The reader will note that this guide appears to deal exclusively with dense linear algebra. While the presented infrastructure does indeed target this domain, it should be noted that many applications that lead to sparse linear algebra problems generate sparsity in the matrix such that there are dense sub-blocks. We have tried to make sure that none of the design decisions for PLAPACK inherently prevent the presented kernels from being used in such a situation. Indeed, in [] we hint at the fact that PBMD may provide a common view for data distribution for parallel dense, sparse iterative, and sparse direct methods.