Out-of-order execution on reconfigurable heterogeneous MPSOC using particle swarm optimization

In a Multiprocessor System On Chip design (MPSOC), the Out-of-order (OoO) execution scheme makes a vital role in task-level parallelism circuit. Scheduling and mapping the tasks on the processing elements are the important challenges of OoO execution. At this level, the major drawback of current programming models is that programmers need to handle the task assignments manually, which would seriously increase the burden of programmers. The Analysis of inter-task dependency is the main challenge for OoO in task level parallelism. Instruction level score boarding algorithm and Tomasulo algorithm are used in analysing inter-task dependences. The prototyped field programmable gate array (FPGA) fabric is composed of both software based static and dynamic implementation on top of a heterogeneous MPSOC. By combining the heterogeneous MPSOC with reconfigurable computing the computational capabilities of heterogeneous MPSOC will greatly improve. The trade-off between static and dynamic approaches can also be analysed. In this paper, modified score boarding algorithm is used along with particle swarm optimization technique. Particle swarm optimization improves the efficiency and performance.