Comparative Analysis of Programming Models for Reconfigurable Hardware Systems

Abstract

Field programmable gate array (FPGA) based reconfigurable computing sys
tems are shown to have great potential for accelerating computationally 
intensive applications. To date, however, these systems have had to be pro
grammed with specialized hardware design skills, making them less accessi
ble. These models and tools, which aim at simplifying FPGA development, are 
examined in this article, and the ease of use, performance, and generational 
efficiency in producing hardware designs are compared among them. This 
has allowed the use of reconfigurable hardware through high level synthe
sis (HLS) tools without having in depth hardware design knowledge. We will 
demonstrate imperative, functional, and graphical programming paradigms 
via Impulse C, Mitrion-C, and DSPLogic. Through analysis of the programming 
models, development workflows and results obtained across multiple bench
mark applications, we can discern the tradeoffs between performance and 
productivity for reconfigurable computing.