Vector Models for Data-Parallel Computing

ion,level of, 24access-fixed code, 150, 217access-restricted code, 149, 217algorithm,biconnectivity, 114binary search, 94closest pair, 98halving merge, 54k-D tree, 96line-drawing, 50line-of-sight, 40, 105linear-systems, 118matrix-vector multiply, 117maximal-independent-set, 114maximum-flow, 114merge-hull, 103minimum-spanning-tree, 109outer-product, 123quickhull, 101quicksort, 46simplex, 122sparse matrix-multiply, 123split-radix-sort, 43algorithmic models, 1, 37collection-oriented, 5processor-oriented, 5algorithms,branch-and-bound, 49computational geometry, 93graph, 109numerical, 5, 117all-closest-pairs, 98allocating,elements to processors, 24, 48APL, 5, 10, 131compiling, 194APL2, 10append, 65, 94, 224append-bits, 223application of program to data, 157apply-to-each, 135array ordering, 91 back solution, 118backward scans, 217implementation of, 203barrier synchronization, 9Batcher’s bitonic sort, 44BBN Butterfly and Monarch, 202biconnectivity algorithm, 114binary search, 94bit pipelining, 198bit-vector machine, 30bitonic merge, 27bitonic sort, 44boolean circuit, 2, 4comparison to V-RAM, 28level in, 28bounds on element complexity, 153bounds on step complexity, 153branch-and-bound, 49branch-packing, 144, 154Brent’s scheduling principle, 57, 195bridges, 104butterfly networks, 28 c-rank-split, 70capacity on an edge, 83carry bit, 198changes from thesis, 216chess-playing, 49, 53closest pair algorithm, 98, 146CM-Lisp, 163code replicating, 143collapse, 168, 182