Re: [Algorithms] Cache-oblivious layout (was: BSP tree, or what?)

[Willem H. de B. <wi...@wh...>]

----- Original Message ----- 
From: "Marc B. Reynolds" <mar...@or...>
>
> Around 2001 (for a previous employer) I did some toy experiments ordering
> large octtree memory via 3D-Hilbert curve indices and was seeing 
> worthwhile
> performance gains.

That sounds interesting. Could you expand on this a bit? In what order
are the nodes stored this way? What was the performance gain, in what
kind of situations did it behave well/badly ?

Cheers,
Willem