Our multiple microprocessor approach to application processing provides an open architecture that permits computing resources to be expanded and customized to meet simulator/training requirements Migrating computing resources from the minicomputer to the microprocessor environment results in an enormous savings in simulation costs. These savings are realized in the initial hardware investment and again in the reduction of facility requirements, including power, space and cooling. This was particularly evident in our parallel microprocessor implementation of the rotorcraft flight model at McDonnell Douglas Helicopter Company, Mesa, AZ, for engineering simulation.
Our rotorcraft simulation was migrated from the GOULD 32/9780 to the Motorola 68020 microprocessor environment. The required frame rate exceeded the capabilities of a single microprocessor. Therefore, the rotorcraft simulation, previously running in serial, was divided into three small models running in parallel on three microprocessors: the main rotor/tail rotor and equations of motion, the stabilization systems, and the engines. The VME bus was selected as the backplane for the system because of high speed data transfer capabilities of this bus and the large selection of devices available for VME. The Motorola 68020 CPU was selected as the target microprocessor. Development was done on the VAX/VMS system using FORTRAN 77 and C cross-compilers for the MC68020 target. Although it was initially uncertain whether the flight model could remain stable running on parallel processors, the parallel portage resulted in insignificant variance from the serial model. Timing results were easily within the required frame rate. This paper discusses the advantages of the microprocessor approach over the minicomputer approach for our rotorcraft flight simulation, the migration of the serial flight model to parallel processing, and how this approach can further enhance rotorcraft flight simulation.