mental ray on DARWIN cluster?

[aglick]

I've looked into this...

There were a spate of academic research papers written in the past decade about how to implement distributed cinematic frame rendering on a clustered environment.

I'm pretty sure that the general concensus is that up till now, this has not been a very efficient approach.

It turns out that distributing/calculating the data required to render mutliple frames on multiple machines is most efficiently handled by a standard multi-node ethernet network.

However, There IS currently some news from SGI/Microsoft/Autodesk/PipelineFX(Qube!) about clustered HPC computing for archetectual/engineering applications such as Bioinformatics, computational fluid fluid dynamics, finite element analysis, etc....

Also, it looks like there IS something happening again on "clustered" frame rendering...but I really have my questions about the cost effectiveness of this approach...

http://www.sgi.com/company_info/newsroom/press_releases/2007/june/gobelins.html


Adam
BOXXlabs

[davegraham]

Ainkaboot (http://www.ainkaboot.co.uk/render-farms.php)and I are working on pushing out clustered rendering and, for what it's worth, Hanni (the Ainkaboot guy) is on the Dr. Queue development board. We're using a specially customized version of KlustOS and render nodes based on Core 2 Duos. :)

anyhow, like Adam said, costs can definitely scale with this type of technology but, at a certain point it does make sense.

cheers,

dave

[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.