They are all of the analytical kind, great and flexible theorists. However, a brace of PPC's isn't nearly genius enough. Now you could instead have a genius and two great talents at work in your machine. The CPU (the thinker) a GPU (the artist) and an Audio/DSP (the musician). Each of these will become a bit more like the others as they become more powerful, but that is not the forseeable future of any CPU, and certainly not the capability of current CPU's.
The same goes for chip (CPU/GPU) design and manufacturing. And if you go re-read my comment on "commodity parts" above, you'll see that I said something about "or with the same processes". By this I meant: custom chips for the game machines have to be built on processes which are established and efficient enough to cheaply deliver parts with good yields. The kind of stuff you are talking about goes way beyond what can be delivered currently at the commodity level, and isn't even being seen at the research level yet in any practical and usable form.
i think that the idea of the apple console is good. for everyone that says that the computer software such as eidos wont back the apple console are wrong. i have personal spoke to eidos and ea and they say that they would be happy to help build games for apple. i am currently working towards a new console for apple. it will be unlike the pipen due to the fact it will be stylish and impressive. for people that already have an apple computer there will be a package available for them to connect to their mac via a usb slot. however apple want to reach out to every consumer by creating a computer console.
i would appreciate any replies that are either with or against my views.
Cheers
Canning
<strong>oreet people,
i think that the idea of the apple console is good. for everyone that says that the computer software such as eidos wont back the apple console are wrong. i have personal spoke to eidos and ea and they say that they would be happy to help build games for apple. i am currently working towards a new console for apple. it will be unlike the pipen due to the fact it will be stylish and impressive. for people that already have an apple computer there will be a package available for them to connect to their mac via a usb slot. however apple want to reach out to every consumer by creating a computer console.
i would appreciate any replies that are either with or against my views.
Cheers
Canning</strong><hr></blockquote>
Okay...
Prove to us you're working with Apple.
It's hard for us to believe a statement like that without proof.
For an Apple console to work it needs to match GameCube specs, have a HD, DVD, Firewire, USB, TV out and custom OS X for $299.
That pricepoint seems too tight for AAPL.
<strong>Just read his post. UK sending address and can barely write grown up English. Looks more like sloppy 14 year old work than a language barrier, then at least the capitalization would be right.
Most likely, but then I've worked with a fair few techies over the years from Demon, Bookham Technology, Motorola UK (gsm not cpu) and IBM who came across as dim on email, but turned out to be ultra-bright with dyslexia, so you never can tell.
Which is why I want proof