Technical rendring question

[gaggle]

Totally off :)

Your graphics-card does plenty of computations that could likely be helpful in rendering a scene, but it's oriented strictly towards real-time computations to a monitor. As it stands, it's really hard getting data from the card for instance. It has this massive data-bus that can transfer tons and tons of data, but only one way.. "that's dumb" some might say, but the only reason a graphicscard is so good at what it does is because it's been super-specialized in doing this one thing. Oh, note that when I say getting data back from the card I don't mean the rendered image itself, you can save that output, but graphicscards only offer a very small array of things it can compute (for instance, off the top of my head, it could do backface culling and other misc. polygon-data-preparation, but it wouldn't be able to render some specialized volumetric effects.. it just wouldn't know how to) .. anyway, since it could help with only some of the computations, the resulting raw scene-data would have to be sent back to the computer for further processing, and that's not doable.

Also, it renders out stuff at the expense of precision too: you might've noticed in games or the MAX viewports that things can flicker if they get far away, because the way the graphicscard handles depth-sorting ("this part is infront of this part, but behind this other part") is done with very limited precision.. again, because that makes it go faster.

These two things, one-way data and lack of precision, means you can't, and wouldn't want to, use your graphicscard for rendering out scenes. Exept when it comes to making previews, which essentially is like letting your graphicscard handle the rendering. It's not pretty, not pretty at all, but it's mad mad fast.

Note that this is all with the current crop of graphicscard, nVidia in particular is trying to take their next-generation cards up to a level where they could actually substitute a CPU. As far as I know they've got it going pretty well, it's not perfect, but it's slowly getting there. They're accomplishing that mainly on two fronts as far as I know, a) by increasing the precison the card renders with, and b) by making the processor on the graphicscard (GPU) so versatile so they can essentially "do anything".

Hooray, all done ranting now :)

[Chaos Maximus]

thanks, i dont mind being compleetly off, and im happy to be informed. thanks, that does make sense thanks for the info.

guess ill have to cough up the money for more memory and that second processor at some point. that will be fun im sure.

thanks, have fun

[CGTalk Moderation]

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.