PDA

View Full Version : nv30 and r300: the next generation of 3d processing


Array
08-13-2002, 08:49 PM
http://tech-report.com/etc/2002q3/nextgen-gpus/index.x?pg=1

Goon
08-13-2002, 09:44 PM
Cool.
But the effects seems on the production end seem to be less spectacular. Someone in another thread mentioned that polygon pushing power hasn't really advanced that much. This article focused mainly upon devices to replace high poly counts, ie. displacement maps, shaders. All this is great for games in which the data has already been baked, but unless cg software fully incorporates pixel and vertex shading to display its shader and particle effects, these advances seem to have little impact whilst smaller increases are made in clock speed and bandwidth.
Am i immmensly wrong here?

Lorecanth
08-14-2002, 09:58 AM
I don't think anyone is getting the grasp of what this really means. Renderman files compiled into opengl multipasses= when we hit the button we can watch full lighting previews in realtime= renderfarms start being created out of mulitple graphics boards.

No more waiting. This doesn't apply to just games but to every facet of the 3D industry.

From above article

"Note that this doesn't mean that technical directors at the film studios will have to learn a new language -- there will be translators that will go from existing languages. Instead of sending their RIB code to the renderfarm, you will send it to a program that decomposes it for hardware acceleration. They will return image files just like everyone is used to.
Multi chip and multi card solutions are also coming, meaning that you will be able to fit more frame rendering power in a single tower case than Pixar's entire rendering farm. Next year.

I had originally estimated that it would take a few years for the tools to mature to the point that they would actually be used in production work, but some companies have done some very smart things, and I expect that production frames will be rendered on PC graphics cards before the end of next year. It will be for TV first, but it will show up in film eventually. "

Marcel
08-14-2002, 11:08 AM
Originally posted by Lorecanth
I don't think anyone is getting the grasp of what this really means. Renderman files compiled into opengl multipasses= when we hit the button we can watch full lighting previews in realtime= renderfarms start being created out of mulitple graphics boards.

No more waiting. This doesn't apply to just games but to every facet of the 3D industry.


These new cards are the first consumer graphics cards that have the ability to render with a precision high enough to use for production.
That doesn't say that it renders in realtime though. Using multiple passes to render complex shaders will probably slow it down a lot. (But still much faster than using your CPU)

I am wondering wheter these cards will also be able to speed up more complex rendering tasks. Can they be programmed to cast soft shadows for example or even create caustics and radiosity?
I hope they can, else such a card would be severly limited for rendering purposes. What good is 4 fps rendering when you can't reach the quality that you need?

ambient-whisper
08-14-2002, 12:12 PM
its just hype for now. calm down till they release it. if they indeed release something thats as revolutionary as they claim... then great:) but until then its just building up hype around something we really havetn personally had much insight into yet. i really doubt the next get videocards willbe able to do what they claim in real time. itll be a first gen for that technology. and will definitely take a while to get adopted by the industry.

remember the geforce 1? environment mapped reflections...gf2 bump mapping...( man not many games use that one STILL ) per pixel shaders? its just now getting around. but thats after a year and a half of the cards inception. and still its not used as widely as it should be.

im giving the industry about 2-3 years until this technology gets a foothold in the industry.

Marcel
08-14-2002, 12:26 PM
Ofcourse it's all a big hype, but I'm hoping it contains a bit of truth.

GPU's are completely designed for doing 3d stuff, unlike the avarage CPU, that's why I think it will be faster. Absolutely not realtime rendering, but I will be very content if these cards will do half of what they are claiming right now. :)

Edit: These cards can probably be compared to the Renderdrive system: pretty fast but limited featurewise.

Array
08-14-2002, 05:01 PM
these cards are NOT like the renderdrive. the renderdrive is a hard coded renderer. the nv30 and r300 promise full programability.

if you want to read more, i have a thread a little bit further down which talks about a plugin that the r300 based cards will use to render maya scenes.

Goon
08-14-2002, 06:36 PM
Seems i stuck my neck out a little far.
The point i was trying to make is that for tasks such as modelling and animation where you deal with superhigh poly counts, these cards are still struggling. Yes it is wonderful to have pixel shaders and bump mapping, but the difference in raw performance between my Gf4 Ti and Gf2 really hasn't seemed that great. Both bog down pretty quickly. Yes there have been substantial increases in these areas but it seems that the newer cards are focusing increasingly on work arounds to achieve the same effects.
Oops i did it again.:D

pearson
08-16-2002, 07:03 PM
I definately agree with you, Goon. I work in games, with scenes no larger than 20k polys. Up until a couple of months back I was using a Geforce 256. It chugged if I turned on the textures (I use Maya), and chugged some even without textures. Then they got me a Quadro 4. I was excited to have all that power, but in reality, it still chugs with the textures on. Not quite as bad, but not nearly the jump in performance I was expecting for a 3 generation jump! (geforce 1 to a geforce 4)

CGTalk Moderation
01-13-2006, 03:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.