Architecture-independent Loop Parallelisation
Architecture-independent programming and automatic parallelisation have long been regarded as two different means of alleviating the prohibitive costs of parallel software development. Building on recent advances in both areas, Architecture-Independent Loop Parallelisation proposes a unified approach to the parallelisation of scientific computing code. This novel approach is based on the bulk-synchronous parallel model of computation, and succeeds in automatically generating parallel code that is architecture-independent, scalable, and of analytically predictable performance.
What people are saying - Write a review
We haven't found any reviews in the usual places.
Data Dependence Analysis and Code Transformation
Communication Overheads in Loop Nest Scheduling
9 other sections not shown
abstract syntax tree affine transformation application architecture-independent array block array reference automatic parallelisation block-cyclic broadcast initialisation loop broadcast loop nest BSP schedule communication and synchronisation communication cost communication overheads coordinates corresponding data dependence data partition decomposition dependence analysis dependence graph direction vectors distance vectors endfor execution flow dependences footprint forall Gaussian elimination imperative programs implementation integer iteration points iteration space iterative scheduling loop body loop indexed loop interchange loop nest scheduling loop parallelisation loop skewing LU decomposition main computation loop main statement nest in Figure obtained optimal outermost loop paral parallel code parallel computer parallel loops parallel schedule parallel version parallelisation techniques parameters perfect loop nest performance potential parallelism processor programming pseudocode rectangular tiles scheduling techniques sequential code step strongly connected components subscripts superstep template-matching scheduling Theorem tightly nested loop tiling matrix tion tool topologically sorted uniform-dependence loop nest upper bound variable wavefront block scheduling