Emulation of microprocessor memory systems using the RAMP design framework

Abstract

With the computer hardware industry and the academic world focused on multiprocessor systems, the RAMP project is aiming to provide the infrastructure for supporting high-speed emulation of large scale, massively-parallel multiprocessor systems using FPGAs. The RAMP design framework provides the platform for building this infrastructure. This research utilizes this design framework to emulate various microprocessor memory systems through a model built in an FPGA. We model both the latency and the bandwidth of memory systems through a parameterized emulation platform, thereby, demonstrating the validity of the design framework. We also show the efficiency of the framework through an evaluation of the utilized FPGA resources.