The Query optimization section is dedicated to explaining concepts relevant to the process of optimizing queries: block computation, cell-by-cell mode, arbitrary shaped sets, varying attributes, sparse and dense calculations, types of cache, and so on.
Block-computation mode, also known as subspace mode, is a process of evaluating cells in blocks. It is a mode that is many times faster than the cell-by-cell mode. Most MDX functions (http://tinyurl.com/Improved2008R2) are optimized to work in that mode. However, by applying non-optimized calculations we can prevent that, accidentally or on purpose.