"When you look at graphics, ray tracing is a niche. It’s great if you have highly reflective surfaces or if you have transparency with reflection and refraction. Or when you have silhouette edges or soft shadows. But for a lot of other things, you want rasterization. You will see hybrid systems with rasterization and ray tracing only when you need it. You don’t want to burden the system with a lot of special purpose hardware that costs a lot."
Great article. It explains alot of where were going
"When you look at graphics, ray tracing is a niche. It’s great if you have highly reflective surfaces or if you have transparency with reflection and refraction. Or when you have silhouette edges or soft shadows. But for a lot of other things, you want rasterization. You will see hybrid systems with rasterization and ray tracing only when you need it. You don’t want to burden the system with a lot of special purpose hardware that costs a lot."
Great article. It explains a lot of where were going
That just doesn't hold up to the reality as i understand it JD. Please correct me if i am wrong here but i was under the impression that doing that sort of thing (Ray tracing) in software carried one hell of an overhead as far as computing power needed to do it in software is concerned. Which was the reason given for not expecting a full ray tracing unit any time soon back when larrabee was first mentioned.
I would have thought doing it in hardware when needed (if possable) would be the easier cheaper and prefered method ?
I'm thinking along the lines of ATI and its rasterisation units, tried it being done back in the shaders and it didnt work so in came the hardware units. Or am i being to simplistic with this ?
I mean i cant see it being overly expensive having some extra parts on the core otherwise ATI wouldn't have been putting a tessellation unit on to there products for no real reason for so long.
The article is stating that to have a full RT unit would be too costly, so thered be compromising, maybe some dedicated HW , or done in SW, but thatd be with a hybrid scenario, where you could get away with a minimal amount of dedicated HW.
So yes, dedicated HW is the way to go, but in a hybrid form, wouldnt be so costly, as itll be mainly rasterization, therell be a smaller amount for those times when RT is used.
To do it in SW, it really couldnt be done due to latency, and match the ongoing rasterization, IMHO.
------------------------------I went drifting, thru the capitols of tin, where men cant walk and cant freely talk, and sons turn their fathers in
Reply to jaydeejohn
My point is that he is proposing to do in software what others are/may be going to do in hardware. Given the overhead we all know this costs and given that ATI have already shown that you can put redundant features on to a solution and still easilly beat the oposition price wise, well it just seems wrong to me. Im reading between the lines a bit here but it sounds very much as though he is saying Nvidia wont use hardware at all.
" My intellectual curiosity makes me want to find out what they are doing. But the more pragmatic part of me says we can do in software what they’re proposing to do with special hardware. Why would we want to put special purpose hardware in that won’t be used when you’re not doing ray tracing?"
Now of course he knows many times more than i do about it but if the hardware part of it is that expensive then given the global financial situation at the moment they had better all not bother just yet.
If you have any more links on the subject i would love to see them.
I got the overall thought from reading it, that RT will be read in SW, and have set aside HW when requested. I think thats what hes trying to say here. Itll still have huge costs involved, but more wisely? used.
To me, thats the only way youll be able to blend the 2 together. Having just permanent dedicated HW, sitting doing nothing til needed is a cost Im sure none of the gpu makers will pursue. It goes directly against the trend of having everything unified, and reducing dedicated as much as possible.
This does open the door for Caustic, ala Ageia, for a dedicated HW for RT, if its that important/costly.
Doing RT thru shaders? If its broken down to certain levels, it could be done. Time will tell. Between RT and Physx, plus rasterization, all of todays "good enuff" cards certainly wont be good enuff, if games start using/requiring these things
------------------------------I went drifting, thru the capitols of tin, where men cant walk and cant freely talk, and sons turn their fathers in
Reply to jaydeejohn
Well i guess we will just have to wait and see how it plays out. Im going to look into how this RT hardware works and what it involves specifically.
I need to understand it better so i can get it straight in my head as to what the differance is with ATI having a Tessalation unit when its not really needed and nvidia saying that RT hardware that isnt really needed would be to costly. As i said before the seeming similarity of the two and the fact that it didnt/dosent stop ATI making a decent card at a good price is bothering me.
You are about to answer a thread that has been inactive for more than 6 months. If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.