CPU + GPU plans

[stevegh]

Will the GPU RT engine be pushed to support as much as possible with respect to CPU RT before release? This may sound like "ok steve dumb question, of course we will try" but the more I use it the more I think it's necessary to be a viable solution, even for previewing. I know you (Vlado) have talked about throwing out warning messages when the hybrid mode is enabled, but I think really if it is enabled you have to limit what the CPU is doing so that the image comes out correctly. And even for non-hybrid mode, having to have a separate workflow (for render setup and materials) would really be a pain in the butt, some of us had seen this with CPU RT.

What CPU RT was great for in my case was lighting. I could throw a material override chrome material on, setup all my cards in real time, and even preview animation. But since you guys have put a crap ton of work into these things, and they really are fast, I would love to invest in some hardware to fully take advantage of your hard work. That said, how far off do we think GPU RT will be given the known limitations? Will it be close and things just need to be implemented, or will time not really matter here until GPU boards start upping the RAM?

Rock on,
-S

[vlado]

I know you (Vlado) have talked about throwing out warning messages when the hybrid mode is enabled, but I think really if it is enabled you have to limit what the CPU is doing so that the image comes out correctly.

Yep, this is already being worked on for the next beta update.

What CPU RT was great for in my case was lighting. I could throw a material override chrome material on, setup all my cards in real time, and even preview animation. But since you guys have put a crap ton of work into these things, and they really are fast, I would love to invest in some hardware to fully take advantage of your hard work. That said, how far off do we think GPU RT will be given the known limitations? Will it be close and things just need to be implemented, or will time not really matter here until GPU boards start upping the RAM?

It is not only RAM that is limited; the GPUs currently have some limitations, which make it impossible (or impractical) to implement everything from the CPU version.

Best regards,
Vlado

[stevegh]

Totally understand and thanks Vlado. The main reason I'm posting this is that IRay for max 2011 is going to ship soon, and they are really pushing it as a production renderer for dummies (ie hey look just push render and its good to go!). While I in no way want RT to be like this, I do really like the idea of a fully unbiased renderer running on the GPU that is as integrated as V-Ray is in max. So other than RAM limitations and some of the computational limitations, is it possible to make RT GPU+CPU a fully functional unbiased rendering solution like so many others out there (Arion-FryRender, Octane, etc...).

[Dariusz Makowski (Dadal)]

Its interesting what they say about Iray... Very curious if that is actually true. And I dont need hundreds of GP graphic card to render it out humhhh gona be interesting September !

Quite curious about the physics... if the light is actually getting bounced foreva or like we got in mental ray, for 3 bounces or so mhmhmh

[vlado]

The main reason I'm posting this is that IRay for max 2011 is going to ship soon, and they are really pushing it as a production renderer for dummies (ie hey look just push render and its good to go!)

And V-Ray RT and V-Ray RT GPU are different in what way? I'm not too worried iray - you will find that many things that work in the regular mental ray renderer do not work in iray either.

So other than RAM limitations and some of the computational limitations, is it possible to make RT GPU+CPU a fully functional unbiased rendering solution like so many others out there (Arion-FryRender, Octane, etc...).

V-Ray RT and V-Ray RT GPU are fully functional unbiased engines right now. They already supports more features than some of the renderers you mention.

Best regards,
Vlado