Analysis of speedup as function of block size and cluster size for parallel feed-forward neural networks on a Beowulf cluster | IEEE Journals & Magazine | IEEE Xplore