I wonder how long it will be until we achieve the required 450 million ray segs per second to achieve raytracing...
...I'm guessing just a few more years ;).
Originally Posted by EK-MDi I wonder how long it will be until we achieve the required 450 million ray segs per second to achieve raytracing...
...I'm guessing just a few more years ;).
Not sure if thats sarcastic or whatnot, but if a single core P4 3.2 can do 100 million, then a quad core core 2 duo should be able to hit the 400 million easily considering how much higher a performance it has over the P4 line with lower clock speeds and a much higher FSB
Originally Posted by DarkLord7854 Not sure if thats sarcastic or whatnot, but if a single core P4 3.2 can do 100 million, then a quad core core 2 duo should be able to hit the 400 million easily considering how much higher a performance it has over the P4 line with lower clock speeds and a much higher FSB
Yup, so dual core conroe may not be enough. But quad core would be.
This article is fantastic, it's one of the few things I've read in the tech world in the last 3-6 months that's got me enthusiastic and hopefull about the future. ATI being bought by AMD shook stuff up, quite possibly not for the better, but at least we can rely on Intel to bring us the capability to raytrace. Having seen some pictures of raytraced stuff, I gotta say, it blows even things like cyrsis way outta the water. You're talking near photo realism here.
If this is where the future is, then things are looking realisticly peachy.
I think that would only be the case if the GPUs are designed around raytracing, or at least designed to be compatible. I was under the impression that they're not, but then again I could be way off there.
Anyways, excellent article. Like spec, I'm quite excited over the prospects of raytracting.
I do remember reading an article on why AMD aquired ATI If AMD and Intel are hurriedly working on micro-cores in order to be the first to market, it will be the ultimate in parallelism. like the author states in the article, raytracing is easily paralleled. so if you've got 52 micro cores, even at slower speeds (say 1 ghz) raytracing in realtime will be a snap.
Surely if it were that simple, and linear in scaling, someone with a 4 processor opteron board, with dual cores in each would be showing off their real time raytracing by now? It's not as if we've got to sit around waiting for Kentsfield for quad cores. There are systems out there available now, that can have 8 cores. There's more to a game than the 3d, and raytracing is not the only physics calc going on either. Then there's the AI etc... oh, and sound, and interfacing, and whole loads of other stuff... Nah... I'm going to be the sceptic in this thread, and sit here like a great big party pooper, and say "Bah Humbug" a great deal... I think it's gonna be some times before this happens.
I've just heard too many predictions go horribly wrong in IT. For all the great changes since my first ever PC... a Pentium 200MMX, a great deal has also not changed. Conroe and Kentsfield are good, but they're not revolutionary, and it's not as if there aren't quad core alternatives that have been available for some time. Maybe having four physical chips is not as efficient as having them on the same die, but surely Kentfield will not be THAT much faster than four fast opterons on the same board? Am I missing something about parallelism here?
I'll just sit back and watch. I'm assuming Conroe and Kentsfield will be interchangable anyway, so it's not as if it's a big deal to see what happens when it's released, and people can jump on the core2duo bandwagon now, and upgrade later hopefully.
4 cores tho!! Even with the new Intels having lower power requirements, I'm starting to shudder at the thought of how much power these rigs will consume.
Originally Posted by DarkLord7854 Not sure if thats sarcastic or whatnot, but if a single core P4 3.2 can do 100 million, then a quad core core 2 duo should be able to hit the 400 million easily considering how much higher a performance it has over the P4 line with lower clock speeds and a much higher FSB
Yeah true, but thats assuming you use the entire core for graphics and nothing else. I guess if there was some sort of dual quad-core mobo planned for the future you could do it!
Pookeyhead, bear in mind that we are only just beginning to believe of the possibility of real-time ray tracing with desktop PCs. In the past, raytracing has been firmly in realm of realistic (i.e. not real-time) graphics rendering. With intel and AMD churning out multi-core processors and the fact that AMD has bought a graphics company this would well be the start of a possible future of real time raytracing being done on our PCs. We're not saying it'll be here any time soon, just that it has sunddenly become a real possibility.
About the article, I don't get how games could be raytraced without textures. Most raytracing demos you see don't use textures but for a game surely they would be essential right?
Great article; my mind has been poking at the very tips of these concepts on occasion; I dont have the gift of being able to tie it all together and saying it so eloquently. Great job ;)
Originally Posted by Pookeyhead Surely if it were that simple, and linear in scaling, someone with a 4 processor opteron board, with dual cores in each would be showing off their real time raytracing by now? It's not as if we've got to sit around waiting for Kentsfield for quad cores. There are systems out there available now, that can have 8 cores. There's more to a game than the 3d, and raytracing is not the only physics calc going on either. Then there's the AI etc... oh, and sound, and interfacing, and whole loads of other stuff... Nah... I'm going to be the sceptic in this thread, and sit here like a great big party pooper, and say "Bah Humbug" a great deal... I think it's gonna be some times before this happens.
I've just heard too many predictions go horribly wrong in IT. For all the great changes since my first ever PC... a Pentium 200MMX, a great deal has also not changed. Conroe and Kentsfield are good, but they're not revolutionary, and it's not as if there aren't quad core alternatives that have been available for some time. Maybe having four physical chips is not as efficient as having them on the same die, but surely Kentfield will not be THAT much faster than four fast opterons on the same board? Am I missing something about parallelism here?
I'll just sit back and watch. I'm assuming Conroe and Kentsfield will be interchangable anyway, so it's not as if it's a big deal to see what happens when it's released, and people can jump on the core2duo bandwagon now, and upgrade later hopefully.
4 cores tho!! Even with the new Intels having lower power requirements, I'm starting to shudder at the thought of how much power these rigs will consume.
Great article though: Well written and amusing.
Wouldn't it be safe to assume that a system using two or more quad core proc's should be able to realistically handle a game using raytracing? Sure power consumption would be a monster, but with these 5.25" 250w supplemental psu's and 1kw main PSU's it shouldn't be THAT much of problem for the well off enthusiast. Right?
I'm personally under the semi-informed/semi-educated opinion that games featureing raytracing are 5yrs or less in the future. Would I stake a sizeable bet to that? HELL NO! But I'd put a decent figure to it.
You would still need to create light sources and define what wavelength and amplitude of light they produce. Instead of having a graphic texture attached to them surfaces would have to be defined by what wavelengths they reflect in certain areas (although the net result is a fairly similar effect as most surface profiles would resemble textures, and bump-mapping to scatter the light rays effectively would probably feature as well.
Volumes would be given an index of refraction so that air and water interfaces would work realistically, although getting water to ripple correctly would still present some challenges - fluid dynamics is a nasty thing to simulate. Dust in the air would create additional work for the processor
The way I'm thinking about it, games would be made in a similar way - the only difference is in how the graphics engine renders things. If anything it may become slightly easier, as instead of HDR and all that having to be created by additional effects it would occur as a natural feature of the raytracing process.
The more I think about it the more issues it throws up, but they can all be solved with sufficient computational muscle - the emergence of which is, after all, what kicked this whole debate off :)
Disclaimer: All of the above is total guesswork by someone who may well not have a clue what they are on about
If raytracing kicks off in a big way am I right in thinking Nvidia will no longer have a market for it's graphics cards as the CPU will do all the calculations?
Originally Posted by Kipman725 If raytracing kicks off in a big way am I right in thinking Nvidia will no longer have a market for it's graphics cards as the CPU will do all the calculations?
Not really, dedicated raytracing hardware already exists, but is restricted to the professional market, and as such are crushingly expensive for even the most enthusiastic home user.
Anyway, there is going to be a huge legacy market considering how long fans keep playing their favourite games. Though it could get interesting, should the changeover happen in some Twilight Zone-esk overnight conspiracy of silence fashion. "Open-GL and Direct 3D compatible cards? I'm not certain what your talking about sir, but might I be able to interest you in this new Nvidia RayForce 6000"
I'd imagine the changeover will be a gradual thing, with extra effects being added as they can be done easly, little bits of raytracing code that can work on the current graphics hardware, or on the CPU using software renderers. I'd also imagine that support for the old raster standards will be continued for quite some time. Of course it may be that adding in a raytracing board may be the next path for dual card graphics, a current style card and a new style raytracing card.
Personally I'm rather salivating at the thought of dedicated raytracing hardware coming down to the consumer level, being able to use in a dedicated coprocessor in rendering jobs alongside the CPUs sounds like a great way to reduce rendering times. Or it could be tied into the realtime preview for modeling packages, to be able to what the model would really look like, rather than the half textured approximations that are all current modelling packages can manage.
Personally, the thought of things looking that realistic, make me amazed. The idea that quad cores could start stuff like that going makes me rather excited.
Originally Posted by Zayfod Anyway, there is going to be a huge legacy market considering how long fans keep playing their favourite games. Though it could get interesting, should the changeover happen in some Twilight Zone-esk overnight conspiracy of silence fashion. "Open-GL and Direct 3D compatible cards? I'm not certain what your talking about sir, but might I be able to interest you in this new Nvidia RayForce 6000"
I may be totally off the mark here, but aren't graphics cards just processor cores that are optimised for graphics processing? I.e. theoretically could a powerful enough processor emulate a graphics card?
If this was possible, maybe through software emulation (yes, I know you're all wincing at the idea of software acceleration but a game from five years ago could probably be emulated on a core 2 duo), then raytracing cards may also be able to adapt to standard graphics processing at a significant loss of efficiency/power
Comments 1 to 25 of 49
typo :P
Im still running an x800xl i can only dream of things like this.
...I'm guessing just a few more years ;).
Not sure if thats sarcastic or whatnot, but if a single core P4 3.2 can do 100 million, then a quad core core 2 duo should be able to hit the 400 million easily considering how much higher a performance it has over the P4 line with lower clock speeds and a much higher FSB
This article is fantastic, it's one of the few things I've read in the tech world in the last 3-6 months that's got me enthusiastic and hopefull about the future. ATI being bought by AMD shook stuff up, quite possibly not for the better, but at least we can rely on Intel to bring us the capability to raytrace. Having seen some pictures of raytraced stuff, I gotta say, it blows even things like cyrsis way outta the water. You're talking near photo realism here.
If this is where the future is, then things are looking realisticly peachy.
Anyways, excellent article. Like spec, I'm quite excited over the prospects of raytracting.
;)
but I can't not say:
I've just heard too many predictions go horribly wrong in IT. For all the great changes since my first ever PC... a Pentium 200MMX, a great deal has also not changed. Conroe and Kentsfield are good, but they're not revolutionary, and it's not as if there aren't quad core alternatives that have been available for some time. Maybe having four physical chips is not as efficient as having them on the same die, but surely Kentfield will not be THAT much faster than four fast opterons on the same board? Am I missing something about parallelism here?
I'll just sit back and watch. I'm assuming Conroe and Kentsfield will be interchangable anyway, so it's not as if it's a big deal to see what happens when it's released, and people can jump on the core2duo bandwagon now, and upgrade later hopefully.
4 cores tho!! Even with the new Intels having lower power requirements, I'm starting to shudder at the thought of how much power these rigs will consume.
Great article though: Well written and amusing.
Yeah true, but thats assuming you use the entire core for graphics and nothing else. I guess if there was some sort of dual quad-core mobo planned for the future you could do it!
About the article, I don't get how games could be raytraced without textures. Most raytracing demos you see don't use textures but for a game surely they would be essential right?
Great article; my mind has been poking at the very tips of these concepts on occasion; I dont have the gift of being able to tie it all together and saying it so eloquently. Great job ;)
Wouldn't it be safe to assume that a system using two or more quad core proc's should be able to realistically handle a game using raytracing? Sure power consumption would be a monster, but with these 5.25" 250w supplemental psu's and 1kw main PSU's it shouldn't be THAT much of problem for the well off enthusiast. Right?
I'm personally under the semi-informed/semi-educated opinion that games featureing raytracing are 5yrs or less in the future. Would I stake a sizeable bet to that? HELL NO! But I'd put a decent figure to it.
So if raytracing was a reality now, how would it effect how games are made?
Volumes would be given an index of refraction so that air and water interfaces would work realistically, although getting water to ripple correctly would still present some challenges - fluid dynamics is a nasty thing to simulate. Dust in the air would create additional work for the processor
The way I'm thinking about it, games would be made in a similar way - the only difference is in how the graphics engine renders things. If anything it may become slightly easier, as instead of HDR and all that having to be created by additional effects it would occur as a natural feature of the raytracing process.
The more I think about it the more issues it throws up, but they can all be solved with sufficient computational muscle - the emergence of which is, after all, what kicked this whole debate off :)
Disclaimer: All of the above is total guesswork by someone who may well not have a clue what they are on about
Not really, dedicated raytracing hardware already exists, but is restricted to the professional market, and as such are crushingly expensive for even the most enthusiastic home user.
Anyway, there is going to be a huge legacy market considering how long fans keep playing their favourite games. Though it could get interesting, should the changeover happen in some Twilight Zone-esk overnight conspiracy of silence fashion. "Open-GL and Direct 3D compatible cards? I'm not certain what your talking about sir, but might I be able to interest you in this new Nvidia RayForce 6000"
I'd imagine the changeover will be a gradual thing, with extra effects being added as they can be done easly, little bits of raytracing code that can work on the current graphics hardware, or on the CPU using software renderers. I'd also imagine that support for the old raster standards will be continued for quite some time. Of course it may be that adding in a raytracing board may be the next path for dual card graphics, a current style card and a new style raytracing card.
Personally I'm rather salivating at the thought of dedicated raytracing hardware coming down to the consumer level, being able to use in a dedicated coprocessor in rendering jobs alongside the CPUs sounds like a great way to reduce rendering times. Or it could be tied into the realtime preview for modeling packages, to be able to what the model would really look like, rather than the half textured approximations that are all current modelling packages can manage.
http://www.zen85473.zen.co.uk/300px-Glasses_800_edit.png
(Full size @ http://upload.wikimedia.org/wikipedia/commons/e/ec/Glasses_800_edit.png)
Personally, the thought of things looking that realistic, make me amazed. The idea that quad cores could start stuff like that going makes me rather excited.
here is the video
i dont think 2007 but 2008 sound like when we start moving towards it and 2009 will be big year for raytracing
If this was possible, maybe through software emulation (yes, I know you're all wincing at the idea of software acceleration but a game from five years ago could probably be emulated on a core 2 duo), then raytracing cards may also be able to adapt to standard graphics processing at a significant loss of efficiency/power