Redsharc: A Programming Model and On-Chip Network for Multi-Core Systems on a Programmable Chip

The reconfigurable data-stream hardware software architecture (Redsharc) is a programming model and network-on-a-chip solution designed to scale to meet the performance needs of multi-core Systems on a programmable chip (MCSoPC). Redsharc uses an abstract API that allows programmers to develop systems of simultaneously executing kernels, in software and/or hardware, that communicate over a seamless interface. Redsharc incorporates two on-chip networks that directly implement the API to support high-performance systems with numerous hardware kernels. This paper documents the API, describes the common infrastructure, and quantifies the performance of a complete implementation. Furthermore, the overhead, in terms of resource utilization, is reported along with the ability to integrate hard and soft processor cores with purely hardware kernels being demonstrated. 1. Introduction Since the resources found on FPGA devices continue to track Moore’s Law, modern, high-end chips provide hundreds of millions of equivalent transistors in the form of reconfigurable logic, memory, multipliers, processors, and a litany of increasingly sophisticated hard IP cores. As a result, engineers are turning to multi-core systems on a programmable chip (MCSoPC) solutions to leverage these FPGA resources. MCSoPC allow system designers to mix hard processors, soft processors, third party IP, or custom hardware cores all within a single FPGA. In this work, we are only considering multi-core systems with a single processor core, multiple third party IP cores, and multiple custom hardware cores. A major challenge of MCSoPC is how to achieve intercore communication without sacrificing performance. This problem is compounded by the realization that cores may use different computational and communication models; threads running on a processor communicate much differently than cores running within the FPGA fabric. Furthermore, standard on-chip interconnects for FPGAs do not scale well and cannot be optimized for specific programming models; contention on a bus can quickly limit performance. To address these issues, this paper investigates Redsharc—an API and common infrastructure for realizing MCSoPC designs. Redsharc’s contribution has two parts. First, introduction of an abstract programming model and API that specifically targets MCSoPC is presented. An abstract API, as described by Jerraya and Wolf in [1], allows cores to exchange data without knowing how the opposite core is implemented. In a Redsharc system, computational units, known as kernels, are implemented as either software