Accelerating Manycore Simulation by Efficient NoC Interconnection Partition on FPGA Simulation Platform

Abstract

manycore architecture is the trend of system design. However, manycore simulators face the important issue than general multicore simulators is simulation speed. Several studies provide a FPGA simulation methodology to solve it, but they almost do it by providing a new internal design of core. And some of them don\'t care the correctness without interconnection simulation. This paper provides a PVCT module and NIP methodology to solve these two issues. According to our experiment, we could provide a 91.4 MIPS simulation performance averagely with Splash2 benchmark by three physical cores. The number of three is decided by capability of Xilinx XC2V8000 FPGA chip.