previous index next
Here are the numbers. This table shows the average number of required collapse weight re-computations for a particular edge before it is actually collapsed. As you can see the number for a Progressive (Tetrahedral) Mesh is roughly constant at 240 re-computations per edge. And we see that the number required for the Block Iteration algorithm is significantly less, although it varies depending on the target tetrahedral count, but more on that later.

We've profiled our code and determined that this weight re-computation dominates the running time for our Block Iteration Algorithm (approximately 80%) so the savings on re-computation compared with a Progressive Mesh is considerable.