Instruction sets for Parallel Random Access Machines
The computational powers of time-bounded Parallel Random Access Machines (PRAMs) with different instruction sets are compared. A basic PRAM can perform the following operations in unit-time: addition, subtraction, Boolean operations, comparisons, and indirect addressing. Multiple processors may concurrently read and concurrently write a single cell. This thesis establishes that the class of languages accepted in polynomial time on a PRAM(*, {up arrow}, {down arrow}) contains the class of languages accepted in exponential time on a nondeterministic Turing machine (NEXPTIME) and is contained in the class of languages accepted in exponential space on a Turing machine. Efficient simulations are presented of PRAMs with enhanced instruction sets by sequential RAMs with the same instruction sets; also simulations of probabilistic PRAMs by deterministic PRAMs, using parallelism to replace randomness. Also given are simulations of PRAM(op)s by PRAMs, where both the simulated machine and the simulating machine are exclusive-read, exclusive-write machines.
- Research Organization:
- Illinois Univ., Urbana, IL (USA)
- OSTI ID:
- 7129054
- Country of Publication:
- United States
- Language:
- English
Similar Records
Simulation of parallel random access machines by circuits
An NC algorithm for recognizing tree adjoining languages