END OF AN ERA, FRACTALFORUMS.COM IS CONTINUED ON FRACTALFORUMS.ORG

Here a rendering of a Mandelbulb (probably the most non-commercial object ever) is used to promote a technology conference and in particular an emerging companies summit related to parallel computing with CUDA.

http://www.nvidia.com/object/emerging_companies_summit.html

Uh, was this particular rendering taken from David Makin? Is its use okay in this context?

It looks like Daniel White's image. It is his style of rendering.

Er yep, that one's mine I think

They didn't get permission, but one can't automatically blame Nvidia since it's possible they just bought it from someone who ripped the picture and put it on an iStockPhoto type site.

It's pretty funny, because I'm only using the raw CPU, no GFX card anyway atm.


What do you think will be more suited to fast flexible raytracing (either GI or DI), the upcoming Fermi/GF100, or Intel's Larrabee?

Larrabee is no longer quite so upcoming, unfortunately. I think Larrabee would be a better architecture for ray tracers when you aim for more sophisticated simulation of light, or more sophisticated/complex data structures and algorithms. Fermi, on the other hand, would have more raw power, and would probably outrun Larrabee whenever you can apply brute force or some approximating cheats.

Both are interesting approaches to massively data parallel programming. But in my not so humble opinion, both these chips are trying to fly before they can run. We still haven't gotten all the fundamentals of the SIMD (Single Instruction Multiple Data) paradigm right.

Edit: had omitted one word, thus losing the meaning.

That picture is worth a GTX 480, no less.