It's far from a trivial problem anyway, and there will always be major tradeoffs as advecting across a large domain is intrinsically not friendly to divisions that don't require either immense overhead or obtrusive management, so they might very well have something mind-blowing cooking.
The claim of "first" though is entirely out of place technologically or commercially, they could have played it better. Different partitioning and boundary management techniques and tools with different approaches to the problem have been available to everybody with some cash at hand for a while now. On Tue, Aug 20, 2013 at 10:01 AM, Ciaran Moloney <[email protected]>wrote: > You're right. As I hit send I remembered the R&H talk at Siggraph 2009 > where they lamented that Houdini's distributed solver wasn't quite up to > snuff (visible discontinuities between domains) and that was for RnD back > in 2008. Point being, this stuff aint all that new. > >
