Illustration of the fundamental difference between the computing speeds that can be achieved on isolatitude and nonisolatitude point sets. In order to be able to perform the necessary computational work in support of multimillion pixel spherical data sets one has to resort to isolatitude structures of point sets/sky maps, e.g., HEALPix. Moreover, future requirements are already fairly well established—measurements of the CMB polarization will require huge multielement arrays of detectors and will produce data sets characterized by a great multiplicity (~1000) of sky maps. Since there are no computationally faster methods than those already employed in HEALPix and global synthesis/analysis of a multimillion pixel map consumes about 10³ s of CPU time on a standard serial machine, the necessary speed‐up will need to be achieved via optimized parallelization of the software.