CONTEXT-FREE PARSING ON O(N) PROCESSORS

The growing availability of multiprocessors makes it interesting to treat compilation itself as an algorithm to be parallelized. One approach is to allocate a processor to each symbol or token of the source program and find massively parallel algorithms for the various phases of compilation. The well-known Cocke-Younger-Kasami (CYK) algorithm for parsing context free languages runs in O(n3) time on a uniprocessor. The CYK algorithm can be adapted to run on O(n) processors in O(n2) time. The number of processors in the present generation of multiprocessors makes this a reasonable practical approach.