| In this thesis, vectorization of a computation elimination approach to Marching Cubes isosurface extraction is studied. The vectorization is achieved by using SIMD within a Register (SWAR) capabilities on general-purpose microprocessors. The computation elimination approach uses the branch-on-need-octree (BONO). The methods to achieve vectorization as well as experiments that explore using vectorized BONO with vectorized Marching Cubes are described. The approach described provides better performance than using just vectorized Marching Cubes alone on general-purpose microprocessors. |
