Run-Time Optimized Reconfiguration Using Instruction Forecasting

The extensive use of reconfigurable computing devices has imposed a new category of processors, the dynamic instruction set processors (DISPs) that customize their instruction sets dynamically to the application needs. One of the major drawbacks of DISPs is the reconfiguration time needed to alter the instruction set, which is directly added to the program execution time discouraging the use of DISPs especially for time critical processing applications. This paper introduces a methodology for optimizing reconfiguration time through instruction forecasting and presents the results obtained when applying this method to Medium Access processing systems that execute time critical network tasks.