A quasi-cycle accurate timing model for binary translation based instruction set simulators

Abstract

Software defines the functionality of today's Cyber-Physical Systems (CPS). Many product innovations are based on software and thus the complexity of software, even when running on platforms equipped with small microprocessors, is increasing dramatically. This calls for adequate embedded software integration testing, even before the actual hardware platform is available. The application of virtual platforms for functional validation, that allows simulating CPS running real target platform application code on a generic host computer, is currently being adopted by the industry. Since the correct behavior of a CPS not only depends on the correctness of computation but also on its timeliness, virtual platforms contain a certain notion of time. This work focuses on enhancing OVP processor models by a quasi-cycle accurate timing model. This paper demonstrates and evaluates the accuracy of the proposed timing model against real hardware measurements for the Xilinx MicroBlaze and ARM Cortex-M0 processors. Results show a mean error of 0.16% for the MicroBlaze and 0.72% for the ARM Cortex-M0 processor over all considered benchmarks, which is a clear improvement compared to previous published work.