A parallel Monte Carlo simulation for gaseous electronics on a dynamically reconfigurable multiprocessor system

Abstract

The Parallel Monte Carlo method implemented on a dynamically reconfigurable multiprocessor system is presented for simulating the evolution of an assembly of electrons interacting with a background gas under the influence of an electric field. The number of electrons keep increasing due to the electron impact ionization process. Since each electron can be traced independently, the simulation is inherently parallel. However, the Monte Carlo simulation is prohibitively expensive when a significantly large number of particles are needed to achieve satisfactory statistics. Hence, a low-cost multiprocessor system is constructed by grouping a number of Inmos T800 Transputers with a dynamically reconfigurable interconnection network implemented with an Inmos C004 crossbar switch. Both the hardware and software are discussed in detail. The performance of the Monte Carlo simulation on the multiprocessor system shows that this special reconfigurable system is cost-effective for investigations of gaseous-electronics problems which requires a considerable amount of computer time.