Parallel parsing of arithmetic expressions

摘要

Parallel algorithms for parsing expressions on mesh, shuffle, cube, and cube-connected cycle parallel computers are presented. With n processors, it requires O( square root n) time on the mesh-connected model and O(log/sup 2/ n) time on others. For the mesh-connected computer, the author uses a wrap-around row-major ordering. For the shuffle computer, he uses an extra connection between adjacent processors, and thus four connections per processor are required.